Volume 4, Number 1

the state of the Internet

1st Quarter, 2011 report

Get the most out of the State of the Internet with our new Data Visualization
www.akamai.com/stateoftheinternet

Average Peak Connection Speed

50,000

40,000

GERMANY SPAIN JAPAN UNITED STATES HONG KONG AUSTRALIA

kbps

30,000

20,000

10,000

Q1 08

Q2 08

Q3 08

Q4 08

Q1 09

Q2 09

Q3 09

Q4 09

Q1 10

Q2 10

Q3 10

Q4 10

Q1 11

Source: Akamai

View trends over time for key metrics from the report across the top 100 countries/regions

Compare trends across countries/regions

save & print customized graphs

letter From the editor

With the publication of this edition, akamais State of the Internet report enters its fourth year. over the course of the previous three years, weve used our unique vantage point on the Internet and the incredible volume of data collected by the akamai Intelligent Internet platform to track the growth of Internet usage and Internet connection speeds around the world, trends regarding where Internet attacks are coming from and what these attacks are targeting, and the growing use of mobile devices to access the Internet.
While the content in the report covers the first quarter of 2011, callouts within the various sections highlight historical perspectives and trends seen over the last three years related to the various metrics. In addition, alongside this quarters report, we are launching a new data visualization tool, available at www.akamai.com/stateoftheinternet, which allows users to select metrics, time frames, and geographies of interest, and then generate (and download) graphs of the associated data. the tool currently includes the top 100 countries/regions by unique Ip address count, and we plan to expand it to include state-level data for the united states in the future. though this report covers the first quarter of 2011, several Internet-related events of note occurred during the second quarter, including:
the shutdown of Internet service in syria, apNIC implementing austerity measures

following similar outages in egypt and libya in the first quarter;

World Ipv6 Day, which was intended to be a

around the assignment of Ipv4 address blocks from a rapidly dwindling pool of available space;
and streaming of the royal Wedding of prince

test flight of Ipv6 across a number of leading Web properties;

William and Catherine middleton, which broke traffic records on sites across the Web.

In next quarters report, we will look at the impact these events had on the Internet, as well as continuing to expand the scope of content within the report, especially around security- and mobile-related topics.

David belson

table of Contents
table of Contents
ExEcutivE Summary SEction 1: SEcurity 1.1 attack traffic, top originating Countries 1.2 attack traffic, top ports SEction 2: intErnEt PEnEtration 2.1 unique Ipv4 addresses 2.2 Ipv4 address space exhaustion 2.3 Ipv6 adoption SEction 3: GEoGraPhy Global 3.1: Global average Connection speeds 3.2: Global average Connection speeds, City View 3.3: Global average peak Connection speeds 3.4 Global average peak Connection speeds, City View 3.5 Global High broadband Connectivity 3.6 Global broadband Connectivity 3.7 Global Narrowband Connectivity SEction 4: GEoGraPhy unitEd StatES 4.1 united states average Connection speeds 4.2 united states average Connection speeds, City View 4.3 united states average peak Connection speeds 4.4 united states average peak Connection speeds, City View 4.5 united states High broadband Connectivity 4.6 united states broadband Connectivity 4.7 united states Narrowband Connectivity SEction 5: mobilE connEctivity 5.1 attack traffic From mobile Networks, top originating Countries 5.2 attack traffic From mobile Networks, top ports 5.3 Connection speeds & Data Consumption on mobile Networks 5.4 mobile traffic Growth as observed by ericsson 5.5 3G Data traffic patterns of multiple Connected Device types as observed by ericsson SEction 6: nEtwork outaGES and diSruPtionS 6.1 egypt 6.2 libya 6.3 Japan 6.4 Georgia/armenia SEction 6: aPPEndix SEction 7: EndnotES
4

5 6 6 7 8 8 9 10 11 12 13 15 16 18 19 20 21 21 22 22 23 23 24 25 26 26 27 28 31 31 34 34 35 36 37 38 39

2011 akamai technologies, Inc. all rights reserved

executive summary
akamais globally distributed network of servers allows us to gather massive amounts of information on many metrics, including connection speeds, attack traffic, and network connectivity/availability/latency problems, as well as traffic patterns on leading Web sites. each quarter, akamai publishes a state of the Internet report. this report includes data gathered from across akamais Intelligent Internet platform during the first quarter of 2011 about attack traffic, broadband adoption, and mobile connectivity, as well as trends seen in this data over time. In addition, this quarters report also includes insight into the state of Ipv4 exhaustion, Ipv6 adoption, and several high profile Internet outages/ disruptions seen in the first quarter.
Attack Traffic
During the first quarter of 2011, akamai observed attack traffic originating from 199 unique countries around the world. myanmar was the top attack traffic source, accounting for 13% of observed attack traffic in total. the united states and taiwan held the second and third place spots respectively, accounting for just under 20% of observed attack traffic combined. attack traffic concentration was lower than in the fourth quarter of 2010, with the top 10 ports seeing 65% of observed attack traffic, including a set of attacks that may have been looking to exploit the Internet privacy tool tor as a means of hiding their tracks. position, with 72% of connections to akamai occurring at 5 mbps or greater. Delaware also maintained the highest average connection speed, at 7.5 mbps, as well as the highest average peak connection speed across the united states, at 30.1 mbps. riverside, California was the united states city with the highest average connection speed (7.8 mbps) in the first quarter, and North bergen, NJ had the highest average peak connection speed (40 mbps).

Mobile Connectivity
reviewing first quarter observed attack traffic from known mobile networks, overall attack traffic concentration remained fairly consistent from the prior quarter, with the top 10 countries generating just under three-quarters of the observed attacks. the targeted ports were largely similar to the overall port list, and port 445 continues to be the target of a significantly higher percentage of attacks as compared to the other ports in the top 10. In the first quarter of 2011, average measured connection speeds on known mobile providers around the world ranged from just over 6 mbps down to 163 kbps. average peak connection speeds on mobile providers around the world ranged from 22.7 mbps to just over 1 mbps. looking at content consumption metrics, users on seven providers consumed, on average, more than one gigabyte (1 Gb) of content from akamai per month, while users on 77 additional providers downloaded more than 100 mb of content from akamai per month during the first quarter. In addition, based on data collected by ericsson, mobile data traffic saw 130% yearly growth in the first quarter, and is now more than double the measured volume of voice traffic.

Internet and Broadband Adoption

akamai observed a 5.2% increase (from the fourth quarter of 2010) globally in the number of unique Ipv4 addresses connecting to akamais network, growing to over 584 million. From a global connection speed perspective, south Korea recorded the highest level of high broadband (>5 mbps) connectivity, with 60% of connections to akamai at speeds above 5 mbps. south Korea also achieved the highest average connection speed at 14.4 mbps. Hong Kong maintained its position as having the highest average peak connection speed, where the per-Ip address maximum connection speed was averaged across all of the Ip addresses seen from each country. Cities in Japan and south Korea continued to hold many of the top spots in the rankings of highest average and average peak connection speeds by city. In the united states, Delaware remained in the top

2011 akamai technologies, Inc. all rights reserved

security
akamai maintains a distributed set of agents deployed across the Internet that monitor attack traffic. based on the data collected by these agents, akamai is able to identify the top countries from which attack traffic originates, as well as the top ports targeted by these attacks. (ports are network layer protocol identifiers.) this section provides insight into attack traffic, as observed and measured by akamai, during the first quarter of 2011.
1.1 Attack Traffic, Top Originating Countries
During the first quarter of 2011, akamai observed attack traffic originating from 199 unique countries/ regions, down from 207 at the end of 2010. as shown in Figure 1, the first quarter saw several changes in the list of the top 10 attack traffic sources, with myanmar making its first appearance in the history of the report, India appearing for the first time since the fourth quarter of 2009, and Hong Kong appearing for the first time since the third quarter of 2008. among the countries/ regions more frequently seen on the top 10 list, the united states and taiwan were responsible for higher percentages of attack traffic as compared to the prior quarter, while russia, China, brazil, romania, and India all saw their percentages decline quarter-over-quarter. this sudden appearance of myanmar on the list of top attack traffic sources is certainly unusual, and appears to be related to attack traffic targeting port 80 observed by akamai in late February and early march. Interestingly, myanmar managed to be responsible for 13% of the observed attack traffic in the first quarter even though only 25 unique ports were targeted, and of that, over 45% of the attacks targeted port 80. (Contrast that with the united states, with 10% of the observed attack traffic and tens of thousands of targeted ports very strongly indicative of general port scanning activity, as opposed to specifically targeted attacks.) a Web search for the Ip address blocks from myanmar that were observed to be originating the attacks returned reports on tracking sites ipillion.com and bizimbal.com of others seeing similar attack traffic from these Ip address blocks as well.1 aggregating observed attack traffic at a continental level, we find that nearly half of the observed attack traffic came from the asia pacific/oceania region, nearly 30% came from europe, and just over 20% came from the americas.

seCtIoN 1:

Country/Region

Q1 11 % Traffic

Q4 10 %

1 2 3 4 5 6 7 8 9 10

Myanmar United States Taiwan Russia China Brazil India Hong Kong Romania Italy Other

13% 10% 9.1% 7.7% 6.4% 5.5% 3.8% 3.3% 2.5% 2.5% 36%

N/A 7.3% 7.6% 10% 7.4% 7.5% 2.1% 0.3% 2.6% 3.6% 45%

10

4 5

2 6 7

Figure 1: Attack Traffic, Top Originating Countries/Regions

2011 akamai technologies, Inc. all rights reserved

1.2 Attack Traffic, Top Ports

attack traffic concentration among the top 10 ports continued to drop from the concentration seen in the fourth quarter of 2010, with the top 10 ports responsible for just 65% of the observed attacks (down from 72% in the fourth quarter). perpetual top target port 445 (microsoft-Ds) dropped nearly 25% from the prior quarter, and ports 23 (telnet) and 22 (ssH) also saw significant percentage declines. However, port 80 (WWW) saw attack traffic levels over 7x higher than at the end of 2010, and the percentage of attacks targeting port 443 (Https/ssl) also saw a massive increase over the prior quarter. as noted above, it is likely that the growth in attack traffic targeting port 80 and port 443 is related to the attacks observed to be originating from myanmar and Hong Kong. the ongoing decline of attacks on port 445 continues to underscore the success of efforts to mitigate the threat posed by the Conficker worm, which is now over three years old. a report released2 by the Conficker Working Group in January 2011 claimed success in ultimately stopping Conficker from communicating with its creator, thus preventing it from updating into newer and more dangerous variants, though it also noted that Conficker still resides on anywhere from four million to 13 million computers across the world. as shown in Figure 2, in addition to port 443s first appearance in the list, port 21 appears on the top ports list for the first time this quarter as well. While officially

assigned to the File transfer protocol (Ftp), several online security resources3 also note that the port is used by a number of trojans malware hidden on a computer system that can steal information or harm the system. port 9050 appears on the list for the first time in the first quarter, ostensibly replacing the unassigned port 9415 that appeared on the list in the fourth quarter of 2010. While officially assigned4 to Versiera agent listener (an enterprise network management & monitoring tool), it appears that Internet privacy tool tor may also use port 9050 for soCKs proxy purposes.5 (that is, for general proxying of tCp connections.) In reviewing ports targeted by the top 10 countries/regions, it appears that nearly all of the observed attacks on this port came from the united states, though it only accounted for 5.8% of the attacks observed from the united states. as such, it may represent attackers based in the united states looking to hide their tracks by leveraging the anonymity afforded by connecting through tor. When reviewing the top ports targeted by attacks originating in China, it is interesting to note that the top three targeted ports (1433, 3389, 445) accounted for just over 20% of the first quarter attacks observed originating from the country, and are all used by microsoft software/ protocols. port 22 (ssH) and port 3306 (mysQl) round out the top 5 within China, possibly indicating that attacks targeting these two ports are searching for systems with weak passwords that can be exploited for the installation of malware, or for use as members of a botnet.

Port

Port Use

Q1 11 % Traffic

Q4 10 %

445 80 443 23 22 1433 25 9050 21 135 Various

Microsoft-DS WWW (HTTP) HTTPS/SSL Telnet SSH Microsoft SQL Server SMTP Versiera Agent Listener FTP Microsoft-RPC Other

34% 11% 4.7% 4.1% 3.3% 1.7% 1.6% 1.5% 1.5% 1.5% 35%

47% 1.5% 0.2% 11% 6.2% 1.1% 0.4% < 0.1% 0.3% 1.1%

Other 35%

Microsoft-DS 34%

Microsoft RPC 1.5% FTP 1.5% Versiera Agent Listener 1.5% STMP 1.6% Microsoft SQL Server 1.7% SSH 3.3%

WWW 11%

Figure 2: Attack Traffic, Top Ports

HTTPS/SSL 4.7% Telnet 4.1%

2011 akamai technologies, Inc. all rights reserved

Internet penetration
2.1 Unique IPv4 Addresses
through a globally-deployed server network, and by virtue of the approximately one trillion requests for Web content that it services on a daily basis, akamai has unique visibility into levels of Internet penetration around the world. In the first quarter of 2011, over 584 million unique Ip addresses, from 237 countries/regions, connected to the akamai network 5.2% more Ip addresses than in the fourth quarter of 2010, and 20% more than in the first quarter of 2009. although we see more than half a billion unique Ip addresses, akamai believes that we see well over one billion Web users. this is because, in some cases, multiple individuals may be represented by a single Ip address (or small number of Ip addresses), because they access the Web through a firewall or proxy server. Conversely, individual users can have multiple Ip addresses associated with them, due to their use of multiple connected devices. as shown in Figure 3, nine of the top 10 countries remained consistent with the prior quarter, with Canada ceding its place on the list to Italy. all of the countries on the list saw quarterly growth, with Italys 11% increase leading the way (and besting Canadas 0.6% increase, which dropped it to 11th place globally). Yearly growth across all of the top 10 countries was strong as well, with double digit percentage increases seen in all of the countries except France, which turned in a still respectable increase of nearly 7%. after showing yearover-year growth rates above 30% each quarter during 2010, Chinas growth appears to have slowed a bit in the first quarter of 2011, dropping slightly to 27%. Concentration among the top 10 continued to be consistent with prior quarters, with those countries still accounting for nearly 70% of the observed Ip addresses. In looking at the long tail, there were 186 countries/ regions with fewer than one million unique Ip addresses connecting to akamai in the first quarter of 2011, 134 with fewer than 100,000 unique Ip addresses, and 31 with fewer than 1,000 unique Ip addresses. the counts for all three thresholds were up slightly quarter-over-quarter.

seCtIoN 2:

Country/Region

Q1 11 Unique IP Addresses

QoQ Change

YoY Change

7 5 10 1 8

1 2 3 4 5 6 7 8 9 10

Global United States China Japan Germany France South Korea United Kingdom Brazil Italy Spain

584,821,069 142,605,731 73,587,347 41,233,145 34,785,032 24,010,722 22,538,305 22,333,025 14,153,991 13,632,661 12,915,356

5.2% 3.9% 9.4% 4.3% 2.8% 3.9% 2.3% 0.7% 4.6% 11% 3.7%

20% 10% 27% 24% 12% 6.8% 35% 11% 24% 28% 15%

2 9

6 3

Figure 3: Unique IPv4 Addresses Seen By Akamai

DID YOU KNOW?

The number of unique IPv4 addresses seen by Akamai grew from 323 million in Q1 2008 to 584 million in Q1 2011 up 80% over the three year period.

2011 akamai technologies, Inc. all rights reserved

2.2 IPv4 Address Space Exhaustion

on January 31, the Internet assigned Numbers authority (IaNa) distributed two of the remaining seven /8 blocks (comprising 16.8 million Ip addresses per block) of Ipv4 addresses to apNIC, the regional Internet registry (rIr) for the asia-pacific region.6 this distribution ultimately led to an event that took place on February 3, at which the five remaining /8 blocks of Ipv4 addresses were distributed to representatives of the five rIrs one block to each. this final exhaustion of the central pool of Ipv4 address space had been anticipated for quite some time and is considered to be a significant milestone for the Internet. With this exhaustion, each rIr now has a finite pool of Ipv4 addresses that it can allocate to network service providers and carriers within its region. each rIr will ultimately face exhaustion of its local pool of addresses and is adopting strict rules around requests for new address space and transfers of existing Ipv4 address space. one such transfer made the news at the end of march, when microsoft was required to satisfy the american registry for Internet Numbers (arIN) transfer policies to receive the address space it agreed to buy from bankrupt telecom gear vendor Nortel 666,624 legacy Ipv4 addresses for $7.5 million usD.7

However, John Curran, Ceo of arIN, noted that at some point in the not-too-distant future, it will become more cost-effective for most users to acquire and use free Ipv6 addresses than to buy legacy addresses, and the bottom will quickly fall out of the Ipv4 aftermarket.8 on april 15, apNIC released a statement noting that it had reached its final /8 Ipv4 address block, bringing the organization to what it termed stage three of Ipv4 exhaustion in the asia-pacific region.9 In this stage, each new or existing apNIC account holder is only eligible to request and receive delegations totaling a maximum of 1024 addresses (a /22) from the apNIC Ipv4 address pool, assuming it meets specific criteria.10 Figure 4 was included in a February 2011 blog post11 by Geoff Huston, Chief scientist at apNIC, and shows predicted exhaustion dates (as of that date) for the other rIrs. In a message12 to the North american Network operators Group (NaNoG) mailing list, Huston explained that it is a probabilistic graph that shows the predicted month when the rIr will be down to its last /8 policy (whatever that policy may be), and the relative probability that the event will occur in that particular month. (Note that this graph has since been updated, and a more recent version can be found at http://ipv4.potaroo.net.)

100 ... 60 50 Probability (%) 40 30 20 10 0

Jan 11

Jul 11

Jan 12
IaNa

Jul 12

Jan 13

Jul 13
rIpeNCC

Jan 14
arIN

Jul 14
laCNIC

Jan 15

Jul 15

Jan 16

apNIC

aFrINIC

Figure 4: Projected RIR IPv4 Exhaustion Dates (source: http://www.potaroo.net/ispcol/2011-02/rir.jpg)

2011 akamai technologies, Inc. all rights reserved

Internet penetration (continued)

2.3 IPv6 Adoption
Figure 5 shows Ipv6 adoption during the first quarter among the top one million13 Web sites as ranked by alexa. the graph shows the percentage of these top sites available via both Ipv4 and Ipv6 as measured by a monitor on Comcasts network, in cooperation with a project run by professor roch Guerin at the university of pennsylvania.14 Ipv6 reachability of these sites appeared to remain fairly constant at approximately 0.25% through the first half of the quarter but jumped suddenly to the 3% range in mid-February. In an e-mail exchange, professor Guerin noted that this increase in reachability was due to Google white-listing Comcast for Ipv6 connectivity as a result, the Comcast monitor was able to reach many blogspot.com hosts over Ipv6. (blogspot.com is the domain name used by blogger, a blogging tool owned by Google.) Guerin further noted that if these blogspot.com hostnames were removed from the results, then Ipv6 reachability at the end of the quarter would be approximately 0.3%, a level similar to that seen by monitors running at the university of pennsylvania. this jump in reachability due to whitelisting of a network provider also highlights the fragmented interconnectivity of the Ipv6 Internet in Googles case, it has a number of requirements that network providers must meet before they can access Google services over Ipv6.15 Internet security firm arbor Networks has also studied Ipv6 adoption and associated traffic levels. leveraging atlas, arbors distributed sensor network that is a collaborative effort with over 100 service providers distributed across 17 countries, arbor has examined Ipv6 traffic growth trends over time and periodically publishes blog posts16 examining its findings. Figure 6 is based on a graph published by arbor that focuses on Ipv6 traffic in six atlas participant network providers that can examine native Ipv6 traffic. the graph shows Ipv6 as an average percentage of all inter-domain traffic in these six providers, and though it only goes through the first half of the quarter, it shows that aggregate Ipv6 traffic volumes generally ranged between 0.1 and 0.2 percent of Internet traffic. arbor notes that this range corresponds with observations made by Google17 and ams-IX (the amsterdam Internet exchange). However, as the identities of these providers have not been published, it is not clear whether the observed Ipv6 traffic volumes are broadly representative of other providers it has been suggested that levels of Ipv6 traffic are higher on providers that are actively marketing Ipv6 services and that are taking an active role in Ipv6 rollout.

seCtIoN 2:

3.50 3.25 3.00 2.75 2.50 2.25 2.00 1.75 1.50 1.25 1.00 0.75 0.50 0.25 0.00 mar 11 apr 11 Feb 11 Jan 11

0.35 0.30 0.25 Traffic (%) 0.20 0.15 0.10 0.05 0.00
1/2/11 1/9/11 1/16/11 1/23/11 1/30/11 2/6/11 2/13/11

Reachability (%)

sites accessible via both Ipv4 and Ipv6

Figure 5: IPv6 Reachability Among Top 1 Million Web Sites (source: http://v6monitor.kangaroo.comcast.net:8180/monitor/)

Figure 6: IPv6 Traffic as a Percentage of All Traffic in Six Selected Carriers (Source: http://asert.arbornetworks.com/2011/04/six-monthssix-providers-and-ipv6/)

10

2011 akamai technologies, Inc. all rights reserved

Geography Global
by virtue of the approximately one trillion requests for Web content that it services on a daily basis through its globally-deployed server network, akamai has a unique level of visibility into the connection speeds of end-user systems and, consequently, into broadband adoption around the globe. because akamai has implemented a distributed network model, deploying servers within edge networks, it can deliver content more reliably and consistently at those speeds, in contrast to centralized competitors that rely on fewer deployments in large data centers. For more information on why this is possible, please see akamais How Will The Internet Scale? white paper18 or the video explanation at www.akamai.com/whytheedge.
the data presented within this section was collected during the first quarter of 2011 through akamais globally deployed server network and includes all countries/ regions that had more than 25,000 unique Ip addresses make requests to akamais network during the first quarter. (Note that the 25,000 unique Ip address threshold is a significant change from the 1,000 unique Ip address threshold that was used in the past we believe that this new, higher threshold will enable us to better address the unfair comparison of extremely small countries with much larger countries.) For purposes of classification in this report, the broadband data included below is for connections greater than 2 mbps, and high broadband is for connections of 5 mbps or greater. In contrast to the high broadband and broadband classifications, the narrowband data included below is for connections to akamai that are slower than 256 kbps. Note that the percentage changes reflected below are relative to the prior quarter(s). (that is, a Q4 value of 50% and a Q1 value of 51% would be reflected here as a 2% increase.) a quarter-over-quarter change is shown within the tables in several sections below in an effort to highlight general trends, and year-over-year changes are shown to illustrate longer-term trends. as noted in previous editions of the State of the Internet report, in July 2010, the united state Federal Communications Commission (FCC) revised its working definition of broadband to include download speeds of at least 4 mbps. We have considered aligning the definition of broadband within this report with the FCCs. However, additional research has shown that the term broadband has varying definitions across the globe Canadian regulators are targeting 5 mbps download speeds,19 whereas the european Commission believes citizens need download rates of 30 mbps,20 while peak speeds of at least 12 mbps are the goal of australias National broadband Network.21 as such, we believe that redefining the definition of broadband within the report to 4 mbps would be too united states-centric, and we will not be doing so at this time. as the quantity of HD-quality media increases over time, and the consumption of that media increases, end users are likely to require ever-increasing amounts of bandwidth. a connection speed of 2 mbps is arguably sufficient for standard-definition tV-quality content, and 5 mbps for standard-definition DVD quality video content, while bluray (1080p) video content has a maximum video bit rate of 40 mbps, according to the blu-ray FaQ.22 In addition to providing data on average connection speeds, we continue to report average peak connection speeds23 around the world, from a country/region, state, and city perspective. this metric can provide insight into the peak speeds that users can likely expect from their Internet connections. Finally, traffic from known mobile network providers will be analyzed and reviewed in a separate section of the report; mobile network data has been removed from the data set used to calculate the metrics in the present section.

seCtIoN 3:

2011 akamai technologies, Inc. all rights reserved

11

Geography Global (continued)

3.1 Global Average Connection Speeds
after remaining flat in the fourth quarter of 2010, the global average connection speed saw healthy quarterly growth in the first quarter of 2011, increasing nearly 10% to just over 2 mbps, as shown in Figure 7. In addition to this strong global growth, four countries within the top 10 saw quarterly growth of 10% or more, with Ireland seeing the greatest increase, at 16%. (Irelands increase allowed it to displace Canada from the top 10 in the first quarter, forcing it down to twelfth place.) Globally, over 40 countries/regions saw average connection speeds increase by 10% or more in the first quarter. only three countries/regions within the top 10 saw quarterly declines, all of which were fairly modest. all of the countries/regions in the top 10, as well as the united states (placing fourteenth), continued to maintain average connection speeds that exceeded the high broadband threshold of 5 mbps. the global average connection speed saw very healthy growth year-over-year as well, increasing 23%. Yearly growth of 20% or more was also seen in three other countries (south Korea, the Netherlands, and belgium), and growth in excess of 10% or more was seen in another three countries in the top 10 (Czech republic, switzerland, and Ireland), as well as in the united states. Globally, year-over-year increases were observed in over 110 countries/regions, with nearly 100 seeing at least double-digit percentage gains, while a dozen saw annual growth rates in excess of 100%. However, of these dozen, the united arab emirates had the highest average connection speed at 3.9 mbps, so even nominal increases in average connection speeds can equate to significant percentage changes. During the first quarter, 36 countries/regions had average connection speeds of 1 mbps or less. the slowest of this set was libya, at 328 Kbps. Note that the shift to requiring 25,000 unique Ip addresses to qualify for inclusion in section 3 has shifted the perspective here, dropping approximately 40 additional countries from consideration, and excluding perennial connection speed laggards such as Cuba and mayotte.

seCtIoN 3:

Country/Region

Q111 Avg. Mbps QoQ Change YoY Change

9 10 8 14 6

7 5 1 3 2

1 2 3 4 5 6 7 8 9 10 14

Global South Korea Hong Kong Japan Netherlands Romania Czech Republic Latvia Switzerland Belgium Ireland United States

2.1 14.4 9.2 8.1 7.5 6.6 6.5 6.3 6.2 6.1 5.6 5.3

9.7% 5.0% -1.7% -2.7% 7.6% -4.9% 14% 6.7% 10% 11% 16% 4.7%

23% 20% 2.1% 2.7% 25% 4.9% 19% 0.4% 17% 29% 14% 14%

Figure 7: Average Measured Connection Speed by Country/Region

12

2011 akamai technologies, Inc. all rights reserved

3.2 Global Average Connection Speeds, City View

as we have done in previous editions of the State of the Internet report, in examining average measured connection speeds at a city level, we have applied filters for unique Ip address count (50,000 or more seen by akamai in the first quarter of 2011) and academic institutions (removing data from known academic networks). In addition, as with the other data sets used in section 3 of this report, traffic from known mobile networks has been removed as well. as shown in Figure 8, Japanese cities tokai, shimotsuma, and Kanagawa topped the list of fastest cities in the first quarter, with average connection speeds of 13.2 mbps, 12.9 mbps, and 12.2 mbps respectively. Including these three, 13 cities achieved average connection speeds in excess of 10 mbps. the fastest city in europe was lyse, Norway, at 8.1 mbps, and riverside, California had the highest average connection speed in North america, at 7.8 mbps.

Continuing the trend seen in the previous year, cities in asia continued to dominate the top 100 list in the first quarter, holding two-thirds of the spots on the list. this included 61 cities in Japan, five in south Korea, and Hong Kong. twenty-one cities from North america made the list, including 18 from the united states and 3 from Canada. europe once again accounted for a dozen cities across ten countries (romania was the only european country with more than one on the list it managed three.) In reviewing the full global list of more than 800 cities that qualified for inclusion in this section, the fastest cities in other geographies included pretoria, south africa (africa), with an average connection speed of 1.5 mbps; riverwood, New south Wales, australia (oceania) with an average connection speed of 5.9 mbps; and munro, argentina (south america) with an average connection speed of 3.4 mbps.

DID YOU KNOW?

In Europe, the largest increase in average connection speed was seen in Georgia, which more than tripled over the last three years. Chinas average connection speed has grown by nearly half since Q1 2008, and exceeded 1 Mbps for the first time in Q1 2011. Canadas average connection speed grew over the last three years by nearly 70%, double the growth rate of 35% seen in the United States. Average connection speeds in Chile, Colombia, and Paraguay more than doubled from Q1 2008 to Q1 2011.

2011 akamai technologies, Inc. all rights reserved

13

Geography Global (continued)

Country/Region City Q1 11 Avg. Mbps Country/Region City Q1 11 Avg. Mbps

seCtIoN 3:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

Japan Japan Japan South Korea Japan Japan Japan South Korea Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan South Korea Japan Japan Japan Hong Kong Japan Japan Japan Norway Japan Japan Japan Japan Japan United States United States South Korea United States Romania Japan Japan Japan Japan Japan Canada Japan

Tokai Shimotsuma Kanagawa Seocho Asahi Yokohama Urawa Ilsan Nagano Hiroshima Tochigi Shizuoka Nagoya Ibaraki Toyonaka Chiba Gifu Marunouchi Kyoto Kobe Hyogo Nara Sendai Wakayama Seoul Osaka Yokkaichi Fukuoka Hong Kong Otsu Fukui Hakodate Lyse Fukushima Niigata Niho Matsuyama Tokushima Riverside, CA Staten Island, NY Yongsan San Jose, CA Constanta Tokyo Kochi Hamamatsu Kanazawa Hodogaya Oakville, ON Soka

13.2 12.9 12.2 12.1 11.9 11.7 11.4 11.3 11.2 11.2 10.9 10.7 10.4 9.9 9.9 9.7 9.6 9.6 9.5 9.5 9.3 9.3 9.1 9.0 8.8 8.7 8.6 8.6 8.6 8.6 8.3 8.2 8.1 8.1 8.1 8.0 8.0 7.9 7.8 7.8 7.8 7.8 7.7 7.7 7.7 7.7 7.6 7.6 7.6 7.6

51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100

Japan Japan Japan Japan Japan Japan United States Czech Republic Netherlands Japan Japan United States Japan Japan Portugal Romania Japan Japan Japan Romania Canada Spain United States Belgium United States United States Japan United States Japan United States United States United States United States United States Japan Japan United States Canada United States United States Japan Japan Latvia Japan Austria South Korea Switzerland Japan United States Japan

Yosida Okayama Mito Kumamoto Yamagata Yamaguchi Fremont, CA Brno Amsterdam Utsunomiya Saga Boston Metro, MA Miyazaki Kofu Porto Timisoara Kokuryo Tottori Kagoshima Iasi Victoria, BC Valencia Jersey City, NJ Antwerp Marietta, GA Anaheim, CA Toyama Traverse City, MI Nagasaki Hollywood, FL Spartanburg, SC Santa Barbara, CA Hayward, CA San Mateo, CA Oita Iwaki Oakland, CA Mississauga, ON Fond Du Lac, WI Union, NJ Okidate Naha Riga Akita Salzburg Taegu Zurich Kagawa Trenton, NJ Sapporo

7.5 7.5 7.5 7.4 7.4 7.4 7.4 7.4 7.3 7.2 7.2 7.1 7.1 7.1 7.0 7.0 7.0 7.0 6.9 6.9 6.9 6.8 6.8 6.7 6.7 6.7 6.7 6.6 6.6 6.6 6.6 6.6 6.5 6.5 6.5 6.5 6.5 6.4 6.4 6.4 6.3 6.3 6.3 6.3 6.3 6.3 6.3 6.3 6.3 6.2

Figure 8: Average Connection Speed, Top Global Cities

14

2011 akamai technologies, Inc. all rights reserved

3.3 Global Average Peak Connection Speeds

the average peak connection speed metric represents an average of the maximum measured connection speeds across all of the unique Ip addresses seen by akamai from a particular geography. the average is used in order to mitigate the impact of unrepresentative maximum measured connection speeds. In contrast to the average measured connection speed, the average peak connection speed metric is more representative of Internet connection capacity. (this includes the application of so-called speed boosting technologies that may be implemented within the network by providers, in order to deliver faster download speeds for some larger files.) Note that data from known mobile networks has also been removed from the source data set for this metric. as shown in Figure 9, the global average peak connection speed jumped above 10 mbps for the first time, growing an impressive 20% from the end of 2010, and up an even more impressive 65% from the beginning of 2010. modest growth was seen across eight of the top 10 countries/ regions, and the united states, which placed thirteenth.

south Korea, latvia, and bulgaria all added more than 10% quarter-over-quarter. Japan and the united arab emirates were the only two countries that saw quarterly declines, though neither lost a significant amount. looking at year-over-year changes, the nearly 4x growth seen in the united arab emirates was clearly the most significant, though three european countries grew 50% or more. Hong Kong, romania, and the united states all saw average peak connection speeds more than 30% higher than in the same quarter a year prior. Hong Kong remained the country/region with the highest average peak connection speed, landing just shy of 40 mbps. south Korea and romania also had average peak connection speeds above 30 mbps. the remaining countries in the top 10, as well as the united states, saw peak speeds above 20 mbps in the first quarter. In addition to those listed, five other countries (four in europe plus Canada) had average peak connection speeds above 20 mbps, while an additional 44 exceeded 10 mbps. under the new qualification guidelines, the country with the slowest average peak connection speed was libya, at just 1226 kbps.

Country/Region

Q1 11 Peak Mbps QoQ Change YoY Change

10 7 6 13 9

8 3 2 4 5 1

1 2 3 4 5 6 7 8 9 10 13

Global Hong Kong South Korea Romania Japan United Arab Emirates Portugal Belgium Latvia Bulgaria Netherlands United States

10.6 39.5 36.3 32.7 29.9 25.9 24.9 24.7 24.4 22.4 22.0 21.2

20% 4.2% 12% 3.3% -1.8% -4.8% 8.7% 8.1% 11% 18% 7.1% 4.3%

65% 34% 11% 31% 16% 394% 56% 50% 29% 45% 52% 31%

Figure 9: Average Peak Connection Speed by Country/Region

2011 akamai technologies, Inc. all rights reserved

15

Geography Global (continued)

3.4 Global Average Peak Connection Speeds, City View
as we have done in previous editions of the State of the Internet report, in examining average measured connection speeds at a city level, we have applied filters for unique Ip address count (50,000 or more seen by akamai in the first quarter of 2011) and academic institutions (removing data from known academic networks). In addition, as with the other data sets used in section 3 of this report, traffic from known mobile networks has been removed as well. as shown in Figure 10, nine of the 10 cities with the highest average connection speeds were in Japan. shimotsuma, Japan was the only city with an average peak connection speed in excess of 50 mbps, though tokai, Japan fell short by just 130 kbps. Including shimotsuma, 16 cities had average peak connection speeds at or above 40 mbps. an additional 37 cities had average peak connection speeds in excess of 30 mbps, while the remaining 47 were all above 20 mbps. Cities in asia once again dominated this metric, with the top 100 list including Hong Kong, Dubai, 54 cities in Japan, and 10 in south Korea. the top european city remained Constanta, romania (which rounded out the top 10), and it was joined by six other european cities, including an additional three from romania, as well as one each from portugal, Norway, and the Czech republic. In North america, 27 cities made the top 100 list, including 25 from the united states and two from Canada. In looking at the full global list of over 800 cities that qualified for inclusion, the fastest ones in other geographies included Casablanca, morocco (africa) with an average peak connection speed of 10.4 mbps; Canberra, australia (oceania) with an average peak connection speed of 22.5 mbps; and munro, argentina (south america) with an average peak connection speed of 20.6 mbps.

seCtIoN 3:

DID YOU KNOW?

In Europe, the largest increase (over 300%) in average peak connection speeds from Q1 2008 to Q 2011was seen in Bulgaria and Moldova. Average peak connection speeds in Australia and New Zealand have more than doubled over the last three years. While the average peak connection speed in the United States increased 95% from Q1 2008 to Q1 2011, Mexicos average peak connection speed grew 166% over the same period. Average peak connection speeds in Argentina, Brazil, Chile, Colombia, Paraguay, and Uruguay more than doubled from Q1 2008 to Q1 2011.

16

2011 akamai technologies, Inc. all rights reserved

Country/Region

City

Q1 11 Peak Mbps

Country/Region

City

Q1 11 Peak Mbps

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

Japan Japan Japan Japan Japan Japan Japan Japan Japan Romania Japan Japan Japan Japan Romania United States South Korea South Korea Romania Japan South Korea Japan Hong Kong Japan Japan Japan Japan United States Japan Japan Japan South Korea Japan Japan Japan Japan South Korea Japan Japan United States Japan Japan South Korea Japan Japan Japan Portugal Japan Japan Japan

Shimotsuma Tokai Kanagawa Marunouchi Yokohama Urawa Tochigi Hodogaya Nagano Constanta Soka Chiba Asahi Shizuoka Iasi North Bergen, NJ Taejon Seocho Timisoara Kokuryo Ilsan Ibaraki Hong Kong Nagoya Utsunomiya Hiroshima Mito Staten Island, NY Fukuoka Sendai Kyoto Taegu Kobe Gifu Niigata Yosida Kimchon Yokkaichi Kofu Van Nuys, CA Nara Hakodate Seoul Niho Otsu Fukui Porto Osaka Wakayama Fukushima

50.2 49.9 48.5 48.0 47.8 47.1 44.2 43.8 43.7 43.0 42.4 42.4 41.9 40.2 40.0 40.0 39.5 39.5 38.7 38.3 38.2 37.6 37.5 37.4 37.0 36.2 35.2 35.0 34.8 34.8 34.5 34.4 34.3 34.2 32.9 32.8 32.5 32.4 32.3 32.1 32.0 31.9 31.6 31.5 31.3 31.1 31.1 30.9 30.6 30.4

51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100

United States Japan Japan South Korea Romania South Korea Japan Japan Norway United States South Korea Japan Japan United States United States South Korea United States Japan Japan Japan Japan United States United States United States Czech Republic United States United States United States Japan United States Japan Japan United States Japan United States Japan Japan United States United States United States United States United States Japan Canada UAE United States United States United States Canada Japan

Riverside, CA Yamagata Hamamatsu Suwon Bucharest Yongsan Kanazawa Okidate Lyse Hayward, CA Sangamdong Matsuyama Iwaki San Mateo, CA Hollywood, FL Anyang Arvada, CO Okayama Yamaguchi Tokushima Tokyo Jersey City, NJ Waco, TX Boston Metro, MA Brno Tallahassee, FL Fremont, CA Marietta, GA Sapporo Ogden, UT Akita Hyogo Canton, OH Toyonaka Oakland, CA Kochi Kumamoto Bellevue, WA Santa Barbara, CA Federal Way, WA Spartanburg, SC Mishawaka, IN Morioka Mississauga, ON Dubai Vancouver, WA Saint Paul, MN Union, NJ Kelowna, BC Toyama

30.4 30.3 30.1 29.9 29.8 29.6 29.6 29.5 29.4 29.3 29.1 29.1 28.8 28.8 28.6 28.3 28.2 28.1 28.0 27.7 27.7 27.7 27.6 27.6 27.5 27.4 27.4 27.4 27.4 27.3 27.1 27.0 27.0 27.0 26.8 26.8 26.7 26.6 26.5 26.4 26.3 26.2 26.2 26.1 26.0 26.0 25.7 25.7 25.6 25.4

Figure 10: Average Peak Connection Speed, Top Global Cities

2011 akamai technologies, Inc. all rights reserved

17

Geography Global (continued)

3.5 Global High Broadband Connectivity
In the first quarter of 2011, global high broadband adoption increased almost six percent quarter-over-quarter, with 25% of all connections to akamai occurring at speeds of 5 mbps or more. as shown in Figure 11, south Korea returned to the top slot, starting 2011 with a high broadband adoption level of 60% -- though this is up from the prior quarter, it is down 10% from the start of 2010. Quarterly growth among other countries/regions in the top 10 ranged from strong 26% growth in the Czech republic to sub-1% increases in both the Netherlands and latvia. among the top 10 countries/regions, only Hong Kong, Japan, and romania saw quarter-overquarter declines, though all maintained high broadband adoption levels above 50%. the united states, ranked twelfth globally, saw a modest increase of almost 9% from the prior quarter, ending the first quarter of 2011 at a high broadband adoption rate of 39%. on a year-over-year basis, global high broadband adoption jumped 15%, and yearly growth was seen in eight of the top 10 countries/regions, as well as in the united states. only south Korea and Japan declined year-over-year, while growth in the other geographies in the top 10 ranged from a significant 44% increase in belgium, down to a still respectable 5.4% increase in Denmark. across the rest of the world, 13 countries saw high broadband adoption rates more than double year-over-year, with the united arab emirates growing 900% (to 26% adoption), and argentina increasing over 500% (to a nominal 3.6% adoption rate). looking at high broadband adoption on a global basis, the first quarter saw six countries/regions with more than half of their connections to akamai at speeds greater than 5 mbps this is up from four countries/regions in the prior quarter. beyond that, there were an additional 19 countries/regions (consistent with the fourth quarter of 2010) where more than a quarter of the connections were at high broadband rates, and 17 more (down from 21 in the fourth quarter of 2010) where at least one in ten connections was faster than 5 mbps. of the 55 geographies that qualified for inclusion in this section, China and India were the only two with high broadband adoption rates below 1% they achieved 0.5% and 0.4% adoption respectively. examining the percentage of connections to akamai at speeds above 10 mbps, we find that 6.7% of all connections globally exceeded this rate. south Korea, Japan, and Hong Kong were the only three countries/regions with more than a quarter of their connections in excess of 10 mbps, with adoption rates of 31%, 26%, and 26% respectively. Five other countries, all in europe, had more than 10% of connections to akamai at speeds above 10 mbps. of the countries that qualified for inclusion, turkey (0.6%), mexico (0.3%), and China (0.1%) were the only three with adoption rates for 10 mbps connectivity below 1%.

seCtIoN 3:

Country/Region

% Above 5 Mbps QoQ Change

YoY Change

10 2 9 12

1 2 3 4 5 6 7 8 9 10 12

Global South Korea Netherlands Hong Kong Japan Belgium Romania Czech Republic Latvia Canada Denmark United States

25% 60% 56% 55% 55% 52% 51% 48% 44% 44% 43% 39%

5.9% 17% 0.6% -1.4% -4.9% 11% -1.6% 26% 0.5% 5.2% 7.2% 8.8%

15% -10% 29% 18% -7.7% 44% 6.5% 38% 7.3% 29% 5.4% 22%

6 5 7

1 4 3

Figure 11: High Broadband Connectivity, Fastest Countries/Regions

18

2011 akamai technologies, Inc. all rights reserved

3.6 Global Broadband Connectivity

In the first quarter of 2011, global broadband adoption continued to increase slightly, gaining 2% from the end of 2010, to reach 62%. as shown in Figure 12, all of the countries/regions in the top 10 had 90% or more of their connections to akamai occurring at speeds of at least 2 mbps in the first quarter. Interestingly, europe is extremely well represented among the countries with the highest levels of broadband adoption, holding nine of the top 10 slots perennial connectivity strongholds south Korea and Japan rank eighteenth (87%) and twenty-ninth (79%) globally, likely indicating that while they have fairly strong levels of high broadband adoption (connections above 5 mbps), there is also a sufficient population of connections below 2 mbps that forced them out of the top 10 for this metric. Nine of the top 10 countries/regions, as well as the united states, saw increased broadband adoption rates quarterover-quarter Hong Kong was the only geography among the top 10 to see a quarterly decline. the increases were relatively modest, with luxembourgs 7% growth the largest of the group. Globally, three countries (ecuador, turkey, and serbia) more than doubled their levels of broadband adoption from the fourth quarter of 2010. on a year-over-year basis, global broadband adoption grew 9.5%, a growth level similar to that seen in the

Czech republic and romania. luxembourg had, far and away, the largest yearly increase of the top 10 countries/ regions, growing 33%. at just under 2%, switzerland had the lowest yearly percentage increase of the top 10. similar to the quarterly change, Hong Kong also declined very slightly (down 0.1%), and belgium joined it as well, losing a minor 1.1%. Globally, 17 countries more than doubled their levels of broadband adoption as compared to the start of 2010, from a massive 2000% increase in oman to 113% growth in the ukraine. (However, just over 30,000 broadband Ip addresses were seen by akamai from oman in the first quarter, as compared to just over 2 million from the ukraine, so the relative growth levels must be considered accordingly.) In the first quarter of 2011, 11 countries/regions (up from nine in the fourth quarter of 2010) saw broadband adoption levels of 90% or better. another 42 countries/regions (down from 53 in the prior quarter) had at least half of their connections to akamai at 2 mbps or more, 14 additional countries/regions had broadband adoption of at least 25%, and another 12 countries/regions had at least one in ten connections to akamai at 2 mbps or more. (these counts are lower than seen in prior quarters due to the new unique Ip address count thresholds for inclusion.) of the countries/ regions that qualified for inclusion, Venezuela had the lowest level of broadband adoption, at 1.7%.

Country/Region

% Above 2 Mbps QoQ Change

YoY Change

8 4

1 2 3 4 5 6 7 8 9 10 32

Global Bulgaria Czech Republic Switzerland Romania Luxembourg Hong Kong Hungary Germany Netherlands Belgium United States

62% 95% 94% 93% 93% 92% 92% 91% 91% 91% 90% 77%

2.0% 4.7% 3.3% 1.1% 0.7% 7.0% -1.6% 3.6% 3.7% 1.0% 0.1% 2.9%

9.5% 6.6% 9.5% 1.8% 9.9% 33% -0.1% 16% 12% 8.1% -1.1% 8.2%

10 5 32 3 7 1

Figure 12: Broadband Connectivity, Fast Countries/Regions

2011 akamai technologies, Inc. all rights reserved

19

Geography Global (continued)

3.7 Global Narrowband Connectivity
the impact of the change in the threshold for inclusion has been referenced in the prior subsections, but the impact of the change is most evident for the narrowband metric, as the countries that have been present in the top 10 list for the last three years, in general, no longer qualify for inclusion. readers will note that the countries in the top 10 are different than have been seen previously, and that the united states global rank is significantly higher, due to fewer countries/regions appearing on the overall global list. as shown in Figure 13, the global level of narrowband adoption declined nicely in the first quarter of 2011, with 3.3% of all connections to akamai at speeds below 256 kbps. this adoption level is 15% lower than at the end of 2010 and 36% lower than at the start of that year. among the countries appearing in the top 10 in the first quarter of 2011, all but two saw quarterly declines. While Indias 0.1% increase in narrowband adoption is notable, but not significant, the 252% increase seen from libya in the first quarter is certainly of concern. However, we believe that this jump could potentially be related to extreme congestion caused by the government-imposed multi-day Internet shutdowns, during which traffic into and out of the country was severely restricted. (this is covered in more detail in section 6.2 of this report.) From a year-over-year perspective, only Indonesia and India saw increased levels of narrowband adoption, and the levels of growth are fairly significant for both countries. though narrowband adoption levels grew 69% year-over-year in Indonesia, it appears that the government there is looking to actively improve the connectivity situation within the country, signing the Jakarta Declaration For meaningful broadband on april 14, 2011.24 the declaration seeks to bring the benefits of broadband-enabled services rapidly and meaningfully to at least 30% of Indonesian society by 2014 through new wireless network technologies, affordable broadband-capable connected devices, the completion of a national fiber-optic backbone infrastructure, and public/private partnerships. India is also planning to take aggressive steps to improve broadband connectivity throughout the country, with the proposed construction of a usD $13 billion national broadband network,25 intended to connect all cities, towns and villages with a population of more than 500 in two phases targeted for completion by 2012 and 2013. published reports 26 indicate that the network would be capable of supporting connection speeds of up to 10 mbps in 63 of the countrys metropolitan areas and larger cities by 2014, while speeds of up to 4 mbps would be offered in 352 additional cities. of the countries/regions that qualified for inclusion, 18 recorded narrowband adoption levels below the global figure of 3.3% in the first quarter or 2011. Half of those saw narrowband adoption below 1%, with France recording the lowest level, at 0.3%.

seCtIoN 3:

Country/Region

% Below 256 kbps

QoQ Change

YoY Change

1 2 3 4 5 6 7 8 9 10 28

Global Lebanon Uzbekistan Libya Bolivia Nigeria Iran Indonesia Nepal India Syria United States

3.3% 61% 54% 52% 51% 40% 38% 38% 36% 35% 20% 2.0%

-15% -7.7% -19% 252% -14% -8.1% -2.4% -13% -40% 0.1% -22% -20%

-36% -11% -36% -37% -25% -23% -17% 69% -36% 32% -5.1% -25%

10 1 28 3 4 5 6 7 9

2 8

Figure 13: Narrowband Connectivity, Slowest Countries/Regions

20

2011 akamai technologies, Inc. all rights reserved

Geography united states

the metrics presented here for the united states are based on a subset of the data used for section 3 and are subject to the same thresholds and filters discussed within the prior section. (the subset used for this section includes connections identified as coming from networks in the united states, based on classification by akamais edgescape27 geolocation tool.)
4.1 United States Average Connection Speeds
Delaware continued to be the fastest state in the union, as shown in Figure 14, with an average connection speed of 7.5 mbps, up 3.5% from the fourth quarter of 2010. seven other states among the top 10 joined it in seeing quarterly increases, with both Virginia and California growing 10% or more. across the country, 39 states saw average connection speeds increase quarter over quarter, from 13% in California to just 0.3% in Nevada. among the top 10, only two states saw average connections speeds decline quarter-over-quarter, with rhode Island dropping 1.1%, and utah shedding 0.1%. a total of 11 states plus the District of Columbia had lower average connection speeds than in the prior quarter, from the 0.1% losses in utah and Illinois to the more significant 16% decline in Vermont. From a yearly perspective, Delaware was once again the only state among the top 10 to experience a yearover-year decline, though this quarters decrease, at 0.7%, was lower than the 6% drop seen in the fourth quarter of 2010. Growth among the remaining states in the top 10 was, by and large, very solid, with seven of the states growing average connection speeds by more than 10% year-over-year. across the whole country, 42 states posted yearly increases in average connection speeds, with growth of 10% or more in 23 of those states. louisiana lost the least ground as compared to the beginning of 2010, dropping just 0.3%, while the District of Columbias 27% yearly drop was the most significant.

seCtIoN 4:

State

Q1 11 Avg. Mbps

QoQ Change

YoY Change

4 3 9 10 6 1 8

1 2 3 4 5 6 7 8 9 10

Delaware Rhode Island Wisconsin New Hampshire Connecticut Indiana Maine Virginia California Utah

7.5 6.8 6.0 6.0 5.9 5.8 5.7 5.7 5.6 5.6

3.5% -1.1% 7.7% 2.5% 6.3% 4.8% 2.8% 10% 13% -0.1%

-0.7% 18% 16% 2.4% 7.8% 11% 15% 24% 15% 11%

7 2 5

Figure 14: Average Measured Connection Speed by State

DID YOU KNOW?

The average connection speed in perennial speed leader Delaware has grown by less than six percent in total over the last three years.

2011 akamai technologies, Inc. all rights reserved

21

Geography united states (continued)

4.2 United States Average Connection Speeds, City View
as with the Global average Connection speeds, City View presented in section 3.2, connections from known academic and mobile networks were removed from the underlying data set for this metric, and the 50,000 unique Ip address filter was used for this view as well. as shown in Figure 15, riverside, Ca once again topped the list, with an average connection speed of 7.8 mbps. (While the top three cities all have an average connection speed of 7.8 mbps, this is due to rounding riversides speed is actually just above that mark, while staten Island and san Jose fall just below it). the average connection speeds in the top 10 cities all exceeded the high broadband threshold of 5 mbps. In the first quarter of 2011, Californias dominance of the average connection speeds by city list continued to falter, with Californian cities taking only four of the top 10 slots. the east Coast is represented by five cities across New York, New Jersey, massachusetts, Georgia, and Florida, and rounding out the list once again is traverse City, michigan.

seCtIoN 4:

4.3 United States Average Peak Connection Speeds

Consistent with its standing as the fastest state in the nation, Delaware broke the 30 mbps barrier for average peak connection speed in the first quarter of 2011, gaining 5.7% to reach 30.1 mbps. as shown in Figure 16, the remainder of the states maintained average peak connection speed levels above 20 mbps in the first quarter. Quarterly changes among the top 10 were largely positive, with eight of the top 10 seeing quarter-over-quarter increases, from New Hampshires 1.6% growth to a solid 11% increase in maine. across the country, 33 states improved their average peak connection speeds as compared to the fourth quarter of 2010 the greatest change was seen in ohio, with 15% quarterly growth. the only two states in the top 10 to see their average peak connection speeds decline quarter-over-quarter were rhode Island and Hawaii, which lost 2.7% and 3.7% respectively. across the country, they were part of a group of 17 states, plus the District of Columbia, that saw lower average peak connection speeds in the first quarter. losses ranged from 0.4% in arkansas to a much larger 14% drop in Vermont. From a year-over-year perspective, however, all of the states in the top 10 saw higher average peak connection speeds as compared to the same period a year prior. Growth among the group was rather strong as well, with three states seeing yearly growth above 30% and four states seeing yearly growth above 20%. New Hampshire was the only state within the top 10 that grew less than 10% year-over-year, falling just shy at 9.7%. across the country, all 50 states saw higher average peak connection speeds in the first quarter of 2011 than they did in the first quarter of 2010, with 45 states increasing by 10% or more. only the District of Columbia lost ground year-over-year, dropping 14%. as noted in previous reports, the average peak connection speed metric represents, in essence, the speed that end users Internet connections are capable of. Given that the long-term trends for this metric were positive across all states (except for the District of Columbia), and that the long-term average connection speed trend across the united states was positive for most states, it is not unreasonable to draw the conclusion that the state of broadband connectivity in the united states continued to improve heading into 2011.

City

Q1 11 Avg. Mbps

1 2 3 4 5 6 7 8 9 10

Riverside, CA Staten Island, NY San Jose, CA Fremont, CA Boston Metro, MA Jersey City, NJ Marietta, GA Anaheim, CA Traverse City, MI Hollywood, FL

7.8 7.8 7.8 7.4 7.1 6.8 6.7 6.7 6.6 6.6

Figure 15: Average Measured Connection Speed, Top United States Cities by Speed

22

2011 akamai technologies, Inc. all rights reserved

State

Q1 11 Peak Mbps

QoQ Change

YoY Change

5 7 8 9 1 4 10

1 2 3 4 5 6 7 8 9 10

Delaware Rhode Island Maine Virginia New Hampshire Hawaii Wisconsin Indiana New York North Carolina

30.1 27.0 24.3 23.5 23.3 23.2 22.6 22.6 22.4 22.0

5.7% -2.7% 11% 10% 1.6% -3.7% 8.8% 1.7% 8.2% 7.7%

16% 21% 33% 33% 9.7% 17% 24% 20% 22% 36%

3 2

Figure 16: Average Peak Connection Speed by State

4.4 United States Average Peak Connection Speeds, City View

topping the list of cities in the united states with the highest average peak connection speeds are cities in New Jersey and New York, at 40 mbps and 35 mbps respectively. similar to the city view of average connection speeds in the united states discussed above, Figure 17 highlights that cities in California hold only four of the top 10 slots for the average peak connection speed as well. the list is rounded out by cities in Florida, Colorado, New Jersey, and texas. In addition to North bergen with its 40 mbps average peak connection speed, three other cities achieved average peak connection speeds in excess of 30 mbps none of the cities in the top 10 in the fourth quarter of 2010 reached this level. all of the remaining cities in the top 10 once again had speeds in excess of 20 mbps.

4.5 United States High Broadband Connectivity

In line with the generally positive trends seen across the united states for average and average peak connection speeds, quarterly changes among the top 10 states for high broadband adoption were all positive in the first quarter of 2011, including three states (Wisconsin, Virginia, Indiana) with growth of more than 10%, as shown in Figure 18. across the entire country, 13 other states also saw quarterly increases of 10% or more, with ohios 32% jump leading the pack, and south Dakota and alaska improving high broadband adoption levels in excess of 20% (25% and 23% respectively). only eight states and the District of Columbia saw high broadband adoption levels decline quarter-over-quarter, with most of the losses relatively modest, except for Hawaiis unexpected 23% decline. as compared to the beginning of 2010, high broadband adoption levels among the top 10 states were generally positive, except for a drop of just over 9% in massachusetts. Very strong growth was seen in rhode Island (72% increase) and maine (69% increase), with Wisconsin and Virginia also growing 30% or more year-over-year. across the whole united states, New Jersey had the largest increase in high broadband connectivity, growing a massive 172% year-over-year. Including those states mentioned above, 28 states saw high broadband adoption grow more than 10% since the same period a year earlier. overall, 42 states experienced a positive yearly change, while seven states and the District of Columbia saw a yearly decline.

City

Q1 11 Peak Mbps

1 2 3 4 5 6 7 8 9 10

North Bergen, NJ Staten Island, NY Van Nuys, CA Riverside, CA Hayward, CA San Mateo, CA Hollywood, FL Arvada, CO Jersey City, NJ Waco, TX

40.0 35.0 32.1 30.4 29.3 28.8 28.6 28.2 27.7 27.6

Figure 17: Average Peak Connection Speed, Top United States Cities by Speed

2011 akamai technologies, Inc. all rights reserved

23

Geography united states (continued)

State % Above 5 Mbps QoQ Change YoY Change

seCtIoN 4:

6 5 9

4 7 2 1 8 10

1 2 3 4 5 6 7 8 9 10

Delaware Rhode Island New Hampshire Maine Wisconsin New York Massachusetts Virginia Indiana South Carolina

72% 65% 58% 45% 43% 43% 42% 42% 41% 41%

7.2% 5.1% 7.3% 1.9% 18% 9.0% 4.2% 15% 12% 7.8%

0.3% 72% 7.4% 69% 31% 6.3% -9.1% 46% 18% 29%

Figure 18: High Broadband Connectivity, Fastest U.S. States

(Vermonts high broadband adoption remained flat yearover-year.) of the states that had lower high broadband adoption, only maryland and the District of Columbia lost more than 10%, down 11% and 25% respectively.

4.6 United States Broadband Connectivity

as Figure 19 illustrates, broadband adoption rates among the top 10 states continued to be remarkably consistent from quarter-to-quarter, with fairly minimal growth seen in nine of the 10 listed states. Four of the states grew less than 1%, while Connecticut remained flat quarter-overquarter. adoption rates among the top 10 remained high, as Delaware, rhode Island, and New Hampshire all had more than 90% of their connections to akamai at speeds above 2 mbps in the first quarter, with Delaware holding steady at the 97-98% range. across the whole country, only six states and the District of Columbia had less than half of their connections below the broadband threshold. Quarter-over-quarter changes across the country as a
State % Above 2 Mbps QoQ Change YoY Change

whole were positive in 39 states, while eight states and the District of Columbia saw broadband adoption rates decline quarter-over quarter. the rates of change were fairly nominal, with increases ranging from 16% in California to just 0.2% in maryland, and losses ranging from 0.2% in south Dakota to 8.8% in the District of Columbia. looking at year-over-year changes, all of the top 10 states saw broadband adoption levels increase as compared to the first quarter of 2010, with growth ranging from just 0.1% in first-place Delaware to 9.0% in second-place rhode Island. across the whole country, New Jersey saw a massive increase for this metric as well, growing 147% year-over-year. Fourteen additional states saw yearly growth of 10% or more, while 38 states overall saw year-over-year growth in their percentage of connections to akamai at speeds over 2 mbps. eleven states and the District of Columbia declined year-overyear, with the District of Columbia seeing the largest decline at 28%. arizona saw no change year-over-year.
7 3 10 9 5 1 8

6 2

1 2 3 4 5 6 7 8 9 10

Delaware Rhode Island New Hampshire Hawaii Connecticut Maine Vermont South Carolina Indiana Wisconsin

98% 93% 92% 87% 85% 84% 81% 80% 79% 79%

0.4% 0.8% 0.7% 4.0% 1.6% 0.6% 2.8% 4.4% 3.6%

0.1% 9.0% 2.8% 5.6% 2.2% 6.2% 5.1% 7.8% 7.5% 8.6%

Figure 19: Broadband Connectivity, Fast U.S. States

24

2011 akamai technologies, Inc. all rights reserved

4.7 United States Narrowband Connectivity

both the short- and long-term trends in the percentage of connections to akamai at speeds below 256 kbps among the top 10 states strongly indicate an ongoing move to higher speed connectivity. (michigans 1.4% year-over-year increase is the lone standout among this group, though the level of change is sufficiently low as to not be of significant concern.) as shown in Figure 20, all of the states in the top 10 saw narrowband adoption rates decline 10% or more quarter-over-quarter, with half declining more than 30%. In addition, all of the states in the top 10, except for michigan, saw narrowband adoption rates decline 10% or more year-over-year, with half declining 50% or more. looking across the whole country, the quarterly trend also tends to support an ongoing move to higher speed connections, with 48 states and the District of Columbia all seeing lower levels of narrowband adoption quarterover-quarter, and all losing 10% or more. some of the most significant shifts were seen in states with fewer than

1,000 unique Ip addresses connecting to akamai at speeds of 256 kbps or less, so a shift of a comparatively small number of Ip addresses can have a big impact. While not as supportive as the quarterly trends, the yearly trends observed in the first quarter also generally indicate a shift away from low-speed connections over time. the District of Columbia and 45 other states saw narrowband adoption rates decline year-over-year, while the remaining five, including michigan, saw nominal increases. In the first quarter, only the District of Columbia had a narrowband adoption level above 5%. Consistent with the prior quarter, 11 states saw narrowband adoption rates of 1% or less. Delaware remained the state with the lowest percentage of connections to akamai below 256 kbps, with just 0.1% at that speed. However, with just a few hundred unique Ip addresses connecting to akamai at that speed, the adoption rate could be quite volatile over time, as the shift of just a few unique Ip addresses to faster or slower speeds could cause a significant change in the adoption rate.

State

% Below 256 kbps

QoQ Change

YoY Change

8 3 4 10 6 9 5 1

1 2 3 4 5 6 7 8 9 10

District Of Columbia Alaska Iowa Colorado Ohio Michigan Georgia Washington Illinois Missouri

5.4% 3.9% 3.9% 3.6% 3.5% 3.1% 3.1% 2.9% 2.8% 2.7%

-31% -23% -18% -18% -32% -18% -32% -10% -34% -34%

-27% -51% -33% -12% -21% 1.4% -54% -56% -54% -56%

7 2

Figure 20: Narrowband Connectivity, Slowest U.S. States

DID YOU KNOW?

In Q1 2008, Washington state had the highest percentage (21%) of connections to Akamai at speeds under 256 kbps, while in Q1 2011, the level had dropped to just below 3%.

2011 akamai technologies, Inc. all rights reserved

25

mobile Connectivity
building on the data presented in previous editions of the State of the Internet report, akamai continues to identify additional mobile networks for inclusion in the report, as well as filtering out networks subsequently identified as having proxy/gateway configurations that could skew results. the source data in this section encompasses usage not only from smartphones, but also laptops, tablets, and other devices that connect to the Internet through these mobile networks. In addition, this edition of the State of the Internet report includes insight into mobile traffic growth and data traffic patterns contributed by ericsson, a world-leading provider of telecommunications equipment and related services to mobile and fixed network operators globally. akamai and ericsson have partnered to develop the first ever end-to-end solution to address performance, scalability, and availability of mobile content and applications on a global scale.28
as has been noted in prior quarters, the source data set for this section is subject to the following constraints:
a minimum of 1,000 unique Ip addresses connecting

seCtIoN 5:

5.1 Attack Traffic From Mobile Networks, Top Originating Countries

In reviewing Figure 21, we find that the distribution of attack traffic sourced in mobile networks during the first quarter of 2011 had a fairly similar distribution to that seen in the prior quarter, though some countries saw slightly higher percentages, while others were slightly lower. Italy remained responsible for the largest percentage of observed attacks, but dropped to 25% (from 30%) this quarter. of the top 10 countries, eight of them were also found on the list last quarter the united Kingdom and russia dropped out of the top 10, while argentina and australia replaced them on the list. overall attack traffic concentration dropped slightly as compared to the fourth quarter of 2010, with the top two countries responsible for 34% of observed attacks (down from 40%), while the top 10 countries were the source of just under three-quarters of observed attacks.

to akamai from the network in the first quarter of 2011 was required for inclusion in the list.
In countries where akamai had data for multiple

network providers, only the top three are listed, based on unique Ip address count.
the names of specific mobile network providers have

been made anonymous, and providers are identified by a unique ID.

Data is included only for networks where akamai

believes that the entire autonomous system (as) is mobile that is, if a network provider mixes traffic from fixed/wireline (Dsl, cable, etc.) connections with traffic from mobile connections on a single network identifier, that as was not included in the source data set.
akamais edgescape database was used for the

geographic assignments.

26

2011 akamai technologies, Inc. all rights reserved

Country/Region

Q1 11 % Traffic

1 2 3 4 5 6 7 8 9 10

Italy Chile Malaysia Australia Poland China United States Hungary Lithuania Argentina Other

25% 9.0% 7.7% 7.2% 6.4% 5.7% 3.9% 3.4% 3.1% 2.9% 26%

1 8 3

4 2 10

Figure 21: Attack Traffic from Mobile Networks, Top Originating Countries/Regions

5.2 Attack Traffic From Mobile Networks, Top Ports

In the first quarter of 2011, nine of the top 10 ports targeted by attack traffic sourced in mobile networks were the same as in the fourth quarter of 2010. In this quarter, port 5900 (VNC server) dropped from the list, replaced by port 443 (Https/ssl). the appearance of port 443 on this list is in line with the massive growth in overall attack traffic targeting the port noted in section 1.2 above. as shown in Figure 22, attack concentration continued to grow in the first quarter, with port 445 responsible

for 80% of the observed attacks, and the top 10 ports responsible for just over 97% of observed attacks (up from 96% in the fourth quarter). Interestingly, China was the only country among the top 10 that did not originate any attack traffic targeting port 445 it was the most targeted port across attacks from the other nine countries. as we have noted in prior reports, we believe that the observed attack traffic that is originating from known mobile networks is likely being generated by infected pC-type clients connecting to wireless networks through mobile broadband technologies, and not by infected smartphones or similar mobile connected devices.

Port

Port Use

Q1 11 % Traffic

445 23 135 80 1433 139 22 443 3389 4899 Various

Microsoft-DS Telnet Microsoft-RPC WWW Microsoft SQL Server NetBIOS SSH HTTPS/SSL Microsoft Terminal Services Remote Administrator Other

80% 5.1% 3.2% 2.0% 1.6% 1.4% 1.3% 1.0% 0.9% 0.7% 2.8%

HTTPS/SSL 1.0% SSH 1.3% NetBIOS 1.4% Microsoft SQL Server 1.6% WWW 2.0% Microsoft-RPC 3.2% Telnet 5.1%

Microsoft Terminal Services 0.9% Remote Administrator 0.7% Other 2.8%

Figure 22: Attack Traffic from Mobile Networks, Top Ports

Microsoft-DS 80%

2011 akamai technologies, Inc. all rights reserved

27

mobile Connectivity (continued)

5.3 Connection Speeds and Data Consumption on Mobile Networks
In the first quarter of 2011, a mobile provider in poland (pl-4) was the provider with the highest average connection speed, at just over 6.1 mbps, bumping the faster provider in the fourth quarter of 2010 (Gr-1) down to second place. readers will note that a provider in belgium (be-3) is listed as having an average connection speed of over 15 mbps, which should put it into first place. However, we believe that this provider may be leveraging a caching or gateway architecture that would be artificially inflating this figure; we will consider removing this provider from the underlying data set in future editions of this report. In reviewing the average connection speeds of the 105 providers listed in Figure 23, excluding the aforementioned belgian provider, we find that pl-4 was the only provider with an average connection speed in the high broadband (>5 mbps) range, while 24 total providers had average connection speeds in the broadband (>2 mbps) range. an additional 45 mobile providers had average connection speeds of 1 mbps or more. the mobile provider with the slowest average connection speed in the first quarter continued to be sK-1, though its average connection speed grew to 163 kbps, up from 134 kbps in the prior quarter. In reviewing quarterly changes, it appears that none of the listed providers saw a doubling of average connection speeds, though Gr-2 came close, growing its average connection speed nearly 97%. only four of the listed providers (Gr-2, es-1, ID-1, uY-1) increased their average connection speeds by 50% or more quarter-over-quarter. Five providers recorded average connection speed growth of under 1%, while two others remained unchanged from the fourth quarter of 2010. In looking at yearly trends, three providers saw year-over-year average connection speed increases of more than 400%, with ukrainian provider ua-1s 481% increase the largest. Interestingly, three of the listed providers saw no change from the same period last year. In reviewing the average peak connection speed data for the first quarter of 2011, we find that a mobile provider in the united Kingdom (uK-3) once again had the highest average peak connection speed, at 22.7 mbps (again, excepting provider be-3 from consideration). of the listed mobile providers, 26 had average peak connection speeds in excess of 10 mbps, while an additional 44 had average peak connection speeds above 5 mbps; all delivered average peak connection speeds of 1 mbps or more. this quarter, a mobile provider in south africa (Za-1) had the lowest average peak connection speed, observed to be just over 1 mbps. In reviewing quarterly changes, only a single provider (Cl-3) saw its average peak connection speed double quarter-over-quarter. However, provider pt-1 in portugal, did come close, though, growing 95%. the largest quarterly average peak connection speed decline was seen on provider NC-1 in New Caledonia, which lost just over 52%. Yearly changes were generally strong, with average peak connection speeds more than doubling at nearly 50 providers. the largest year-over-year declines were seen on providers uK-3 in the united Kingdom and CN-1 in China, which lost 34% and nearly 40% respectively. For the first quarter of 2011, we found that, once again, users of seven mobile providers consumed, on average, more than one gigabyte (1 Gb) of content from akamai per month. (excepting provider Ca-1, which has been confirmed to be using a proxy architecture.) users on an additional 77 providers around the world downloaded more than 100 mb of content from akamai per month during the first quarter, while users on 20 other providers downloaded fewer than 100 mb. (It is interesting to note that provider be-3 is in this group of 20 while its connection speeds would seem to indicate the presence of a proxy, its download volumes do not at least not a direct proxy, though it may be caching content, which could help explain the discrepancy.) Consumption grew more than 100% quarter-over-quarter on only a single provider (pt1), while the largest quarterly decline was seen on slovenian provider sI-1, which lost nearly 65%. Hong Kongs HK-1 saw the largest year-over-year increase in content consumption, gaining over 840%, with provider pt-1 the only other one increasing download volumes more than 500% since the first quarter of 2010.

seCtIoN 5:

28

2011 akamai technologies, Inc. all rights reserved

Country/Region

ID

Q1 11 Avg. kbps

Q1 11 Peak kbps

Q1 11 Avg. MB/ month

Country/Region

ID

Q1 11 Avg. kbps

Q1 11 Peak kbps

Q1 11 Avg. MB/ month

AFRICA Egypt Morocco Nigeria South Africa ASIA ASIA China Hong Kong Hong Kong Indonesia Israel Japan Kuwait Malaysia Malaysia Pakistan Qatar Saudi Arabia Singapore Singapore Singapore South Korea Sri Lanka Taiwan Taiwan Thailand EURoPE EUROPE Austria

EG-1 MA-1 NG-1 ZA-1 CN-1 HK-2 HK-1 ID-1 IL-1 JP-1 KW-1 MY-1 MY-3 PK-1 QA-1 SA-1 SG-3 SG-4 SG-5 KR-1 LK-1 TW-1 TW-2 TH-1 AT-1

482 1118 267 364 1135 1325 2618 459 1873 1613 1857 590 879 876 2061 2021 1382 1858 650 1176 701 1280 742 453 3392

3001 9815 4174 1024 2870 7251 16358 8070 7945 8240 8428 3861 6387 5644 8697 8768 7341 8804 5772 4134 5311 6187 3308 4154 14619

245 414 391 176 162 492 3744 2966 79 168 276 152 366 572 209 310 672 309 467 101 307 136 154 90 195

Austria Belgium Belgium Belgium Czech Republic Czech Republic Czech Republic Estonia France Germany Germany Germany Greece Greece Hungary Hungary Ireland Ireland Ireland Italy Italy Italy Lithuania Lithuania Moldova Netherlands Netherlands Norway

AT-2 BE-1 BE-3 BE-2 CZ-1 CZ-3 CZ-2 EE-1 FR-2 DE-1 DE-2 DE-3 GR-1 GR-2 HU-2 HU-1 IE-1 IE-2 IE-3 IT-3 IT-2 IT-4 LT-2 LT-1 MD-1 NL-2 NL-1 NO-2

2649 2623 15366 1744 1639 3296 849 1058 1988 843 3988 1520 4560 798 2307 1651 2685 1732 1734 2913 3565 1030 1543 2248 1484 2212 1529 1717

15464 10229 43141 4358 7925 9176 4599 5279 7084 3516 11735 6468 17794 4823 11935 11835 12531 12925 12816 12304 17303 6720 9684 12395 6437 5749 4016 5988

756 482 19 20 80 235 169 311 1420 78 1970 141 390 155 126 193 489 632 788 568 437 215 378 525 129 30 31 61

Figure 23: Average and Average Peak Connection Speed, Average Megabytes Downloaded per Month by Mobile Provider

DID YOU KNOW?

During the first quarter of 2011:

In the United States, for the first time, 51% of the devices sold were smartphones. Globally, the average is 26%. One-third of all smartphones sold were sold in the United States. Smartphones now account for 80% of the revenue of all phones sold in the United States.

[Source: http://www.chetansharma.com/blog/2011/05/09/us-mobile-data-market-update-q1-2011/]

2011 akamai technologies, Inc. all rights reserved

29

mobile Connectivity (continued)

seCtIoN 5:

Country/Region

ID

Q1 11 Avg. kbps

Q1 11 Peak kbps

Q1 11 Avg. MB/ month

Country/Region

ID

Q1 11 Avg. kbps

Q1 11 Peak kbps

Q1 11 Avg. MB/ month

EURoPE (CoNTINUED Norway Poland Poland Poland Portugal Romania Russia Russia Russia Slovakia Slovakia Slovenia Spain Spain Spain Turkey Ukraine Ukraine United Kingdom United Kingdom United Kingdom NORTH AMERICA NoRTh AMERICA Canada Canada Curacao El Salvador El Salvador

NO-1 PL-1 PL-2 PL-4 PT-1 RO-1 RU-3 RU-4 RU-2 SK-1 SK-2 SI-1 ES-1 ES-3 ES-2 TR-1 UA-1 UA-2 UK-3 UK-2 UK-1 CA-2 CA-1 CW-1 SV-2 SV-1

1127 3742 1456 6151 711 674 1452 2571 857 163 2133 1825 3385 950 958 1934 1045 1619 4206 2413 1605 1051 3174 564 1601 1044

4900 11952 7323 14893 4311 3947 5235 10058 3777 1301 9340 6474 19654 5327 3235 9098 3101 6709 22703 11194 11275 2738 20058 3551 8870 5627

65 141 77 141 217 85 125 333 92 36 1825 35 417 134 804 217 69 128 105 969 677 614 23404 295 655 300

El Salvador Guatemala Guatemala Nicaragua Puerto Rico United States United States United States OCEANIA oCEANIA Australia Australia Guam New Caledonia New Zealand SOUTH AMERICA SoUTh AMERICA Argentina Argentina Bolivia Brazil Brazil Chile Chile Colombia Paraguay Paraguay Uruguay Uruguay Venezuela

SV-3 GT-2 GT-1 NI-1 PR-1 US-2 US-1 US-3 AU-3 AU-1 GU-1 NC-1 NZ-2 AR-1 AR-2 BO-1 BR-1 BR-2 CL-4 CL-3 CO-1 PY-2 PY-1 UY-1 UY-2 VE-1

622 1059 893 1278 2230 1092 1759 1007 1601 1201 538 674 1445 638 752 214 741 792 960 1502 1000 307 564 1842 456 752

3237 6791 5434 7426 9478 3930 4468 2964 8109 10149 2595 2074 8131 4066 4629 3004 4163 4678 6766 14242 6893 3312 5441 15788 3879 5094

348 686 188 608 2249 39 103 547 243 1640 80 515 544 94 152 191 156 174 467 203 283 314 172 279 67 171

Figure 23 (Continued)

30

2011 akamai technologies, Inc. all rights reserved

5.4 Mobile Traffic Growth As Observed by Ericsson

In mobile networks, the access medium (spectrum) is being shared by different users in the same cell. It is important to understand traffic volumes and usage patterns in order to enable a good customer experience. ericssons presence in more than 180 countries, and its customer base representing more than 1,000 networks enables ericsson to measure mobile voice and data volumes. the result is a representative base for calculating world total mobile traffic in 2G, 3G, and 4G networks (not including DVb-H, WiFi, and mobile Wimax). these measurements have been performed for several years, pointing at a stable trend of traffic growth. However, the measurements of data and voice traffic in these networks (2G, 3G, 4G/lte) around the world show large differences in traffic levels between markets and regions, and also between operators due to their different customer profiles. as illustrated in Figure 24, mobile data surpassed voice on a global basis in Q4 2009. this finding is based on ericsson measurements from live networks covering all regions of the world. the growth rate in mobile data traffic from Q1 2010 to Q1 2011 was 130%.

5.5 3G Data Traffic Patterns of Multiple Connected Device Types As Observed by Ericsson
a significant part of mobile Internet traffic is transferred over 3G mobile access networks. Figure 25 shows the volume (average values for networks with the smallest and the highest usage) of monthly data traffic per subscriber delivered over 3G networks for laptop, tablet and smartphone devices (including android-based phones and apple iphones). as the figure shows, the average monthly traffic volumes per subscription over 3G access are undoubtedly the highest for laptop terminals (1 7 Gb) followed by tablets (250 800 mb) and smartphones (80 600 mb).
7 GB / month / subscriber 6 5 4 3 2 1 0 laptop tablet smartphone

Figure 25: Monthly Traffic Volumes in 3G Mobile Broadband Networks per Subscriber

400
Voice Data

Total (UL + DL) monthly traffic (Exabyte/month)

350

300

250

200

150

100

50

0 Q1 07 Q2 07 Q3 07 Q4 07 Q1 08 Q2 08 Q3 08 Q4 08 Q1 09 Q2 09 Q3 09 Q4 09 Q1 10 Q2 10 Q3 10 Q4 10 Q1 11

Figure 24: Total Monthly Mobile Voice and Data as Measured by Ericsson

2011 akamai technologies, Inc. all rights reserved

31

mobile Connectivity (continued)

using 3G data traffic measurements performed by ericsson from all major regions of the world, Figure 26 shows how the most widely used online applications contribute to overall mobile Internet traffic volumes, and how these contributions vary by the type of connected device, based on estimated worldwide average values from all measured networks. Note that WiFi offload traffic is not included in underlying data for Figures 25 or 26. regardless of device type, online video (30 40%) is the largest contributor to traffic volumes, followed by Web
100% GB / month / subscriber 80% 60% 40% 20% 0% laptop
online audio online Video Web browsing

seCtIoN 5:

browsing (20 30%). on laptop-type devices, the amount of file sharing traffic can also be significant. on tablets and smartphone devices, online audio, e-mail, software downloads, and social networking traffic are also important contributors of 3G data traffic. It is also interesting to examine the difference in mobile data traffic patterns generated by laptops, tablets and smartphones. Figure 27 shows sample network traffic for a 24-hour period (one day), where each horizontal line corresponds to one subscriber, and shading along these lines represent data traffic via 3G access from these subscribers along the timeline. as illustrated by the figure, laptops have a few longer sessions mainly during daytime and the evening, but at dawn, most laptops are turned off. an examination of the underlying data for a selected five-minute period shows that active sessions for laptop-type devices are characterized by longer bursts of intensive usage from interactive applications (such as online video and Web browsing) and shorter low bandwidth data transmissions from background applications (such as instant messaging). In contrast, tablet and smartphone devices usually have frequent and short sessions typically during the whole day, sometimes showing a periodic nature. as shown in Figure 28, an examination of these sessions shows that they consist of low bandwidth background data transmission bursts (such as presence updates and periodic email checking), interspersed with a few more intensive interactive usage bursts.

tablet
social Networking email File sharing

smartphone
software Downloads other

Figure 26: Average Application Volume Shares in Mobile Broadband Networks for Different Device Types

Figure 27: Mobile Broadband Traffic Patterns Visualized

32

Each Line = Traffic of One Randomly Selected Subscriber

2011 akamai technologies, Inc. all rights reserved

Each line = traffic of one randomly selected subscriber

1 < 1 kbps 1 10 kbps

2 Time (min.) 10 100 kbps

4 100 kbps 1 Mbps

5 > 1 Mbps

Figure 28: Data transmission patterns from laptops, tablets, and smartphones as examined over a five-minute period

It is interesting to note that tablet traffic patterns over some portion of tablet and smartphone traffic 3G mobile networks are much closer to smartphone is offloaded from 3G to WiFi. traffic patterns than to laptop traffic patterns. For ex monthly data subscription caps are often smaller ample, on tablets, one could expect online video usage for tablets and smartphones than for laptops more similar to smartphones than to laptops. these traffic pattern similarities could potentially be due to several the smaller screens on smartphones and tablets different factors: (as compared to laptops) may result in a more limited video and Web browsing experience 3G access for laptops is often used as the main avenue for Internet access (replacing a wired connection), while tablets and smartphones are often used as secondary devices.

Smartphone
2011 akamai technologies, Inc. all rights reserved

Tablet

Laptop

33

Network outages and Disruptions

In the first quarter of 2011, Internet outages or disruptions of note occurred in several countries around the world due to government action in response to protests, natural disasters, or oddly enough, simply scavenging for scrap metal. In the sections below, we review some notable outages as seen by trends in traffic delivered by akamai into these countries.
6.1 Egypt
according to published reports, in late January, the egyptian government took the unprecedented step of cutting off nearly all of the countrys Internet access in response to widespread protests.29 the countrys government ordered each of the four major Internet service providers within the country to shut down all international connections to the Internet on January 28. the shutdown was significant, cutting off an estimated 20 million egyptian Internet users; as a blog post from Internet monitoring firm renesys noted, every egyptian provider, every business, bank, Internet cafe, website, school, embassy, and government office that relied on the big four egyptian Isps for their Internet connectivity is now cut off from the rest of the world.30 the egyptian government ended the shutdown several days later, on February 2. as shown in Figure 29, the volume of traffic being delivered by akamai to users in egypt dropped precipitously concurrent with these providers being shut down, and remained at near-zero levels for the duration of the outage. Interestingly, upon restoration of Internet access, traffic peaked at a level slightly higher than that seen immediately prior to the shutdown, likely related to pent-up demand for news, social networking, and other types of content from egyptian users.

seCtIoN 6:

2/ 1/ 20 11

1/ 27 /2 01 1

1/ 28 /2 01 1

1/ 29 /2 01 1

1/ 30 /2 01 1

Figure 29: Multiday Internet Outage in Egypt lasted from January 27 through February 2

34

2011 akamai technologies, Inc. all rights reserved

1/ 31 /2 01 1

2/ 2/ 20 11

6.2 Libya
Following the Internet outages in egypt that occurred in response to widespread protests, political unrest in libya drove two brief disruptions in Internet connectivity in libya in the third week of February, followed by a longer disruption that started in early march. according to published reports, two brief outages occurred during the February 18-20 period, after which Internet traffic returned at a level 60-80% of that seen prior to the disruption.31 the graph of akamai traffic delivered into libya during the two week February 14-28 period, shown in Figure 30, has two clearly identifiable outage periods during the 18th-20th, and the peak traffic levels after those outages were a fraction of those seen in the days before.

additionally, on march 3rd, as shown in Figure 31, akamai traffic delivered into libya fell to near-zero levels, and remained that way for over a week. a blog post from Internet monitoring firm renesys found that, during this period, nearly every host inside libya had become unresponsive. renesys noted, You could attempt to ping them, send a traceroute along the path to them, try to retrieve pages, try to look up domain names ... but in nearly every case, there was no response. 32 similar to what occurred in egypt, the libyan government apparently ordered libya telecom & technology to throttle the flow of traffic to the point of uselessness, rather than turning it off entirely33 this would likely explain why akamais traffic into the country did not drop completely to zero during this period.

2 4/ /1

1 01 2

2 5/ /1

1 01

/2 16 2/

01

1 1 1 1 1 1 1 1 1 1 1 1 1 01 01 01 01 01 01 01 01 01 01 01 01 01 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 1 28 27 26 25 24 23 22 21 20 19 18 17 3/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/

Figure 30: Two brief disruptions in Internet connectivity in Libya occurred during the third week of February

/2 /1

1 01 3

1 01 /2 /2

/ /3

1 20

2 4/ 3/

01

1 5/ 3/

20

11 3

0 /2 /6

11 3

/ /7

20

11 3

/ /8

20

11 3

11 20 9/ /

Figure 31: A longer disruption in Libyas Internet connectivity occurred in early March

2011 akamai technologies, Inc. all rights reserved

35

Network outages and Disruptions (continued)

proximately 27% immediately after the earthquake occurred but began to recover shortly thereafter. While on march 11, a magnitude 8.9 earthquake struck not shown in the figure, akamais monitoring indicated northeast Japan and spawned a tsunami with waves that 34 that in the days after the earthquake, the peak levels reached up to six miles inland. Despite damaging undersea cables belonging to telecommunications provider KDDI, for traffic delivered to Japan were higher than those seen these natural disasters did not otherwise cause widespread prior to the earthquake, likely due to Japanese citizens turning to the Internet for updated news and information or long-term Internet outages within Japan. as shown in on the disaster, as well as in efforts to find and communiFigure 32, akamai traffic delivered to Japan dropped apcate with friends and family.

seCtIoN 5:

6.3 Japan

1 20 0/ 1 3/

: 15

50 2 0/ /1 1 01

18

4 :1 /2 10 1 01

:3 20

8 /2 10 01 1

23

:0

2 / 10 3/ 20 11

1:

26 3/ / 10 20

11

3:

50 / 10 3/ 20

11

6:

14 3/ / 10 20

11

38 8: 3/ 1 2 0/ 01

11

:0

3/

3/

Figure 32: The earthquake occurred at 05:46 UTC, and resulted in a 27% drop in traffic

DID YOU KNOW?

In January 2008, a pair of cut submarine telecom cables in the Mediterranean just north of Egypt caused severe Internet outages and disruptions in the Middle East, Pakistan and India. In September 2008, Hurricane Ike caused extensive Internet outages across the United States. In December 2008, three key submarine cables in the Mediterranean were severed, which impacted Internet traffic in the Middle East and Indian subcontinent.

36

2011 akamai technologies, Inc. all rights reserved

6.4 Georgia/Armenia
While other notable Internet outages and disruptions in the first quarter were due to natural disasters or government action, published reports indicate that disruptions seen in late march in the eastern european countries of Georgia and armenia had a far more unusual cause. these reports35 claim that a 75-year old Georgian woman searching for scrap metal cut a fiber-optic cable belonging to Georgian railway telecom, which caused 90 per cent of private and corporate internet users in neighboring armenia to lose access Georgia

for nearly 12 hours while also hitting Georgian internet service providers. Figure 33 illustrates patterns in traffic that akamai delivered to Georgia and armenia between march 26th and 31st. as can be seen in the highlighted areas, otherwise cyclical traffic in both countries saw uncharacteristically large declines on march 28, likely related to the disruption in Internet access caused by this severed cable, as Web users were left twiddling their thumbs for up to five hours as the countrys main internet providers were prevented from supplying their normal service.36 Armenia

26. mar

28. mar

30. mar

26. mar

28. mar

30. mar

Figure 33: Note the uncharacteristically large dips in Akamai traffic delivered to Georgia (L) and Armenia (R) on March 28

DID YOU KNOW?

In June 2009, Internet connectivity in Iran experienced disruptions related to unrest around controversial elections within the country. In July 2009, damage to the undersea SAT-3 cable caused Internet connectivity problems in West Africa. In 2010, damage to submarine cables caused Internet outages in Haiti (January), Taiwan (March), the Middle East (April), and Malaysia (April).
37

2011 akamai technologies, Inc. all rights reserved

appendix
Country/Region
* Countries listed with had fewer than 25,000 unique IP addresses connecting to Akamai during the first quarter at this speed. Based on the revised threshold for inclusion, they were not included in the global ranking.

seCtIoN 6:

% Attack Traffic

Unique IP Addresses

Avg. Connection Speed (Mbps)

Peak Connection Speed (Mbps)

% Above 5 Mbps*

% Above 2 Mbps*

% Below 256 kbps*

EURoPE Austria Belgium Czech Republic Denmark Finland France Germany Greece Hungary Iceland Ireland Italy Luxembourg Netherlands Norway Poland Portugal Romania Slovakia Spain Sweden Switzerland United Kingdom ASIA/PACIFIC Australia China Hong Kong India Japan Malaysia New Zealand Singapore South Korea Taiwan MIDDlE EAST Egypt Israel Kuwait Saudi Arabia Sudan Syria United Arab Emirates (UAE) lATIN & SoUTh AMERICA Argentina Brazil Chile Colombia Mexico Peru Venezuela NoRTh AMERICA Canada United States

0.1% 0.1% 0.2% 0.1% 0.1% 1.2% 2.1% 0.2% 1.8% <0.1% 0.1% 2.5% <0.1% 0.2% <0.1% 1.9% 0.2% 2.5% 0.1% 0.8% 0.2% 0.1% 0.7% 0.4% 6.4% 3.3% 3.8% 1.7% 1.0% 0.3% 0.6% 1.2% 9.1% 1.3% 0.5% 0.1% 0.2% <0.1% <0.1% 0.2% 1.5% 5.5% 0.4% 0.6% 0.3% 1.3% 0.3% 0.9% 10%

2,861,052 3,920,493 2,054,599 2,463,216 2,660,860 24,010,722 34,785,032 2,459,685 2,111,588 130,615 1,551,482 13,632,661 173,290 8,166,009 3,030,551 6,575,834 2,526,492 2,531,466 797,784 12,915,356 6,103,986 2,972,087 22,333,025 11,749,126 73,587,347 2,478,786 6,974,771 41,233,145 2,045,067 1,562,272 1,362,513 22,538,305 7,782,733 1,330,239 2,168,339 357,971 2,192,288 29,581 221,394 915,638 4,745,447 14,153,991 2,580,353 2,858,414 8,930,278 738,094 2,308,463 12,583,683 142,605,731

4.4 6.1 6.5 5.6 4.9 3.6 4.7 3.5 5.1 5.1 5.6 3.7 4.5 7.5 5.4 3.6 4.9 6.6 4.8 3.4 5.0 6.2 4.6 3.4 1.0 9.2 0.8 8.1 1.6 3.5 4.2 14.4 4.1 0.8 3.6 1.5 2.0 0.6 1.8 3.9 1.8 1.7 2.6 1.9 1.9 1.4 0.8 5.6 5.3

15.7 24.7 20.7 17.4 16.5 14.2 18.3 17.0 21.4 21.2 18.6 14.9 16.2 22.0 18.5 13.9 24.9 32.7 18.4 15.7 19.1 21.1 17.2 14.7 4.1 39.5 5.2 29.9 8.9 13.7 19.3 36.3 18.3 7.0 15.9 9.0 8.0 5.3 4.1 25.9 10.7 8.9 14.7 9.6 9.5 8.6 4.8 20.1 21.2

24% 52% 48% 43% 31% 13% 27% 9.1% 35% 26% 29% 11% 20% 56% 35% 17% 37% 51% 20% 11% 29% 40% 25% 16% 0.5% 55% 0.4% 55% 2.4% 15% 28% 60% 24% 11% 28% 3.6% 3.5% 4.3% 1.1% 44% 39%

71% 90% 94% 87% 68% 79% 91% 80% 91% 87% 81% 85% 92% 91% 82% 65% 88% 93% 90% 79% 66% 93% 89% 57% 9.3% 92% 4.9% 79% 15% 74% 69% 87% 75% 4.4% 83% 16% 41% 44% 53% 27% 27% 65% 37% 32% 11% 1.7% 88% 77%

0.3% 0.6% 0.9% 0.4% 0.6% 0.7% 0.6% 2.2% 7.9% 35% 1.1% 2.7% 4.5% 0.5% 9.4% 20% 2.3% 9.1% 1.3% 1.5% 9.4% 1.3% 2.0%

38

2011 akamai technologies, Inc. all rights reserved

endnotes
1

seCtIoN 7:

http://bit.ly/jJe5vg http://www.confickerworkinggroup.org/wiki/uploads/ Conficker_Working_Group_lessons_learned_17_June_ 2010_final.pdf http://www.grc.com/port_21.htm, http://isc.sans.org/ port.html?port=21, http://www.ictsc.com/Ip_port21.htm http://www.iana.org/assignments/port-numbers https://www.torproject.org/docs/tor-doc-windows.html.en http://news.cnet.com/8301-30685_3-20030105-264.html http://www.networkworld.com/community/blog/ microsoft-pays-nortel-75-million-ipv4-address http://gcn.com/articles/2011/03/04/Ipv4-aftermarketfor-usused-address-space.aspx?p=1 http://www.apnic.net/publications/news/2011/final-8 http://www.apnic.net/policy/add-manage-policy#9.10 http://www.potaroo.net/ispcol/2011-02/transtools-part1.html http://mailman.nanog.org/pipermail/nanog/2011-February/031788.html http://s3.amazonaws.com/alexa-static/top-1m.csv.zip http://mnlab-ipv6.seas.upenn.edu/monitor/index.html http://www.google.com/intl/en/ipv6/ http://asert.arbornetworks.com/category/ipv6/ http://www.google.com/intl/en/ipv6/statistics/ [Graph time scale narrowed to January 1, 2011 march 31, 2011] http://www.akamai.com/dl/whitepapers/How_will_the_ internet_scale.pdf http://www.crtc.gc.ca/eng/com100/2011/r110503.htm http://ec.europa.eu/information_society/newsroom/cf/pillar. cfm?pillar_id=46 http://www.nbn.gov.au/frequently-asked-questions/#a558

22

http://www.blu-ray.com/faq/ the average peak connection speed metric represents an average of the maximum measured connection speeds across all of the unique Ip addresses seen by akamai from a particular geography. the average is used in order to mitigate the impact of unrepresentative maximum measured connection speeds. In contrast to the average measured connection speed, the average peak connection speed metric is more representative of what many end-user Internet connections are capable of. (this includes the application of so-called speed boosting technologies that may be implemented within the network by providers, in order to deliver faster download speeds for some larger files.) http://www.digitaldivide.org/wp-content/uploads/2011/04/ broadband_declaration_final_version__english_signed1.pdf http://www.telegeography.com/products/commsupdate/ articles/2010/12/09/trai-releases-national-broadband-plan/ Ibid. http://www.akamai.com/html/technology/products/ edgescape.html http://www.akamai.com/ericsson/index1.html http://www.capacitymagazine.com/article/2781237/ egypt-cuts-internet-in-attempt-to-silence-protests.html http://www.renesys.com/blog/2011/01/egypt-leaves-theinternet.shtml http://news.cnet.com/8301-31921_3-20035079-281.html http://www.renesys.com/blog/2011/03/what-libya-learnedfrom-egypt.shtml Ibid. http://articles.cnn.com/2011-03-11/world/japan.quake_1_ hokkaido-tsunami-east-japan-railway?_s=pm:WorlD http://www.telegraph.co.uk/news/worldnews/europe/georgia/8442056/Woman-who-cut-internet-to-Georgia-andarmenia-had-never-heard-of-web.html http://www.guardian.co.uk/world/2011/apr/06/georgianwoman-cuts-web-access

23

24

25

10

26

11

27

12

28

13

29

14

15

30

16

31

17

32

18

33

19

34

20

35

21

36

2011 akamai technologies, Inc. all rights reserved

39

the akamai Difference

akamai makes the Internet work for some of the best-known companies in the world with its solutions for cloud computing, ecommerce, websites, HD video and software-as-a-service. Delivering unmatched performance, scale and security, akamais solutions are built on the akamai Intelligent Internet platform, unique in the industry due to its rich functionality and intelligence and because it is globally distributed across 650 cities, in 72 countries and integrated into about 1,000 of the Internets most important networks. to learn more visit www.akamai.com or follow us on twitter @akamai.

acknowledgements
EDITOR: David belson CONTRIBUTOR: Jon thompson CONTRIBUTOR: patrick Gilmore CONTRIBUTOR: alloysius Gideon EXECUTIVE EDITOR: brad rinklin EXECUTIVE EDITOR: tom leighton Please send comments, questions, and corrections to stateoftheinternet@akamai.com Follow @akamai and @akamai_soti on

Akamai Technologies, Inc.

U.S. Headquarters 8 Cambridge Center Cambridge, ma 02142 tel 617.444.3000 Fax 617.444.3001 u.s. toll-free 877.4aKamaI (877.425.2624) www.akamai.com International Offices unterfoehring, Germany paris, France milan, Italy london, england madrid, spain stockholm, sweden bangalore, India sydney, australia beijing, China tokyo, Japan seoul, Korea singapore
2011 akamai technologies, Inc. all rights reserved. reproduction in whole or in part in any form or medium without express written permission is prohibited. akamai and the akamai wave logo are registered trademarks of akamai technologies, Inc. other trademarks used herein may be owned by other companies and are used for descriptive purposes only. akamai believes that the information in this publication is accurate as of its publication date; such information is subject to change without notice.