Kapil Arora et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol.

5 (3) , 2014, 3232 - 3235

Web Cache Page Replacement by Using LRU and

LFU Algorithms with Hit Ratio: A Case Unification
Kapil Arora1, Dhawaleswar Rao Ch 2
1

M.Tech, Department of Computer Science & Engineering,

Lovely Professional University, Phagwara, India

Assistant Professor, Department of Computer Science & Engineering,

Lovely Professional University, Phagwara, India

ABSTRACT- In the information age, due rapid growth of the

web and its protocols, a huge transfer of information take place
over the Internet, which leads to enormous transmission traffic,
which increases the load, latency time, network traffic, and
bandwidth consumption. These parameters can be considered as
problems arise due to the enormous traffic over the Internet. To
overcome these kinds of traffic problems, we can adopt several
measures and web caching is part of them. Web caching
enhances the performance and offers an efficient access to the
web content. Different algorithms such as least recently used
(LRU), least frequently used (LFU), most recently used (MRU)
etc., can be used. In this paper, we present web cache page
replacement algorithm and comparison between LRU and LFU
using the caching with respect to pages and the system we has
proposed.
Keywords--Architecture,
Performance
Metrics,
Page
Replacement Policy, Least recently used, Least frequently Used.

I.
INTRODUCTION
The World Wide Web (WWW) is increasely being used by
people all over the world and this lead to an enlarge in traffic
in network. Consequently, this high traffic will increase the
server load, latency time and reduce the bandwidth for
competing request. To overcome this situation, Web caching
technique has been used. Web cache reduces the high traffic
over the internet so that user can access the web content
faster. The main purpose of cache is to place the copy of
object near to the client, so that web user can access the
object easily, without the request going to the web server.
There are different points at which we can set up a cache such
as browser, proxy server and close to server. When a user
request for the web page, firstly it is checked in cache, if the
requested web page is available then it send back to the user.
If the web page is not found in the cache, then the request
redirect the web server and sends the response to the client. In
between cache store the requested web page.
Because of the partial size of memory of cache, it becomes so
much hard to save all objects in the memory. So, to resolve
this problem, we have to use the page replacement
algorithms. These algorithms are being used to evict the
object from the cache and create a space for the new web

object to store. Algorithm such as Least Recently Used

(LRU), Least frequently Used (LFU), Lowest Relative Value
(LRV), Most recently used (MRU) are most popular. These
algorithms assist in improving the performance of cache. A
brief description on various web cache page replacement
algorithms will be discussed in this paper.
II.
ARCHITECTURE
To monitor request from client , a web cache is placed
between server and client. Instead of sending the request to
the server, it will serve the request from the cache. This will
improve the latency and saved the bandwidth as well. It will
cache the web document for the future used.
In order to reduce the numerous cost metrics such as average
latency, byte hit ratio, hit ratio, several caching algorithm
have been proposed. Cache page replacement algorithm is
classified into categories.
Publisher

Web
server

Hitsandmisses
resolved

Hitsandmisses
resolved

ISP
ProxyWeb
Cache

Entry
web
cache
Hit
resolved

Direct
resolved
Entry
web
cache

Entry
web
cache
Hit
resolved

Hitsandmisses
resolved

Consumer

Consumer
Local
cache

Consumer

Fig1. Web Cache Architecture [2]

3232

Kapil Arora et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (3) , 2014, 3232 - 3235

III.
PERFORMANCE METRICS
To accomplish the objective, page replacement policy
depends on the several key metrics. Based on this
performance metric we can compare the performance of
different algorithms. These performance metrics are plays a
very important role in the web cache performance calculation.
Such replacement policy aims to optimize performance
metrics. To evaluate the performance of the cache,
performance metrics are being used. The most commonly
used are Hit rate, byte hit rate bandwidth saved, delay saving
ratio are most commonly used.
A. Hit rate
The percentage of all requests object which are found in the
cache instead of transferred from the requested server.
B. Byte hit
The percentage of all data that is transfer straight from the
cache rather than from requested server.
Metric
Hit Ratio
Byte Hit Ratio
Saved Bandwidth
Delay Saving Ratio

Definition
h
f
S .h
S .f
This is directly related to byte hit
ratio.
d .h
d .f
d . 1 h |f
R

Average downloads
time.
Notation:
Si =size of an object i.
fi =The total number of request for the object i
hi =a number of object hit for object i.
di =The delay which are occur to retrieve the object from
server.
R= Set of object which are accessed.
||R||=size of R
Fig 2.Performance metrics [5]

C. Bandwidth Saved
Bandwidth saved is used to reduce the amount of the
bandwidth consumed.
D. Delay Saving Ratio
The measure the latency of fetching an object.
These performance metrics are the heart of web caching
algorithm. These performance metrics are used to evaluate the
performance of the replacement algorithms with respect to
object which are present and requested in memory of cache,
the saved bytes due to no retransmission, the decrease in
latency to retrieve an object which is requested..

I.
Performance metrics parameters
There are some parameters which we inured to evaluate the
performance.
A. Throughput
A number of requests which is generate by the client per
second
B. Hit ratio
Ratio of the total amount of object found in the cache to the
total number of object is requested.
C. Byte hit ratio
Ration of the total amount of bytes which are served by the
cache to total number of bytes which are sent to clients.
D. Cache age
Time after which the cache becomes full.
E. Downtime
Time taken to improve from the cache failures. [1].
II.
Category of cache page replacement policy
Cache page replacement policies is classify in the following
category:
A. Recency-based
This category of algorithm work on the time basis i.e. time to
access the last references of object. The algorithm of this
category is LRU (Least Recently Used) and has been
functional in a number of proxy caching servers.
B. Size-based
The size based cache replacement policy consider the object
size as the basic parameter. LFU-Size based algorithm is
come under this category.
C. Frequency-based
The Frequency-based cache replacement policy is work on
the frequency of the object means that the number of times
an object is accessed. The algorithm of this category is LFU .
D. Function-based
A cost based function is used to determine the functional
based algorithms. It involves the multiple parameters which
are related to performance metric we used. Most indicative
algorithm of this category is Greedy-Dual Size. [3].
VI.
CACHE REPLACEMENT POLICY
Cache replacement policy plays a crucial part in the web
caching. To achieve the high sophisticated cache mechanism,
these replacement algorithms is necessary. These replacement
policies helpful in evicted the object from the cache and build
a new space for the incoming object. Due to limited size of
the cache, a cache cannot store the entire requested object. So,
we use the cache replacement policy to let the room for new
document. This is applicable when the cache is full of object
and then we have to insert the new object in to cache. So we
have to evict object from the cache to make space. There are
different cache replacement policies which are playing an
important role in the web cache. These algorithms are:
A. Least Recently Used (LRU)
Least Recently Used page replacement policy is simple and
easy to use. This algorithm is work on the time-stamp. It
removes the last recently used object that was not used for a

3233

Kapil Arora et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (3) , 2014, 3232 - 3235

longest time when the cache exceeds its maximum size and
put a new object in place of evicted object. In case, if the
cache size is not full then it will insert the object in the cache
memory. For example, let we have 14 page, then the LRU
work as:
7

0
0
7

1
1
0
7

2
1
0
2

0
1
0
2

3
3
0
2

0
3
0
2

4
3
0
4

2
2
0
4

3
2
3
4

0
2
3
0

3
2
3
0

2
2
3
0

1
2
3
1

Fig 3.Example of LRU page replacement policy

B. Least Frequently Used (LFU)
Least Frequently Used (LFU) page replacement policy which
removes page which are least frequent used and then put a
new requested page in free space. It is very simple and easy to
use. In this, we have to maintain a counter which will count
the frequency of the page. The page which is less frequent to
use is replaced by the new incoming page. For example, we
have 14 page and LFU work as:
7

0
7

1
1
0
7

2
1
0
2

0
1
0
2

3
3
0
2

0
3
0
2

4
3
0
4

2
2
0
4

3
2
0
3

0
2
0
3

3
2
0
3

2
2
0
3

1
2
0
1

Fig 4. Example of LFU page replacement policy

C. Most Recently Used (MRU) page replacement policy
The Most Recently used algorithm evicts the most recently
used document from the cache. This algorithm is used where
we have to access the historical information.[2].

Drop From Head

FIFO Queue
Newly
Requested
Page

D. Lowest Relative Value (LRV)

As the name suggest, lowest relative document value is first
evict from the cache. To calculate an utility, to each object we
assign an relative cost value. The object which is having an
very low relative value is evicted first from the cache and a
new entity will place in that space.
E. LRU-MIN
To minimize the replacement, this algorithm keeps a smaller
size of object in the cache. If there is any object with Size S in
the cache, then we follow the LRU algorithm to evict least
recent used object from the cache. If there are no such an
object which is having size S in the cache, then this algorithm
evict the object of size S/2 in the Least recently used order.
LRU-Threshold
This is similar to LRU, but with a subtle difference is that, a
object which is largest than a threshold size is not inert into
the cache.
F. Greedy dual-Size (GD-SIZE)
It is a generalized version of LRU. This algorithm is
concerned when we have an equal size of an object and the
different cost is incurred to retrieve from the secondary
storage [4].

VII.
COMPARISON OF LRU AND LFU
As, we consider the page fault in the case of least recently
used (LRU) and least frequently Used(LFU)
in fig.3 and fig 4.There are 10 page fault in the Least recently
used and there are 9 page fault in the Least Frequently used.
So as we consider the performance, then we can conclude that
LFU perform better than LRU due to the less page fault.
VIII.
PROPOSED SYSTEM
In this, I have included a threshold value i.e. TSD to evict the
historical object which are not been used by the long time. I
have used both recency and frequency in this system. If any
object which are placed in the memory and is not used since
from a long interval of time, then it has to be remove that
object to make a room for the new object.
Frequency division (F.D.) is used to calculate the average
frequency. When a new object is entering, it checks the least
time and check the priority with the frequency division. If the
priority is greater than frequency division than we look for the
second smallest timestamp. If the difference between both
the time-stamp is greater than TSD, then we remove the first
least document else if not, then we calculate the priority with
the frequency division. If the priority is greater than the
frequency division than remove the first document else
second.

Fig. 5. Example of MRU page replacement policy [2]

3234

Kapil Arora et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (3) , 2014, 3232 - 3235

IX.
CONCLUSION
Web cache is used to increase the performance of the system
by reduce the server load in term of the requested object,
latency time. In this paper, we reviewed the performance
metrics to measure the performance of the cache and
parameters which are meant for evaluate the performance of a
Web cache. The page replacement algorithm categories such
as Recency based, frequency based and functional based and
page replacement policy such as LRU, LFU and the
application related to the web cache has been discussed and
we have concluded that LFU is perform better than the least
recently used (LRU).Our system removed the drawback of
LRU and LFU.

New Item

Find cache (m)

with min

Check priority of
cache (m)

No

If p[n] <FD

Find cache[n] with

second smallest T.S[n]

Yes
Yes
Replace cache (m)
with new item

If T.S[n]T.S[m]>TSD

No

Put t[m] =current & p

( ) 1

No

Check p[n] <FD

Yes

END
Replace cache[n]
with new item

Put TS[n] =current &

P[n] =1

REFERENCES
S.V.Nagaraj, Web Caching and Its Applications, Kluwer Academic
Publishers, 2004.
[2] Vinit A. Kakde, Prof. Sanjay K.Mishra , Prof. Amit Sinhal, Mahendra
S. MahalleSurvey of Effective Web Cache
Algorithm,
International Journal of Science
and Engineering Investigations,
Volume 1, Issue 1, February 2012.
[3] George Pallis, Athena Vakali, Eythimis Sidiropoulo, FRES-CAR: An
Adaptive Cache Replacement Policy, IEEE, 2005.
[4] K.Geetha, Dr.N.Ammasai Gounden, Monikandan S.,SEMALRU: An
Implementation of modified web cache replacement algorithm, IEEE,
2009.
[5] Abdullah balamash and Marwan krunz,
An Overview of Web
Caching Replacement Algorithms, IEEE Communications Surveys &
Tutorials, 2004, Vol.6 (2).
[6]. Harshal N. Datir, Yogesh H. Gulhane, P.R. Deshmukh,Analysis and
Performance Evalution of web caching algorithms, International
Journal of Engineering Science and Technology, feb.2011.
[7]. Naizheng Bian and Hao chen, (2008), A Least Page Replacement
Algorithm for web Cache Optimization, Workshop on Knowledge
Discovery and Data Mining.
[8]. Brad Whitehead, Chung-Horng Lung, Amogelang Tapela, Gopinath
Sivarajah,(2008), Experiments of Large File Caching and
Comparisons of Cache Algorithms, IEEE international Symposium on
Network Computing and Applications.
[1]

Fig.6. Proposed System

3235