1 4

CHAPT ER

Switched Multimegabit Data Service

Background
Switched Multimegabit Data Service (SMDS) is a packet-switched datagram service designed for
very high-speed wide-area data communications. SMDS offers data throughputs that will initially be
in the 1- to 34-Mbps range and is being deployed in public networks by the carriers in response to
two trends. The first trend is the proliferation of distributed processing and other applications that
require high-performance networking. The second trend is the decreasing cost and high-bandwidth
potential of fiber media, making support of such applications over a wide-area network (WAN)
viable.
SMDS is described in a series of specifications produced by Bell Communications Research
(Bellcore) and adopted by the telecommunications equipment providers and carriers. One of these
specifications describes the SMDS Interface Protocol (SIP), which is the protocol between a user
device (referred to as customer premises equipment, or CPE), and SMDS network equipment.
The SIP is based on an IEEE standard protocol for metropolitan-area networks (MANs): that is, the
IEEE 802.6 Distributed Queue Dual Bus (DQDB) standard. Using this protocol, CPE such as routers
can be attached to an SMDS network and use SMDS service for high-speed internetworking.

Technology Basics
Figure 14-1 shows an internetworking scenario using SMDS. In this figure, access to SMDS is
provided over either a 1.544-Mbps (DS-1, or Digital Signal 1) or 44.736-Mbps (DS-3, or Digital
Signal 3) transmission facility. Although SMDS is usually described as a fiber-based service, DS-1
access can be provided over either fiber or copper-based media with sufficiently good error
characteristics. The demarcation point between the carriers SMDS network and the customers
equipment is referred to as the subscriber network interface (SNI).

Switched Multimegabit Data Service 14-1

Addressing

Figure 14-1

SMDS Internetworking Scenario

Ethernet

SMDS
network
Central office

Central office

SNI
FDDI

Switch

Router

Trunk

Switch

DS-3
access

Router

Ethernet

DS-3
access
SNI

Router

FDDI

SMDS data units are capable of containing up to 9,188 octets (bytes) of user information. SMDS is
therefore capable of encapsulating entire IEEE 802.3, IEEE 802.4, IEEE 802.5, and FDDI frames.
The large packet size is consistent with the high-performance objectives of the service.

Addressing
Like other datagram protocols, SMDS data units carry both a source and a destination address. The
recipient of a data unit can use the source address to return data to the sender and for functions such
as address resolution (discovering the mapping between higher-layer addresses and SMDS
addresses). SMDS addresses are 10-digit addresses that resemble conventional telephone numbers.
In addition, SMDS supports group addresses that allow a single data unit to be sent and then
delivered by the network to multiple recipients. Group addressing is analogous to multicasting on
local-area networks (LANs) and is a valuable feature in internetworking applications where it is
widely used for routing, address resolution, and dynamic discovery of network resources (such as
file servers).
SMDS offers several other addressing features. Source addresses are validated by the network to
ensure that the address in question is legitimately assigned to the SNI from which it originated. Thus,
users are protected against address spoofingthat is, a sender pretending to be another user. Source
and destination address screening is also possible. Source address screening acts on addresses as
data units are leaving the network, while destination address screening acts on addresses as data
units are entering the network. If the address is disallowed, the data unit is not delivered. With
address screening, a subscriber can establish a private virtual network that excludes unwanted
traffic. This provides the subscriber with an initial security screen and promotes efficiency because
devices attached to SMDS do not have to waste resources handling unwanted traffic.
14-2 Internetworking Technology Overview

S1323a

DS-1
access
SNI

Access Classes

Access Classes
To accommodate a range of traffic requirements and equipment capabilities, SMDS supports a
variety of access classes. Different access classes determine the various maximum sustained
information transfer rates as well as the degree of burstiness allowed when sending packets into the
SMDS network.
On DS-3-rate interfaces, access classes are implemented through credit management algorithms,
which track credit balances for each customer interface. Credit is allocated on a periodic basis, up
to some maximum. Then, the credit balance is decremented as packets are sent to the network.
The operation of the credit management scheme essentially constrains the customers equipment to
some sustained or average rate of data transfer. This average rate of transfer is less than the full
information carrying bandwidth of the DS-3 access facility. Five access classes, corresponding to
sustained information rates of 4, 10, 16, 25, and 34 Mbps, are supported for DS-3 access interface.
The credit management scheme is not applied to DS-1-rate access interfaces.

SMDS Interface Protocol (SIP)

Access to the SMDS network is accomplished via SIP. The SIP is based on the DQDB protocol
specified by the IEEE 802.6 MAN standard. The DQDB protocol defines a Media Access Control
(MAC) protocol that allows many systems to interconnect via two unidirectional logical buses.
As designed by IEEE 802.6, the DQDB standard can be used to construct private, fiber-based MANs
supporting a variety of applications including data, voice, and video. This protocol was chosen as
the basis for SIP because it was an open standard, could support all the SMDS service features, was
designed for compatibility with carrier transmission standards, and is aligned with emerging
standards for Broadband ISDN (BISDN). As BISDN technology matures and is deployed, the
carriers intend to support not only SMDS but broadband video and voice services as well.
To interface to SMDS networks, only the connectionless data portion of the IEEE 802.6 protocol is
needed. Therefore, SIP does not define voice or video application support.
When used to gain access to an SMDS network, operation of the DQDB protocol across the SNI
results in an access DQDB. The term access DQDB distinguishes operation of DQDB across the SNI
from operation of DQDB in any other environment (such as inside the SMDS network). A switch in
the SMDS network operates as one station on an access DQDB, while customer equipment operates
as one or more stations on the access DQDB.
Because the DQDB protocol was designed to support a variety of data and nondata applications and
because it is a shared medium access control protocol, it is relatively complex. It has two parts:

The protocol syntax

The distributed queuing algorithm that constitutes the shared medium access control

CPE Configurations
There are two ways to configure CPE on the SMDS access DQDB, as shown in Figure 14-2. In a
single-CPE configuration, the access DQDB simply connects the switch in the carrier network and
one subscriber-owned station (CPE). In a multi-CPE configuration, the access DQDB consists of the
switch in the network and multiple interconnected CPE at the subscriber site. In this latter
configuration, all CPEs must belong to the same subscriber.

Switched Multimegabit Data Service 14-3

SMDS Interface Protocol (SIP)

Figure 14-2

Single-CPE and Multi-CPE Configurations

Single CPE
Bus A

Router

Switch

SMDS
network

Switch

SMDS
network

Bus B
Multiple CPE

Router

Bus A

Router

Bus B

Bus B

S1324a

Bus A

In the single-CPE case, the access DQDB is essentially just a two-node DQDB subnetwork. Each of
the nodes (the switch and the CPE) transfer data to the other via a unidirectional logical bus. There
is no contention for this bus because there are no other stations. Because of this, the distributed
queueing algorithm need not be used. Without the complexity of the distributed queuing algorithm,
SIP for single-CPE configurations is much simpler than SIP for multi-CPE configurations.

14-4 Internetworking Technology Overview

SMDS Interface Protocol (SIP)

SIP Levels
The SIP can be logically partitioned into three levels, as shown in Figure 14-3.

Encapsulation of User Information by SIP Levels

SDU

SMDS

SIP Level 3 PDU

SIP Level 2 PDU

TRLR

HDR

HDR

HDR

TRLR

TRLR

HDR

TRLR

S1325a

Figure 14-3

HDR = Header
PDU = Protocol data unit
SDU = Service data unit
TRLR = Trailer

Level 3
SIP Level 3 operation involves the encapsulation of SMDS service data units (SDUs) in a Level 3
header and trailer. Level 3 protocol data units (PDUs) are then broken into Level 2 PDUs as
appropriate to conform to Level 2 specifications.
The SIP Level 3 PDU is reasonably complex. It is shown in Figure 14-4.

Field length,
1
in bytes

SIP Level 3 PDU

Number of octets
1

RSVD BEtag BAsize DA

SA

X+
HLPI

4 bits 4 bits 2

X+ HEL

X+

12 9188 0,4

Info+
HE Pad CRC RSVD BEtag Length

S1326a

Figure 14-4

BEtag = Beginning-end tag

BAsize = Buffer allocation size
CRC = Cyclic redundancy check
DA = Destination address
HE = Header extension
HEL = Header extension length
HLPI = Higher-layer protocol identifier
Info+Pad = Information + padding (to ensure that this field ends on a 32-bit boundary)
SA = Source address
RSVD = Reserved
X+ = Carried across network unchanged

Switched Multimegabit Data Service 14-5

SMDS Interface Protocol (SIP)

Fields marked X+ in the figure are not used in the provision of SMDS, but are present in the protocol
to ensure alignment of the SIP format with the DQDB protocol format. Values placed in these fields
by the CPE must be delivered unchanged by the network.
The two reserved fields must be populated with zeros. The two BEtag fields contain an identical
value and are used to form an association between the first and last segments or Level 2 PDUs of a
SIP Level 3 PDU. These fields can be used to detect the condition where the last segment of one
Level 3 PDU and the first segment of the next Level 3 PDU are both lost, resulting in receipt of an
invalid Level 3 PDU.
The BAsize field contains the buffer allocation size.
The destination and source addresses consist of two parts: an address type and an address. In both
cases, the address type occupies the four most significant bits of the field. If the address is a
destination address, the address type may be either 1100 or 1110. The former indicates a 60-bit
individual address; the latter indicates a 60-bit group address. If the address is a source address, the
address type field can only indicate an individual address.
Bellcore Technical Advisories specify how addresses consistent in format with the North American
Numbering Plan (NANP) are to be encoded in the source and destination address fields. In this case,
the four most significant bits of each of the source and destination address subfields contain the value
0001, which is the internationally defined country code for North America. The next 40 bits
contain the binary coded decimal (BCD)-encoded values of the 10-digit SMDS, NANP-aligned
addresses. The final 16 (least-significant) bits are populated with ones for padding.
The higher-layer protocol identifier field indicates what type of protocol is encapsulated in the
information field. This value is important to systems using the SMDS network (such as Cisco
routers) but is not processed or changed by the SMDS network.
The header extension length (HEL) field indicates the number of 32-bit words in the header
extension field. Currently, the size of this field for SMDS is fixed at 12 bytes. Therefore, the HEL
value is always 0011.
The header extension (HE) field is currently identified as having two uses. One is to contain an
SMDS version number, which is used to determine what version of the protocol is being used. The
other use is to convey a carrier selection value providing the ability to select a particular
interexchange carrier to carry SMDS traffic from one local carrier network to another. In the future,
other information may be defined to be conveyed in the header extension field, if required.
The CRC field contains a cyclic redundancy check value.

Level 2
Level 3 PDUs are segmented into uniformly sized (53-octet) Level 2 PDUs, often referred to as slots
or cells. The format of the SIP Level 2 PDU is shown in Figure 14-5.

Field length,
in bits

SIP Level 2 PDU

32

14

352

Access
control

Network
control
information

Segment
type

Message
ID

Segmentation unit

Header

14-6 Internetworking Technology Overview

10

Payload Payload
length
CRC

Trailer

S1390a

Figure 14-5

SMDS Interface Protocol (SIP)

The access control field of the SIP Level 2 PDU contains different values depending on the direction
of information flow. If the slot is sent from the switch to the CPE, only the indication of whether the
PDU contains information or not is important. If the slot was sent from the CPE to the switch, and
if the configuration is multi-CPE, this field can also carry request bits that indicate bids for slots on
the bus going from the switch to the CPE. Further detail on how these request bits are used to
implement distributed queuing media access control can be obtained from the IEEE 802.6 standard.
The network control information field can contain only two possible values. One particular bit
pattern is included when the PDU contains information; another is used when it does not.
The segment type field indicates whether this Level 2 PDU is the beginning slot, the last slot, or a
slot from the middle of a Level 3 PDU. Segment type values are shown in Table 14-1.

Table 14-1

Segment Type Values

Value

Meaning

00

Continuation of message

01

End of message

10

Beginning of message

11

Single segment message

The message ID field allows association of Level 2 PDUs with a Level 3 PDU. The message ID is
the same for all segments of a given Level 3 PDU. On a multi-CPE access DQDB, Level 3 PDUs
originating from different CPE must have different message IDs. This allows the SMDS network
receiving interleaved slots from different Level 3 PDUs to associate each Level 2 PDU with the
correct Level 3 PDU. Successive Level 3 PDUs from the same CPE may have identical message IDs.
This presents no ambiguity, because any single CPE must send all Level 2 PDUs from one Level 3
PDU before it begins sending Level 2 PDUs of a different Level 3 PDU.
The segmentation unit field is the data portion of the PDU. In the event of an empty Level 2 PDU,
this field is populated with zeros.
The payload length field indicates how many bytes of a Level 3 PDU are actually contained in the
segmentation unit field. If the Level 2 PDU is empty, this field is also populated with zeros.
Finally, the payload CRC field contains a 10-bit cyclic redundancy check (CRC) value used to detect
errors over the segment type, message ID, segmentation unit, payload length, and payload CRC
fields. This CRC does not cover the access-control or network-control information fields.

Level 1
SIP Level 1 provides the physical link protocol, which operates at DS-3 or DS-1 rates between the
CPE and the network. SIP Level 1 is divided into two parts: the transmission system sublayer and
the Physical Layer Convergence Protocol (PLCP). The transmission system sublayer defines the
characteristics and method of attachment to the transmission link, that is, the DS-3 or DS-1. The
PLCP specifies how the Level 2 PDUs, or slots, are to be arranged relative to the DS-3 or DS-1
frame, and defines certain management information.
Because the SIP is based on IEEE 802.6, it has the advantage of compatibility with future BISDN
interfaces that will support not only data but video and voice applications as well. However, this
compatibility does cost some protocol overhead, which must be taken into account when calculating
overall data throughput that can be achieved using SIP. Over a DS-3 access DQDB, the total
bandwidth available for Level 3 PDU user data is approximately 34 Mbps. Over a DS-1 access,
approximately 1.2 Mbps can carry user data.
Switched Multimegabit Data Service 14-7

SMDS Interface Protocol (SIP)

The use of the IEEE 802.6 MAN MAC protocol as the basis for the SMDS SIP means that local
communication between CPE on the same access DQDB is possible. Some of this local
communication will be visible to the switch serving the SNI and some will not. The switch therefore
must use the destination address of a data unit to differentiate between data units intended for
SMDS-based transfer and data units intended for local transmission among multiple CPE sharing an
access DQDB.

Network Implementation
Inside the carrier network, the high-speed packet-switching capability required by SMDS can be
provided by a number of different technologies. In the near term, switches based on MAN
technology such as the DQDB standard are being included in a number of networks. A series of
Technical Advisories produced by Bellcore specify standard requirements on network equipment for
such functions as the following:

Network operations
Usage measurement for billing
Interface between a local carrier network and a long-distance carrier network
Interface between two switches inside the same carriers network
Customer network management

As has been noted, the IEEE 802.6 protocol and SIP were intentionally designed to align with the
principal BISDN protocol referred to as Asynchronous Transfer Mode (ATM). ATM and IEEE 802.6
belong to a class of protocols often referred to as fast packet-switching or cell relay protocols. These
protocols organize information into small, fixed-size cells (Level 2 PDUs in SIP terminology).
Fixed-size cells can be processed and switched in hardware at very high speeds. This tightly
constrains delay characteristics, making cell relay protocols useful for video and voice applications.
As ATM-based switching equipment becomes available, this technology will also be introduced into
networks providing SMDS.

14-8 Internetworking Technology Overview