The Technology Guide Series

techguide.com

ATM Interworking
Access
at the Customer
Premises

This guide has been sponsored by

 Table of Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Interworking over ATM Services The Relationship of ATM, Frame Relay and . . . . . .
Leased Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
When designing an enterprise network, corporate network Frame Relay and ATM—Technical Review . . . . . . . . 9
managers must consider the twin demands of fulfilling Interworking Frame Relay and ATM. . . . . . . . . . . . . 16
their bandwidth needs with finding the most economical QoS Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
service available. ATM Interworking Starting at the . . . . . . . . . . . . . .
Customer Premises . . . . . . . . . . . . . . . . . . . . . . . . . 27
RAD offers a range of managed access product solutions, Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . 29
enabling users to transport their corporate data, voice and Appendix 1—Formats for Frame Relay Frame . . . . .
LAN traffic over public Frame Relay and ATM services. and ATM Cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Appendix 2—Frame Relay Standards . . . . . . . . . . . . 33
The ACE family of ATM Concentrators and the MAXcess Appendix 3—ATM Service Categories. . . . . . . . . . . . 34
FRAD product line provide complete Frame Relay-ATM Appendix 4—Mapping Frame Relay Packets . . . . . .
interworking solutions for enterprise networks. Now, it is onto ATM Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
possible to build an integrated corporate network that Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
mixes ATM for offices requiring broadband access and Glossary of Terms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Frame Relay for those needing lower speed solutions. About the Editor…
Jerry Ryan is the Editor of the ATG series of Technology Guides on
Communications and Networking and is the author of numerous tech-
nology papers on various aspects of networking. Mr. Ryan has developed
Frame Relay Frame Frame Relay and taught many courses in network analysis and design for carriers,
Access ATM Access
Voice Relay Voice
Device Network Device government agencies and private industry. He has provided consulting
LAN Network LAN
support in the area of WAN and LAN network design, negotiation with
Customer Premises Customer Premises
carriers for contract pricing and services, technology acquisition, cus-
tomized software development for network administration, billing and
auditing of telecommunication expenses, project management, and
RFP generation. He was the president and founder of Connections
For more information about RAD’s full range of Telecommunications, Inc., a Massachusetts based company specializing
in consulting, education and software tools which address network design
ATM-Frame Relay Interworking access solutions, contact: and billing issues. Mr. Ryan is a member of the Networld+Interop
market@radmail.rad.co.il Program Committee. He holds a B.S. degree in electrical engineering.

This book is the property of The Applied Technologies Group and

is made available upon these terms and conditions. The Applied
Technologies Group reserves all rights herein. Reproduction in whole
or in part of this book is only permitted with the written consent of
The Applied Technologies Group. This report shall be treated at all
times as a proprietary document for internal use only. This book may
not be duplicated in any way, except in the form of brief excerpts or
quotations for the purpose of review. In addition, the information
www.rad.com contained herein may not be duplicated in other books, databases or
any other medium. Making copies of this book, or any portion for any
U.S. Headquarters International Headquarters purpose other than your own, is a violation of United States Copyright
Tel: (201) 529-1100 Tel: 972-3-6458181 Laws. The information contained in this report is believed to be reliable
but cannot be guaranteed to be complete or correct.
Fax: (201) 529-5777 Fax: 972-3-6498250
Copyright © 1998 by The Applied Technologies Group, One Apple Hill,
email: market@radmail.rad.co.il Suite 216, Natick, MA 01760, Tel: (508) 651-1155, Fax: (508) 651-1171
E-mail: info@techguide.com Web Site: http://www.techguide.com
 Contemporary wide area networks are increasingly dependent Another ATM advantage is the ability to consoli-
on ATM to provide flexible, high-speed, quality assured service to date all types of WAN services. Many carriers would
support the variety of network applications, each with their like to merge their commercial services (as well as their
unique set of service demands. ATM is deployed in many internal traffic) into one superior protocol. At the same
ways—as a transparent backbone deeply entrenched in the carrier time, other lower-rate WAN technologies, particularly
infrastructure, as a carrier service, as a private corporate back- frame relay, have also been evolving quietly alongside
bone, and as a homogeneous end-to-end service. Frame relay, the widely-deployed leased line population.
which generally preceded the wide spread deployment of ATM Frame relay is a variable-sized, packet-based tech-
has, in a short time, replaced traditional leased line services for nology, contrasting nicely with the fixed cell philosophy
many business applications. As a result, many networks today of ATM. ATM has evolved with a fixed 48 octet cell as
consist of ATM, frame relay, and leased line services that will be the atomic unit of its protocol. The argument, simply
part of carrier as well as corporate networks for the foreseeable stated, was that if the engineers could make ATM go
future. Supporting diverse applications, including traditional data fast enough, the cells, small as they were, could be
as well as multimedia and voice, these technologies can and delivered so quickly that the appearance of continuity
should be complementary. In order for organizations to fully could be preserved. It was also imagined that frame
benefit from this combination of service technologies, they need relay would merely be a temporary stopgap technology
to be deployed in a manner that assures effective interworking between X.25 and ATM. Running faster than the basic
and honoring of Service Level Agreements (SLAs). This Guide X.25 service, frame relay offered speeds from 56Kbps
examines the issues associated with ATM interworking with up to E1/T1 speeds.
traditional services and offers some guidelines and principles to That was the thinking at the beginning of the 90’s.
assure successful interworking operation and QoS. But then, in 1995, the frame relay community surprised
the telecom world by announcing that frame relay could
now be increased to T3 rates, about the entrance level
of ATM. As a result, frame relay, instead of becoming
Introduction a stopgap technology, is fully able to support long-term
corporate needs. Now, ATM and frame relay are com-
The great hope of ATM emerging as a ubiquitous plementary technologies. Frame relay supplies the end-
answer to all networking prayers has been tarnished by user connectivity and ATM provides the high-speed
the relentless onslaught of LAN evolutions, especially WAN interconnection and QoS support which gives
in the Ethernet sphere. Because of the inability of the user an illusion of seamless end-to-end connectivity.
ATM developers to reduce the cost of desktop connec- In addition to being seen as a high-speed solution
tivity to within the traditional Ethernet cost range, provider, frame relay (and more recently, ATM) can
ATM to the desktop may not occur soon. However, it also be positioned as a solution to decrease leased lines
is developing as the technology of choice for national connectivity costs. Today, many corporate WANs are
and international carriers of very high-speed and bulk implemented using fixed leased lines with data rates
data. The major carriers are deploying OC-192 ranging from 56Kbps to T3 speeds. Companies must
SONET rings throughout North America and the rest pay for the fixed costs of these dedicated links whether
of the world that will make ATM WAN data transfer a they are fully utilized or not. Additionally, the costs of
commonplace reality.

2 • ATM Interworking Access at the Customer Premises Technology Guide • 3

leased lines increase as their distance increases. Frame aspect of the growing importance of the Internet is the
relay permits replacing these fixed links with links that fact that IP networking, which not only drives routers
charge only for the traffic carried. In addition, the costs but is the underlying protocol of the Internet itself, as
are (normally) independent of distance, thus resulting well as of intranets and extranets, is the emergence of
in significant savings in monthly costs if the lines are Class of Service (COS) treatment of the various traffic
not fully utilized. flows that will be overlaid onto IP.
However, leased lines cannot be written off just
yet. First of all, they can guarantee the delivery of
delay-sensitive traffic such as voice and video; frame
The Relationship of ATM with Frame
relay cannot. Secondly, with the deployment of the Relay and Leased Lines
new xDSL technologies as the “final mile” solution to Regardless of their cost disadvantage, leased lines,
the house, the leased line service provider will offer a because of their fully dedicated bandwidth, can guar-
wide variety of rates and speeds. Being able to inte- antee the delivery of delay-sensitive traffic such as voice
grate these into a frame relay network will provide a and video. Frame relay, because it does not have QoS
seamless evolutionary path that protects the users’ standards, per se, and because it is a statistically multi-
investment in legacy equipment while allowing them plexed service, cannot. ATM, on the other hand,
to take advantage of the newly emerging technologies because of its mature class of service distinctions and
of the 90’s. As shall be seen, there is an ATM-based QoS operation, can support leased line traffic with the
model being developed (Circuit Emulation Service— same guarantees for delay sensitive traffic.
CES) over low-speed ATM which will aid in this inte- In a study of access requirements, (WAN Access
gration by giving a quality of service guarantee that Speeds, Sept., 1996) the Yankee Group market research
frame relay cannot match. firm estimated that 82% of all access lines through the
Other forces have been at work in the market. The end of the century would be at T1/ E1 speeds or lower.
exponential growth of the Internet has encompassed Unfortunately, low bandwidth leased lines and frame
hundreds of millions of new users of WAN technologies. relay (i.e., T1/E1 and lower) have typically been too
Internet Service Providers, (ISPs), quickly found that low-speed to be considered for ATM application. But
their initial DS0 and X.25 links were becoming satu- now, in order to support both low-speed leased lines
rated. Thus, the introduction of frame relay was and frame relay, low-speed ATM-based Circuit Emula-
crucial for their planned expansion and crucial for tion Service (CES) can be used for transportation of
the Internet to continue its exponential growth. ATM low speed leased line traffic, while frame relay applica-
deployment, delayed because of its complexity, was tions can be supported by standard Interworking
also too slow in lowering the costs to the end users as functionality.
well as being limited in what service and rates were CES functionality can be provisioned at the carrier’s
provided. Thus, frame relay, with its statistical band- Central Office to provide connections for large numbers
width usage, came along at the right time and at the of low-speed ATM ports which are overlaid onto a sin-
right price to replace the leased lines as the interconnect gle high-speed ATM switch port. By using inexpensive
of choice of the ISPs. As the World Wide Web has customer premise access concentrators, the carriers can
grown, so has the market for frame relay. An important offer the benefits of dedicated private line services with
attractive pricing that takes advantage of ATM’s statis-

4 • ATM Interworking Access at the Customer Premises Technology Guide • 5

tical multiplexing of its links. There is, however, a guide addresses this issue; the successful interoperabili-
penalty to be paid for this compromise. A traditional ty of frame relay/leased lines with ATM.
T1 link provides 24 DS0 channels while the CES T1-
ATM, because of CES overhead, only provides the
equivalent of 20 DS0s. There is a similar overhead in
the European E1 (30 channel) model. The Relationship of ATM,
Frame Relay and Leased Lines
The Importance of Service
Level Agreements (SLAs) The relatively slower speeds and lower costs of
Many factors are causing an increased awareness frame relay are ideally suited as a replacement for
of the need for carriers and vendors of network equip- legacy leased lines and to support local access to the
ment to support their proposed solutions with formal ISPs while the long haul carrying capability of ATM
assurances of satisfactory levels of service. For one, fits nicely into the infrastructure of network providers.
most applications migrating to the network today are Thus, the economies of scale of ATM, which are
mission critical, that is, they cannot tolerate casual or maximized with high bandwidth payloads, can be
less that satisfactory service. Furthermore, the complex- interfaced with the less expensive frame relay deployed
ity and newness of many network technologies leaves at the local level. Implementation and deployment
customers with valid concerns about how well service problems occur as vendors evolve their products to
levels can be predicted or assured. Compounding all successfully interwork these two related technologies.
of this is the nature of interoperating and interworking
SMDS X.25 ATM Frame Relay
networks which are made up of many vendor compo-
nents, each of which is expected to interwork without 800

fail. Because of these issues, the growing trend to nego- 600

Gigadollars

tiate formal service level agreements is a reality. Service

Level Agreements (SLAs) are one of the key elements 400

in assuring the customer the sincerity and commitment

200
of the vendor.
In summary, we can consider the network access 0
1996 1997 1998 1999 2000 2001
environment as having two major perspectives: The
prevalence of leased line and frame relay as a continu- Figure 1: Cell-based Revenue/Segment
ing reality for most businesses as an access method and
the parallel growth of ATM as the dominant backbone
technology of the wide area network. Given the twin The above chart (IDC, 1997) shows the rough
reality of frame relay/leased lines in the network access dollar value of ATM and frame relay purchases over
and ATM in the network backbone, the key question the next few years. For the next decade, frame relay
is how can these two perspectives come together in a will continue to be a significant player in the WAN
successful internetworking capability. This technology arena as will ATM. At the moment, one likely scenario

6 • ATM Interworking Access at the Customer Premises Technology Guide • 7

is that ATM will dominate the global infrastructure, for its maturity as a technology. With the above-mentioned
WAN interworking and frame relay will grow to domi- introduction of T1/E1-ATM, ATM can efficiently
nate the local access into the ATM cloud. This makes provide a universal transport layer for all of these
sense in terms of pricing, existing infrastructure and needs in addition to the value-added benefits of
traffic demands. Quality of Service (QoS) and Traffic Management.
A parallel reality is that regardless of the growing To take advantage of ATM’s built-in QoS and traffic
popularity of frame relay, users will still be deploying management, there is a need to provide an ATM
leased lines for a variety of reasons. As already featured approach (as opposed to an ATM-based
described, ATM can efficiently provide a universal system) to consolidating frame relay and leased lines
transport layer for all of these needs in addition to onto the common ATM substrate. Some carriers are
providing the value-added services of Quality of already consolidating most of the service offerings over
Service and traffic management. To take advantage of ATM. More than likely, users do not realize that their
ATM’s built-in QoS and traffic management capabilities, leased line services are actually ATM CBR services!
there is a need to provide an integrated, interworking
ATM-based approach to consolidating frame relay and
leased line onto the ATM substrate.
Frame Relay and ATM—
The Need for ATM Interworking Technical Review
Interworking can be defined as the fundamental
capability of higher layer network elements to operate
in an integrated and comprehensive manner, fully uti-
A Brief Technical Review of Frame Relay
lizing the basic capabilities of the related services to Frame relay has become particularly popular as
move information through the network across multiple a replacement for legacy leased line applications and
service elements. It is not sufficient to simply share for supporting data and voice connectivity. Frame relay
some level of operational interface so that data can services are provided by all of the major telecommuni-
move through the network. The term “interoperability” cations carriers in the world, and it is likely to become
often provides a less than complete functionality in as ubiquitous as X.25 was in its prime. Frame relay
which certain features are sacrificed to allow cross net- data rates range from 56Kbps to 45Mbps, positioning
work communication. Interworking, on the other hand, it as a natural bridge between analog modem connec-
implies a deep, seamless compatibility in which features tivity and entry-level ATM capabilities. Furthermore,
are not lost, but integrated into the whole. This is the it is assumed that frame relay will run over a highly
case in which ATM interworks with both leased line reliable physical transport medium and therefore will
and frame relay services. The rich functionality of not need the rigorous error detection schemes employed
ATM is specifically designed to fully support them by X.25. Frame relay packets are in the form of variably
without loss of capability. Due to this, some carriers sized frames, which enables it to accommodate data
are already consolidating all of the service offerings packets of different sizes and makes it easy to accom-
over ATM. Finally, users will still be attracted to leased modate any native protocol with little protocol transla-
lines because of pricing, and to frame relay because of tion. This is one of its key advantages. A second

8 • ATM Interworking Access at the Customer Premises Technology Guide • 9

advantage of the variable frame format is that it is well carrier location which forwards all of the traffic across
suited to support “bursty” transmissions, such as one the network to its destination according to a predeter-
would expect in LAN or Internet traffic. mined path. The discard eligibility value is key to keep-
Traditionally, frame relay has not been deemed ing traffic moving smoothly through the network. The
suitable for time-sensitive traffic since packets may be congestion control fields provide mechanisms for throt-
discarded if overloading occurs. In the last couple of tling the traffic back to avoid lost frames. If congestion
years, there has been considerable interest in voice over is still not relieved, a simple priority scheme discards
frame relay. The Frame Relay Forum has accepted two lower priority packets. Note in passing that this implies
Implementation Agreements (IAs), FRF-11 (Voice over that the higher layer application must do the error
Frame Relay) and FRF-12 (Frame Relay Fragmentation). correction; frame relay does not recover from discarded
Basically, the fundamental problem is that frame relay cells.
data PDUs can be very large (up to 8,000 bytes) where-
as voice PDUs (protocol data units) tend to be small The Virtual Circuit
(30 bytes). Thus, data can clog up the network delaying The concept of the virtual circuit allows the net-
voice packets or forcing packet discard. The result is work to operate in a manner similar to a leased line. A
poor voice quality, which is unacceptable to most cor- virtual circuit, whether Permanent (PVC) or Switched
porations today. By implementing the FRF-12 IA, (SVC), provides a route through the network over
data packets can be fragmented into smaller sizes and which the frame travels to reach its destination. This
stalled when voice packets are being transmitted. If route is determined by the carrier according to certain
the vendor combines this with a good data compression standards of performance and qualities of service. A
algorithm that can maintain reliable performance in PVC is negotiated and arranged prior to its use and is
the presence of errors, voice quality can approach that “permanently” assigned. An SVC provides a path that
of leased lines. The cost is, of course, much lower. is set up upon request and after traffic is sent, the VC is
Now that the two IAs are in place, vendors are rolling torn down. The value of a PVC is that no time is lost
out compliant equipment. Thus, frame relay is about in call setup and it is easier to implement. The negative
to become a cost-effective way of delivering voice (up side is that PVCs that are rarely used take up resources
to E1/T1 rates). As shall be seen, ATM services can be that could be assigned elsewhere. SVCs, on the other
used to seamlessly transmit higher rates. hand, require call setup each time they are used,
impacting performance.
The Frame Relay Frame Structure When implementing frame relay, the user can
The frame structure of frame relay is based on acquire as many PVCs as are needed, usually one for
the simple arrangement of the OSI HDLC frame, each logical destination in the network. For example,
which includes a two octet header containing address if PVCs are used to connect remote SNA sites, each
information (DLCIs and EA), congestion information SNA site would be assigned a unique PVC. Similarly,
(FECN and BECN), discard eligibility (DE), and a for VoFR, rather than use a unique DLCI for each
command response value C/R (see Appendix 1). voice circuit (which is possible in a private network but
Although quite simple, this structure is highly effective. prohibitively expensive in a public one as the carriers
The DLCI address maps onto a virtual circuit at the charge by the number of DLCIs), a good vendor will

10 • ATM Interworking Access at the Customer Premises Technology Guide • 11

use a sub-addressing scheme, multiplexing many circuits In an analogous manner to frame relay, ATM
onto one DLCI. also defines a congestion bit so that lower priority cells
Frame relay networks may be private, public, or a can be selectively discarded if the network shows signs
combination of both. Finally, as has been noted above, of overloading. In addition, ATM standards define
the standards for frame relay are driven by the Frame a concept called Quality of Service, which will be
Relay Forum, a consortium of major vendors in the discussed in detail.
field, similar to the ATM Forum. The standards in The ATM model is similar to that of frame relay.
place are listed in Appendix 2. Each device attaches to the ATM service via a User
Network Interface (UNI). ATM interior nodes commu-
nicate via a Network to Network Interface (NNI).
A Technical Review of ATM Networks can be private, public, or a combination of
Asynchronous Transfer Mode (ATM) is a cell- both. Communications are circuit-oriented, as far as
based protocol with a fixed cell size of 53 octets, 48 for the users are concerned. PVCs and SVCs can be set
data and 5 for header information. The basic transfer up as in frame relay. (See Appendix 1 for a detailed
rate, called OC-1, is approximately 51Mbps, enough to look at the ATM cell structure)
accommodate a T3 signal plus some overhead. Higher The ATM protocol is quite rich and entails many
rates are defined in multiples of the basic Optical functions. It defines the rules for the actual data transfer.
Channel (OC) rate. An OC-48 channel, for example, In addition, it also defines the rules for signaling, in
is 48 x 51Mbps or about 2.45Gbps . The basic premise which the VPI/VCI’s temporary and unique address-
on which ATM is based is that if the speed of the ing is set up along with the switching path through
synchronous substrate (SONET, in Europe—SDH) on the cloud to the receiver. Finally, it defines one of the
which the ATM cells are delivered is high enough, traffic most important aspects of the ATM protocol—the
of vastly different characteristics can be transported QoS guarantees.
across a common medium without mutual interference.
Real-time signals such as voice can be segmented into
48 octet cells, sent across the ATM network, and Functional Differences Between
reassembled at the other end quickly enough so that the Frame Relay and ATM
receiver cannot detect any delays. Because the cell size Frame relay delivers frames in the order in which
is fixed, the network can deliver data much faster and they were sent with a high delivery probability. The
more reliably than if variable-sized cells were allowed. underlying assumption is that the physical layer is very
The basic OC-1 rate was selected by the carrier reliable. If frames are corrupted, they are dropped and
industry to manage T3 rates, as mentioned. Recently, it is the responsibility of the higher layers to recover
the introduction of “low-speed” ATM has permitted from the loss by requesting a retransmission of the lost
edge-users to have their DS0 channels combined into portions of the original transmission, if that is an
E1/T1 trunks which are then multiplexed into the stan- appropriate action to take. (In some instances, it is not
dard ATM fabric speed, switched to their destinations, appropriate, such as in voice or video, since lost traffic
and demultiplexed into the original DS0 channels, is simply experienced as static or interference.)
preserving their real-time characteristics.

12 • ATM Interworking Access at the Customer Premises Technology Guide • 13

 Frame relay may also not be suitable for delay- Unstructured CES is intended for equipment that
sensitive traffic if the ordinary frame size is implemented does not support standard framing formats (i.e., clear
up to the 8K octet maximum, since this would cause channel T1/E1 circuits). Thus, a CES-based solution
significant variation in the frame delay, particularly in connects customer premise equipment to a private or
the event of retransmissions. ATM, on the other hand, public ATM network, integrating circuit emulation
because of the small packet size, smooths out the traffic (voice and video) with data.
flow so that the transmission of large pieces of data Although leased lines have been a technology of
does not block the transmission of small real-time choice for decades, as stated earlier, it makes sense to
packets. In the event of errors, retransmissions are support them through ATM interworking. By definition,
accomplished quickly. However, this distinction is leased lines provide constant bit rate, non-interrupted
erased when the frame size is reduced. In some cases, traffic flow and, for this reason, leased lines naturally
when transmitting very small messages (< 48 octets), support voice and video, in addition to certain,demand-
frame relay can actually be more efficient than ATM. ing data applications. When interfaced to ATM, these
While ATM must use a 48 octet, frame relay could use rather straightforward, although non-negotiable charac-
a packet size of just one octet if needed. Indeed, frame teristics must be maintained. That is, the ATM function
relay can handle even SNA traffic quite efficiently, a must allow transparent, non-blocking, non-delayed traffic
communications protocol notorious for its critical flow from end to end for each ATM equivalent virtual
timing requirements. circuit. Leased lines of Nx64Kbps or Nx56Kbps services
Frame relay virtual circuits may be point-to-point, can be aggregated into the backbone through the
point-to-multipoint, or multipoint-to-multipoint (also ATM CES function. In this case, all of the appropriate
called multicast). They may also be arranged into modem functions must be supported and when the
closed user groups for security purposes. Frame relay signal is offloaded at the distant end, it must be recon-
handles congestion a little less precisely than ATM. It stituted in a way that is compatible with the receiving
defines a Committed Information Rate (CIR) which CPE. In a similar way, digital leased lines that operated
is the rate that the network agrees it will support for a up to T1/E1 and higher use CSU/DSU technology
particular frame connection. If data is transmitted in that must be accommodated by the interworking ATM
excess of the CIR, the excess data is vulnerable for backbone. Although successful interworking with frame
discarding under congestion. However, there is no relay has more complexity, interworking with leased
guarantee that the CIR will always be met. ATM, lines is as important to the industry. Figure 2 illustrates
on the other hand, is much stricter in observing its customer premises connected directly into the ATM
guarantees of service. network via on premises ATM access device.

Leased Line—ATM Interworking Voice ATM

Access
E1 / T1
ATM E1 / T1
ATM Voice
LAN Device Access
The ATM Forum has specified two modes for Device LAN

CES; structured and unstructured. Structured mode

CES is intended to emulate point-to-point fractional Figure 2: ATM Access Devices at Customer Premises
T1/E1 circuits, permitting more efficient use of CBR
bandwidth by supporting multiple 64Kbps circuits.

14 • ATM Interworking Access at the Customer Premises Technology Guide • 15

The Benefits of Combining Frame Frame Relay
Relay, Leased Lines and ATM Voice Frame
E1 / T1
Relay IWF ATM Frame Voice
By connecting a frame relay network to an ATM LAN Device IWF
E1 / T1
Relay
Device LAN
backbone, the user can benefit from low end-to-end
delay and low differential delays, which permit the
efficient transport of voice and time-sensitive data such Figure 3: Frame Relay-ATM Network
as SNA. It also enables better control over the cost of with IWF at the ATM Edge

equipment. As bandwidths increase, users can keep

their FRADs at lower volume sites and upgrade the The frame relay devices know nothing about the
central ATM sites, thus preserving their capital invest- ATM backbone connection, because the networking
ment. equipment, normally the ATM WAN switches, provide
Most importantly, by carrying frame relay and the necessary interworking function. Multiple frame
leased line traffic on ATM, those traffic streams can relay networks can be supported by an ATM back-
benefit from the fully provisioned QoS capabilities of bone, making it a scalable high-speed option for frame
ATM. This gives the user the best of both worlds: low relay users and network designers that does not require
cost, right-sized access trunks through frame relay and changes to customer premises equipment. It provides
leased lines, and high-speed, QoS managed wide area a strategic architecture that supports both the flexible
service through ATM. and cost effective access that frame relay provides as
well as the high-speed and throughput capabilities of
ATM in the network infrastructure.
Interworking Frame Relay A second approach (Figure 4) is to place the inter-
networking function inside the device that connects
and ATM directly to the ATM cloud. The advantage of such an
approach is that it enables the provisioning of both
interworking and ATM based QoS assurance right at
The Basic Concept the customer premises. ATM switches also benefit if
Frame relay-ATM network interworking connects this functionality is performed at the customer premises,
separate frame relay networks using an ATM cloud as since it enables the switch to only handle ATM cells.
the backbone (Figure 3). As far as the user is concerned, Furthermore, this approach enables easy upgrading if
the connection is made as if the separated networks the client needs more WAN resources. The IWF device
were one. need only be modified to connect directly to the ATM
cloud if rates greater the E1/T1 are needed.

16 • ATM Interworking Access at the Customer Premises Technology Guide • 17

 performs the ATM Adaptive Layer (AAL) function of
Frame Relay
assembling and disassembling the entire frame relay
VoFR ATM
&
E1 / T1
ATM ATM VoFR
frame into and out of ATM cells, including the com-
LAN IWF E1 / T1
& plete frame overhead. It is similar to encapsulation, or
IWF LAN
tunneling.
Figure 4: Interworking at the Customer Premises
Frame Flag Header User SDU (user traffic) FCS Flag

Frame address
bits maps to
VCI/VPI bits
Frame Relay-ATM Interworking Standard Cell Header Payload Header Payload Header Payload
Beyond providing a means for frame-based user
equipment to access ATM networks at the user to net- Figure 5: Frame Mapping onto ATM Cells
work interface, an important consideration is how to
interwork frame relay networks with ATM networks In translation mode, the frame relay frame is
and thereby interwork the network users. This leads to unbundled and the frame overhead is mapped onto an
two frame relay/ATM interworking scenarios: Network ATM cell, replacing the frame protocol and thereby
Interworking and Service Interworking. These two reducing overhead.
interworking functions provide a method by which
the two technologies can interoperate. Simply stated,
network interworking provides a transport between The Network Interworking
two frame relay devices or entities. Service interwork- Function (NIWF)
ing enables an ATM user to transparently interwork
The Network Interworking Function facilitates the
with a frame relay user, and neither one knows that the
transparent transport of frame relay user data and PVC
far end uses a different technology.
signaling from one frame relay network connected to
By providing a framework within which both tech-
another by an ATM backbone. The frame relay
nologies can interwork, it permits the network manager
frame, including all protocol overhead, is completely
to avoid making the difficult choice of selecting either
encapsulated and placed into a series of ATM cells.
one or the other network to install. In addition, by
The Network IWF function is typically integrated into
having both work together seamlessly, individual net-
either the ATM or the frame relay switch. Each frame
work components can be implemented with either
relay PVC is mapped directly onto an ATM PVC.
technology as circumstances dictate.
There are three major ways to implement the
NIWF. The IWF function can be performed inside the
Transparent vs. Translation Mode frame relay network, attaching as an ATM UNI into
the backbone. Secondly, the ATM connection point
Interworking can either occur in transparent or
can perform the IWF within the switch performing the
translation mode. In transparent mode, the customer
UNI (this is the most common configuration). In this
premises equipment (CPE), typically an ATM access
case, the translation would be performed in the Service
concentrator, accepts the frame relay traffic as is and
Specific Convergence Sublayer of the ATM protocol

18 • ATM Interworking Access at the Customer Premises Technology Guide • 19

stack. Finally, an intermediate device can perform the
ATM
translation; frame relay in, ATM out. Edge Device
Frame Relay

FR
The Service Interworking Device
IWF ATM
IWF
FR
Device
Function (SIWF) Customer
Premises Customer
Premises
The Service Interworking Function does not trans-
port traffic transparently. Unlike the IWF, which loads Figure 6: Frame Relay and ATM Devices
the entire frame relay frame including the frame relay Attached to an ATM Network
protocol elements directly into the AAL layer of ATM,
the SIWF acts as a protocol converter. In this arrange- In the NIWF case, the sending node must be
ment, the frame relay traffic travels through the frame configured to interoperate with the distant node since
relay network to the SIWF. There it is converted into it will deliver an intact frame relay frame to the distant
the appropriate ATM structure and mapped directly end of the ATM network.
onto an ATM PVC. In this approach, the frame relay In SIWF, frame relay PVC status signaling is con-
protocol elements are stripped off and only the data is verted to ATM OAM cells and OAM cells are converted
converted into ATM Cell payload. The details of the to frame relay status signaling. Under this arrangement,
mapping function are illustrated in Appendix 4. when a failure occurs in one network, the other network
Note that in the case of SIWF, the frame relay is notified in its appropriate protocol. Other conditions
PVC is mapped onto an equivalent ATM PVC with such as congestion and discard/eligible/cell loss priority
no knowledge of whether or not the target device is are mapped appropriately. Specifically, the SIWF
attached to a frame relay network on the other side maps the frame relay DLCI onto an ATM VPC/VCI
of the ATM network. In fact, there is no need for the address, the FECN into the payload field and the DE
remote device to be a frame relay attached device (see bit onto the CLP bit. The details of the mappings and
Figure 6). With service interworking, users can coexist the related RFCs can be found in FRF-8. Since the
with or evolve a portion of the existing frame relay frame relay multi-protocol encapsulation procedures
network to ATM, without requiring any other network (RFC 1490) are not identical to those of ATM (RFC
devices to follow suit. Essentially it allows an “as-need- 1483, RFC 1577), SIWF also converts the multi-protocol
ed” deployment of frame relay at any place in the data unit headers from the frame relay format into
network that is deemed appropriate. ATM’s and vice versa.

20 • ATM Interworking Access at the Customer Premises Technology Guide • 21

QoS Issues ATM Service Categories
The ATM Forum has defined 5 categories of
Any realistic network must be able to provide ATM service, namely:
Quality of Service guarantees of some sort. ATM is • Constant Bit Rate (CBR),
the most advanced protocol family in this regard and it
• Real-time Variable Bit Rate (rt-VBR),
is useful to examine it in detail. Because of the high
service competition (caused by deregulation) users are • Non-Real-Time Variable Bit Rate (nrt-VBR),
expecting and demanding better service quality offer- • Unspecified Bit Rate (UBR), and
ings. These are being codified by means of service level
• Available Bit Rate (ABR).
agreements, which precisely define the QoS that is to
be delivered. These are explained more fully in Appendix 3.
It is clear, therefore, that ATM’s QoS is an essential
tool for contemporary network vendors seeking to
comply with demanding SLAs. As a corollary to this,
the emergence of Class of Service (COS) treatment for ATM Quality of Service (QoS) Parameters
TCP/UDP/IP traffic anticipates similar SLA compliance QoS parameters, as defined by the ATM Forum,
capabilities in that environment. Since IP based traffic describe the quality of the Virtual Connection offered
is one of the dominant trends for future networking, its by the carrier. The set of agreed QoS parameters
COS capabilities need to interwork with the equivalent defines the commitment of the carrier to the customer.
QoS capabilities of ATM. This cross mapping function, Each IWF connection can negotiate three important
illustrated in Figure 7, will be an essential product QoS parameters with the ATM switch, namely:
capability. • Cell Loss Ratio (CLR),
• Maximum Cell transfer delay (maxCTD), and

QoS
• Cell Delay Variation (CDV).
IP Router Class of Service ATM

Customer
These are explained more fully in Appendix 3.
Premises Cross Mapping
Function
Both maxCTD and CDV refer to end-to-end
Figure 7: COS-QoS Cross Mapping Function objectives. The negotiation of these QoS parameters
is a function of provisioning (for PVCs) or signaling
(for SVCs). There are other non-negotiable ATM QoS
parameters which relate to errors introduced by the
network facilities or switching equipment.

22 • ATM Interworking Access at the Customer Premises Technology Guide • 23

The Traffic Contract private ATM or traditional service users, traffic shaping
allows flexible and accurate adaptation to the required
When the connection is first set up, a contract is
service. Time sensitive applications like voice can be
established between the customer and the carrier. It is
run over ATM in conformance to the relevant QoS
the formal agreement of the Quality of Service that
parameters. In addition, carriers benefit from traffic
will be provided. Like all contracts, it is binding on
shaping performed at the customer premises, as it
both parties. As long as the customer keeps his traffic
smooths ingress traffic. Such functionality pushed to
within the parameters agreed upon, the carrier is
devices located at the customer site help to solve the
committed to guarantee that Quality of Service.
tradeoff between the bandwith optimization and QoS
The contract is structured in two parts, the
guarantees.
Connection Traffic Descriptor and the Source Traffic
Descriptor. The Connection Traffic Descriptor includes
the negotiated QoS, which implies a conformance Traffic Shaping Defined
definition for the data entering the system. This confor-
Traffic shaping is that value-added processing of
mance is guaranteed by a Usage Parameter Control
a devices’ cell stream to ensure that it conforms to an
(UPC) function at the input edge of the network. It is
idealized definition of traffic flow or to a traffic contract.
often called traffic policing as it insures that the traffic
Traffic is “shaped” to eliminate peaks in a flow that
parameters that have been negotiated are policed.
may be bursty or have excessive cell jitter due to a
The Source Traffic Descriptor includes the Peak
variety of network factors. By smoothing the cell
Cell Rate (PCR), the Sustainable Cell Rate (SCR), the
stream of every connection, a switch’s traffic shaping
Maximum Burst Size (MBS), and for the ABR category,
function can ensure a more predictable traffic profile
the Minimum Cell Rate (MCR). The totality of these
which leads to better fairness, lower cell loss, and less
numbers attempts to define both the maximum and
stress on network resources.
the average bandwidth requirements.
When the connection is first requested, the carrier
switch must determine if it can still meet the QoS level Performance Benefits of Traffic Shaping
for this and all existing connections. This activity is There are some cases where traffic shaping may
called the Connection Admission Control (CAC) func- mean the difference between an application working
tion. Note that the implementation of the CAC func- well or not working at all. This would be true for an
tion is network-specific that is, it is not standardized. ATM end system that outputs cells at the full-line rate,
instead of limiting output to the peak cell rate as
Traffic Shaping defined by the traffic contract. This may not be unusual
since data communications devices usually send data
In addition to traffic monitoring and policing, an
at the full-line rate of the line card. ATM, however,
ATM access device should provide traffic shaping func-
requires more than just segmenting frames or packets
tionality. This allows customers to better utilize ATM
into cells; it requires traffic shaping to ensure that the
services, even at low rates. Both customers and ATM
interval between cells conforms to the peak cell rate,
service providers enjoy improved service since no cells
which may be substantially lower than the full-line rate.
are dropped, even in cases of bit rate adaptation. For

24 • ATM Interworking Access at the Customer Premises Technology Guide • 25

 • There is enormous advantage in being able to assurances of the service provider in specific detail.
shape traffic on a per-circuit basis. It includes contingency understandings and penalty
• Non-conforming traffic can be brought into com- clauses for non-compliance. The SLA is an appropriate
pliance on a per-circuit contract basis to avoid cell companion to the ATM QoS standards. In Interworking
discard at down-stream nodes due to traffic con- between ATM and other service types, it is sometimes
tract violation. the only vehicle for defining and assuring good service.

• Fairness between circuits can be achieved since

each circuit is prevented from bursting into the
output queue at excessive cell rates. ATM Interworking Starting
• The system can perform more accurate resource at the Customer Premises
allocation due to the now-predictable nature of the
shaped traffic. Thus, the switch can avoid over-
provisioning both buffers and bandwidth. Customers can use the above technologies to con-
nect legacy equipment to ATM-based networks. There
are a variety of routers and access devices that permit
ATM Interworking and SLAs connection into either the leased line, frame relay, or
The challenge for ATM interworking is to have a the ATM cloud. Options range from plug-in modules,
strategy for matching the various application types onto which can either bridge or route legacy Ethernets into a
the ATM network. In order to be successful, the various T1/E1 connection, and hence into a frame relay/ATM
products that attempt to interwork with ATM, whether connection, or stand-alone aggregators which can
LAN switches, frame relay access devices, IP routers, accept a variety of inputs.
or simply leased line multiplexers, must have a clear Typical devices include interfaces into common
view of how ATM will deal with the traffic it receives. legacy LANs such as Ethernet/Fast Ethernet or Token
Interworking implies that the subtleties of the traffic Ring. Typical WAN interfaces include T1/E1, frame
handoff, whether through an encapsulation method relay, X.25, and SMDS. Thus, these aggregators can
such as the Network Interworking Function or through accept various legacy inputs and route or bridge them
some protocol conversion arrangement such as the into a frame relay connection.
Service Interworking Function, must be understood
and fully complied with. This will be a major differen- Important Issues
tiator among vendor offerings. The customer should
also recognize that the guarantees implied in the QoS It is certainly true that carriers are investing more
standards in ATM should be codified with the vendor and more in ATM backbone equipment. Sprint, for
through the mechanism of the Service Level Agreement. example, has committed their network to be 100%
The Service Level Agreement, or SLA, is a binding ATM by the end of the decade. AT&T, Worldcom, and
contract between the carrier and the customer to make others are not far behind. Therefore, ATM, predictably,
sure that the desired QoS is delivered. It specifies all is the dominant carrier-based WAN architecture. As a
of the essential requirements of the customer and the result of this, better ATM-based products and services
are sure to emerge to support both traditional services,

26 • ATM Interworking Access at the Customer Premises Technology Guide • 27

such as leased lines for data and voice and robust ATM equipment into their expanding network. As new
interworking products for frame relay and IP based ser- developments occur, older parts can be selectively
vices such as the Internet. replaced without disturbing the flow of application
To take advantage of the dominant trend, the next data traffic. These give the user an evolutionary, scal-
logical step in network evolution will be the development able, transparent migration path that will both protect
of ATM featured services which start right at the cus- invested infrastructure yet permit increasing economies
tomer premises. To enable this to happen successfully, of scale as the field matures.
there must be a concomitant development of dedicated
low cost access devices that will interwork smoothly
with ATM. In order to support the expected high QoS, Summary and Conclusions
this new generation of IWF access devices will need to
be equipped with advanced traffic management tools as In conclusion, ATM interworking, augmented with
well as QoS measurement capabilities. Using the traffic frame relay and leased lines, provides an architecturally
grooming capabilities of these state-of-the-art IWF sound way of designing flexible, cost-effective networks
access devices which can easily scale ATM, support for that will permit incremental evolution while maintaining
leased line services and bandwidth management can be a unified end-to-end view for both applications and
tailored to the customer needs. These access devices end users. As the technologies evolve, we can expect
can easily support voice over ATM using CES. With some quickly developing implementations to become a
the development of the ATM Circuit Emulation reality. It is inevitable, since ATM will certainly pervade
Services (CES), both T1/E1 ports and fractional the WAN infrastructure, that ATM will also expand
T1/E1 links can be collected and transported over deeper and deeper into the customer premises, and
ATM networks, permitting even richer forms of virtually all network applications will migrate into
network and voice connections. some form of ATM interworking. This will provide the
Furthermore, LAN and IP can also be supported customer with a uniform environment, with standard-
and interworked at the customer premises before being ized QoS capabilities across the network, and depend-
transported over ATM. able SLAs to enable the growing demand for mission
Standards groups across the industry are working critical application support throughout the network.
hard to develop ways of mapping the wide variety of
applications and service types into the ATM Class of
Service categories so that QoS standards can be applied Alternative Scenarios
uniformly throughout the network. These developments As the future quickly unfolds, several evolutionary
will naturally enhance the importance of IWF-capable scenarios make sense. First, in connecting into the
customer premises access devices. frame relay or ATM networks, the network user can
Regardless of future developments, however, frame connect local devices at the customer premises into a
relay and leased line interworking is here today as a FRAD or other access device that connects directly
rich set of capabilities. In fact, frame relay, because of into a E1/T1 port and onto the carrier service without
its fundamental similarity to ATM, is the most suitable complex IWF functions. A corollary to this is the direct
to be interworked today. Thus, with these new devices, leased line connection from an ATM access device at a
network designers can effectively integrate legacy

28 • ATM Interworking Access at the Customer Premises Technology Guide • 29

customer premises in which customer traffic is converted
to ATM cells and shipped into the ATM cloud by the Frame Relay IWF
CP or
on-premises access unit. IWF ATM ATM
Leased Lines
A second approach is deployed when the network
user needs to combine heterogeneous networks because Customer
Premises
of price and availability, or because of the need to Demarcation
retain legacy equipment. In this case, multiplexers are
available that combine data and voice over frame relay.
They can interwork locally at the edge of the frame Figure 9: Direct Customer Access into the
ATM Network Using Frame Relay-ATM IWF
relay cloud or have the frame relay carrier perform the
Interworking at the ATM edge, as is illustrated in
Figure 8. Again, legacy equipment is retained, permitting a
controlled evolutionary growth into the future. The
ATM scalability of ATM can easily be leveraged to provide
Data Access
MX
FR CP Edge Device almost unlimited upwards expansion.
VoFR IWF Frame IWF ATM ATM Data
FR Relay ACE Indeed, the emerging advantages of ATM are
Voice
Device becoming clearer with its widespread deployment in
the backbone of the network infrastructure. ATM has
Figure 8: Data and Voice over Frame Relay
undoubtedly become a dominant force. Equally true is
the fact that frame relay is here to stay and meets the
many disparate needs of the user community. It is
In a third phase, as the network user migrates to
inevitable, therefore, that frame relay will continue to
more mission critical and demanding traffic require-
grow dramatically. Even leased lines are maintaining a
ments, with stringent QoS and their associated SLAs,
steady, although not growing presence. Because of this,
the frame relay/ATM interworking function can be
ATM interworking, particularly with frame relay and
performed directly at the customer premises through
leased lines will continue to evolve and become a
IWF- compatible access devices. These devices are
primary paradigm for networks of the future.
used to multiplex local traffic onto frame relay and
from there directly into the ATM service provider
network (Figure 9).

30 • ATM Interworking Access at the Customer Premises Technology Guide • 31

Appendix 1—Formats for Frame Appendix 2—Frame Relay
Relay Frame and ATM Cell Standards
FRF.1.1 UNI Implementation Agreement
8169 Octets Maximum FRF.2.1 NNI Implementation Agreement
Flag Frame Relay Header Information FCS Flag
FRF.3.1 Multiprotocol Encapsulation
Implementation Agreement (MEI)
DLCI C/R EA DLCI FECN BECN DE EA control NLPID Ether Frame
FRF.4 SVC Implementation Agreement
2 Octets 1
FRF.5 Frame Relay/ATM Implementation
Figure 10: Frame Relay Frame with Ethernet Frame Agreement
FRF.6 Frame Relay MIB Implementation
• Flag—(01111110) Agreement
FRF.7 Frame Relay PVC Multicast Service and
• FCS—Frame Check Sequence
Protocol Description Implementation
• DLCI—Data Link Connection Identifier, is a 10 Agreement
bit address
FRF.8 Frame Relay/ATM PVC Multicast
• C/R—Command/Response indicator (2 bits) Service Interworking Implementation
• FECN—Forward Error Congestion Notification Agreement

• BECN—Backwards Error Congestion Notification FRF.9 Data Compression Implementation

Agreement
• EA—Extended address bits
• DE—Discard Eligibility bit FRF.10 SVCs at the NNI Implementation
Agreement
8 7 6 5 4 3 2 1 FRF.11 Voice over Frame Relay (VoFR)
GFC VPI 1 Implementation Agreement
VPI VCI 2
FRF.12 Frame Relay Fragmentation Agreement
VCI 3

VCI PT CLP 4

HEC 5

Payload 48 octets .

53

Figure 11: ATM Cell Format

32 • ATM Interworking Access at the Customer Premises Appendix • 33

Appendix 3—ATM Service Non-Real-Time Variable Bit Rate (nrt-VBR)
This category is similar to the nr-VBR except that
Categories there is no latency requirement (hence the non-real-time
modifier). Applications that might use this category are
The ATM Forum, the body of people charged routers connecting LANs together. Clearly this is a
with setting ATM standards, has defined 5 categories preferred category for carriers.
of ATM service, namely:

Unspecified Bit Rate (UBR)

Constant Bit Rate (CBR) This category is the inverse of the above; applica-
CBR applications need a fixed amount of band- tions that are non-real-time ones with no required
width with low latency. Technically, a CBR service service guarantees. There is no guarantee that the cells
resembles a private analog line or a Time Division will even be delivered! As shall be seen, ATM provides
Multiplexed circuit. The service might guarantee the carrier with the option of dropping certain cells if
10Mbps for an Ethernet link, for example. Some the network gets congested. Some wags refer to this
typical CBR applications are circuit emulation, voice, service as the “send-and-pray” category.
video, and any type of data that is time-defined. CBR
applications are not bursty.
Available Bit Rate (ABR)
This service guarantees the application a minimum
Real-time Variable Bit Rate (rt-VBR) of cell transfers per unit time. However, if there is
These applications require guarantees on the delay excess capacity in the net, the application can take
(hence the “real-time”) and a limited variation on the advantage of it and increase the number of cells sent
delivery of the cells to their destination, but do not per second. ABR applications can expect no latency
require a fixed bandwidth. It is assumed that rt-VBR guarantees and must be able to suffer some cell loss but
applications will be bursty and real-time, as opposed not a considerable amount. Thus, it would be a useful
to CBR which are real-time and constant. There still service for file transfers. If the bandwidth were there,
need to be qualifications on the data as we will see the rate of data sent could be increased to soak up the
(such as a limit on the number of cells presented, excess link capacity. When cells of a higher priority hit
maximum burst size, etc.) Applications that might use the switch, that application flow would be throttled
this category include certain voice and video applica- back to the guaranteed minimum with a possible cell
tions. loss or two. The application would of course have to
do a little retransmission to correct for the lost cell.
Implicit in the standard is that there will be a fair-
allocation scheduling in place at the switch to avoid
one application hogging the extra bandwidth at the
expense of other ABR applications.

34 • ATM Interworking Access at the Customer Premises Appendix • 35

ATM Quality of Service (QoS) Parameters Cell Delay Variation (CDV)
Each ATM connection can negotiate three impor- Cell Delay Variation describes the maximum
tant QoS parameters with the edge switch, namely: variation permitted in the transmission of a series of
• Cell Loss Ratio (CLR), cells. Sometimes it is called cell jitter. Imagine a source
generating a string of cells, evenly spaced in time.
• Maximum Cell transfer delay (maxCTD), and Then the series may either clump up or stretch out
• Cell Loss Ratio (CLR). with gaps, depending on how the various switches
process the string and how congested the net is as the
cells pass through. For time-sensitive applications like
Cell Loss Ratio (CLR)
voice and video, it is vital that this be below the human
CLR is the ratio of lost cells/total transmitted cells. perception time. Thus, this parameter guarantees that
Cells are mainly lost when the switch decides to discard the cells will not distort much in their delivery.
them when the network becomes overloaded. Each
ATM cell of 48 data octets has a special bit called the
CLP bit (Cell Loss Priority bit). If this is set, this cell
will be discarded when the load exceeds the capability Appendix 4—Mapping Frame
of the link. Any cell that does not have this bit set will
proceed; that is if there is still room. Thus, the CLP
Relay Packets onto ATM Cells
bit gives a one-level priority scheme. These traffic types
are distinguished as priority traffic and non-priority Frame RFC-1490 User
Flag FCS Flag
traffic. Cells can be lost in other ways too. How a switch Header Header SDU

decides to discard cells in the presence of congestion is Frame Relay

Frame
vital to the health of the network. It is possible for a DLCI upper C/R 0
DLCI upper FECN BEC DE 1
switch to make matters worse by improper discarding
of cells. In third-generation switches, much research SERVICE
Mapping / Conversion
IWF
has been done to ensure clever discard algorithms.
GFC VPI
VPI VCI ATM
Maximum Cell Transfer Delay (maxCTD) VCI Cell
VCI PT CLP
The Cell Transfer Delay or CTD is the total time HEC
GFC =
VPI =
generic flow control
virtual path identifier
taken for the cell to transit through the ATM network RFC-1483 Header
Figure 12: Title???
VCI = virtual connection identifier
PT = payload type
Cell Payload (first segment of SDU)
from the source (the time the cell leaves the application) CLP = cell loss priority
HEC = header error checksum
to the destination (the time the cell enters the destination CLP = cell loss priority

application). It is the sum of all the ATM transit times

plus the sum of all of the ATM node processing times.
The maximum is the greatest delay that the application
can permit.

36 • ATM Interworking Access at the Customer Premises Appendix • 37

CASE STUDY1—Lower Operation Rural Offices
These offices usually require a low speed connec-
Costs and Future Technology tion (e.g. Fractional E1 or T1) and in order to reduce
costs, all traffic will be concentrated over Frame Relay.
Migration with Frame Relay In many cases Frame Relay is the only service available.
and ATM Internetworking
ATM-Frame Relay interworking objectives
The major objective of organizations when design-
• Minimize the investment in new equipment.
ing their corporate network is to save on communica-
Replace equipment only in offices that need more
tion cost—select the most economical service that meets
bandwidth or where lower cost services are avail-
their needs. Frame Relay is the most economical service
able. Reuse PBX’s, routers and Frame Relay
in many places around the world, however, it is limited
muxes over ATM or interwork with ATM equip-
to E1/T1 rates, therefore it may not fulfill bandwidth
ment.
needs of large regional offices. The ability to mix
high-end services such as ATM in branch offices that • Migrate the headquarters and some of the city
require them, with low cost services such as Frame offices to ATM in order to decrease cost of higher
Relay in all others, offers an economical solution with bandwidth connectivity
future migration path. • Increase the system efficiency by using voice com-
pression and voice switching at the Frame Relay
Today’s Network Structure level

Headquarters • Single vendor solution for both Frame Relay and

The corporation headquarters includes a large ATM
PBX for internal and external voice communication • System management application for the entire
and a router or LAN for internal and external data corporate network
information exchange. The migration to ATM is
• Support future ATM applications with today’s
typically driven by the headquarters that is the first to
equipment (for example video over ATM)
reach the E1/T1 bandwidth limit requiring more
bandwidth (NxE1/T1, E3/T3 or even STM-1), since • Dedicated solutions for small and medium size
most of the traffic is between it and the branch offices. branches to reduce the system cost

City Offices
These offices have PBX for voice and routers for
data, as well as a steady demand for more bandwidth,
as the branch activities are growing. They might
migrate to ATM depending on the amount of band-
width required, ATM cost vs. Frame Relay cost, and
the availability of the service.

38 • ATM Interworking Access at the Customer Premises Case Studies • 39

RAD’s ACE and the MAXcess Product This study shows how RAD’s two families—the
Families Solve Both Frame Relay and ATM ACE family and the MAXcess family— provide a
single vendor solution for the corporate network,
Networking Needs of the Corporation. enabling operation over ATM and Frame Relay net-
works in a cost effective manner. The interworking
Headquarters
between the two services can be done either by the
PBX
`service providers, as Frame Relay-ATM interworking
MX
3000 service, or by RAD’s ACE family. The migration to
Branch A Branch C ATM is done with a minimum investment, since there
ACE-101 MX
PBX
MX
30 PBX is no need to replace the existing equipment. High
ACE-20
3000 bandwidth, support in future services, and flexibility
Frame ATM demands are answered with the combination of the
Branch B Relay Branch D
MX
two RAD families of products.
PBX PBX
300
MX ACE-101
30

CASE STUDY 2—Efficient

The ACE-101, which implements both AAL1
and AAL5 interworking protocols can be used as the and Simple Control of ATM
migration path for the corporate headquarters. The
ACE-101 has a wide range of modular interfaces that
Quality of Service
enable the different users and network equipment to
interconnect and has a link rate between E1 and 155 Following large-scale ATM service trials, and prior
Mbps. The rate range enables the corporation to to rolling out nationwide and international native
increase the network bandwidth connection according ATM services, a major European carrier determined
to the changing needs, without replacing the equipment. the need for a carrier owned and controlled ATM
The traditional PBX and routers can be used or the Network Termination Unit to be provisioned at the
RAD MX-3000 can be used for transmitting and interface between the customer premises and the public
switching compressed voice over Frame Relay. ATM network. This would serve as the demarcation
RAD’s ACE-20 can be used in some of the city point between the ATM environment at the customer
offices where ATM is more economic than Frame premises and the carrier’s network. The device had to
Relay. The PBX and the router, which are already be QoS enabled, offer a superior cost-performance
available in the office, are used for both voice and data ratio, and support a minimum number of VP/VCs (256).
traffic. The carrier also required that flexible LAN/WAN
For areas where Frame Relay is more economical, connectivity options be built into the device, for
MX-300 and MX-30 can be used for concentration maximum on-the-spot compatibility between diverse
of voice and data over the Frame Relay networks. end user equipment and carrier transmission interfaces.
The products also perform voice compression, which The ACE-101 answers all these specifications. In
increase the utilization of the network and saves com- particular, ACE-101 satisfies the carrier’s requirements
munication costs.

40 • ATM Interworking Access at the Customer Premises Case Studies • 41

for implementing Operation, Administration and As ACE-101 demonstrates, the carrier is able to
Maintenance (OAM) support for end-to-end ATM exercise maximum control over ATM parameters,
service monitoring, including Continuity Check (CC), ensure traffic conformity with a low-cost dedicated
Alarm Indication Signal (AIS), and Remote Defect device managed from a central location. For end users,
Indication (RDI). ACE-101 offers modularity—adapting to diverse
ACE-101 also provides Performance Management interfaces—redundancy, and essential traffic shaping
(PM) features, checking peak cell rate, cell delay varia- (buffering of multiservice traffic according to predeter-
tion tolerance, sustainable cell rate, and maximum burst mined priority settings).
size of traffic entering the public network. Loopback The benefits of the ACE-101 family may also be
cells are used to measure cell loss and cell delay para- applied to other ATM service applications, such as at
meters end to end. the boundary between carrier jurisdictions in interna-
Of special interest to the carrier is the ACE-101’s tional networks.
ability to carry out “Reverse PM,” collecting statistics
on what is transmitted by the NTU into the network
as well as information on what is received from the
customer premises into the NTU device.
Policing of ATM traffic is applied to all connections,
preventing users from violating their service contract
by transmitting unauthorized ATM cells into the public
network. In parallel, the carrier implements Policing at
the edge switch to stop illegal user attempts to circum-
vent the ATM NTU.

ATM data Customer Premises

in-band management
ACE-101
Check ingress NMS User
traffic conformance Network
¥ Policing
¥ Monitoring

Switch Switch ACE-101

User User
Network Network
ACE-101
Switch Customer Premises
Customer Premises

Public Network
¥ QoS control
¥ Segment OAM between edges, including:
Fault Management - AIS, RDI, CC, LB
Performance Management - Forward and Backward
Continuity checks

Efficient and Simple Control

of ATM Quality of Service

42 • ATM Interworking Access at the Customer Premises Case Studies • 43

Glossary ABR (Available Bit Rate)—One of five ATM forum
defined service categories. In this service type, the net-
work makes the best effort to pass maximum cells but
AAL (ATM Adaptation Layer)—A collection of without an absolute guarantee for the cells delivery.
standardized protocols that adapt user traffic to the cell Supports variable bit rate data traffic with flow control,
format. The AAL is subdivided into the Convergence a minimum guaranteed data transmission rate and
Sublayer (CS) and the Segmentation and Reassembly specified performance parameters. In turn for regulating
(SAR) sublayer. There are currently 4 types of AALs: user traffic flow, the network offers minimal cell loss
AAL1, AAL2, AAL3/4 and AAL5 to support the of accepted traffic. Traffic parameters are PCR and
various AAL service classes. MCR. QoS parameters are CLR and CER

AAL1(AAL type 1)—Protocol standard used for the B-ICI (Broadband Inter-Carrier Interface)—An
transport of time-dependent Constant Bit Rate (CBR) interface that supports service connections (such as in
traffic (e.g. audio and video) and the emulation of CRS, CES, SMDS, FR) across public ATM networks.
TDM-based circuits (i.e. DS1, E1). Timing information CAC (Connection Admission Control)—An ATM
is required to be exchanged between the source and the control function which determines whether a virtual
destination. AAL1 supports QoS Class A. connection (VC) request should be accepted or rejected.
AAL2 (AAL type 2)—Protocol standard for supporting The decision is based on the QoS required, the network
time-dependent slow or Variable Bit Rate (VBR-RT) resources available and the availability of the connection
connection-oriented traffic (e.g. packetized and com- over the network.
pressed audio and video). Timing information is CAS (Channel Associated Signaling)—Voice
required to be exchanged between the source and the signaling based on bits taken from voice time slots, used
destination. AAL2 supports QoS Class B. by many PABXs.
AAL 3/4 (AAL type 3 and 4)—Protocol standard for CBR (Constant Bit Rate)—One of the five ATM
supporting both connectionless and connection-oriented classes of service, which support the transmission of a
Variable Bit Rate (VBR-NRT) traffic. AAL3 supports continuous bit-stream of information where traffic such
QoS class C while AAL4 supports QoS class D. They as voice and video, require a constant amount of band-
are currently combined into one type. AAL3/4 also width allocated to a connection for its duration.
performs resequencing and cell identification operations.
AAL3/4 services are suitable for supporting interworking CC (Continuity Cell)—A cell used periodically to
with Frame Relay, SMDS and X.25. check whether a connection is idle or has failed.
Continuity checking is one of the OAM function
AAL5 (AAL type 5)—Protocol standard for supporting types for fault management
connection-oriented Variable Bit Rate (VBR-NRT) data
traffic and signaling messages. AAL5 supports QoS CCR (Current Cell Rate)—The currently acceptable
Class X. AAL5 services are suitable for supporting transmission rate for an end-system as defined by RM
interworking with most data networking protocols, such cells within ABR. The field in the RM cell indicates the
as Frame Relay, SMDS, Ethernet and IP. current complying cell rate (i.e. ACR) a user can transmit
over a virtual connection (VC).

44 • ATM Interworking Access at the Customer Premises Glossary • 45

CDV (Cell Delay Variation)—A QoS parameter CLR (Cell Loss Ratio)—A QoS parameter that
that measures the difference between a single cell’s gives the ratio of the lost cells to the total number of
transfer delay (CTD) and the expected transfer delay. transmitted cells.
This parameter is important for time sensitive virtual
Congestion control—A resource and traffic manage-
circuits such as CBR and VBR-RT.
ment mechanism to avoid and/or prevent excessive
CDVT (Cell Delay Variation Tolerance)—Used situations (buffer overflow, insufficient bandwidth) that
in CBR traffic, it specifies the acceptable tolerance of can cause the network to collapse. Congestion control
the CDV (jitter). schemes may be based on fields within the ATM cell
header (CLP, EFCI within the PTI) or may be based on
CER (Cell Error Rate)—A QoS parameter that
a more sophisticated mechanism between the ATM
measures the fraction of transmitted cells that are
end-system and ATM switches. The ATM forum has
erroneous over a specific period of time (i.e. those that
developed a mechanism based on rate control for ABR-
contain errors when they arrive at the destination).
type traffic.
CES (Circuit Emulation Service)—ATM Forum-
CRM (Cell Rate Margin)—A measure of the resid-
defined service that provides a virtual circuit connection,
ual useful bandwidth for a given QoS class, after taking
which emulates the characteristics of a real, constant-
into account the SCR.
bit-rate, dedicated-bandwidth circuit. Traffic over ATM
networks that comply with the other ATM Forum inter- CRS (Cell Relay Service)—A bearer service offered
operability agreements. Specifically, this specification by an ATM network, to the end users that delivers
supports emulation of existing TDM circuits over ATM ATM cells directly.
networks.
CS (Convergence Sublayer)—The sublayer of the
CI (Congestion Indicator)—A field in the RM-cell ATM AAL where traffic is adapted based on its type
that indicates congestion in the network which can before undergoing segmentation into cells (SAR
ultimately lead to the source reducing its allow cell rate process). The CS includes the CPCS and the SSCS.
(ACR).
CTD (Cell Transfer Delay)—A QoS parameter
CIR (Committed Information Rate)—A term that measures the average time for a cell to be trans-
used in Frame Relay, which defines the information rate ferred from its source to its destination over a virtual
the network is committed to provide the user with, connection (VC). It is the sum of any coding, decoding,
under any network conditions. segmentation, reassembly, propagation, processing and
queuing delays.
Classical IP and ARP over ATM—A standardized
model where ATM acts as a high-quality link layer ELAN (Emulated LAN)—The ATM segment of a
transport for higher-level protocols. The classical model virtual LAN (VLAN) based on the ATM forum LANE
is defined in RFCs 1577 (Laubach, 1993) and 1483 standard. A VLAN consists of an ELAN segment along
(Heinanen, 1993), and was first deployed ATM with traditional LAN segment.
architecture.

46 • ATM Interworking Access at the Customer Premises Glossary • 47

EPD (Early Packet Discard)—A procedure for IMA (Inverse Multiplexing over ATM)—
discarding cells related to one user frame to minimize A method to pass ATM traffic over multiple E1/T1
the impact of congestion. Can be implemented on any links while keeping the ATM’s quality of service and
virtual circuit using AAL5 since the PTI field indicates optimize the bandwidth usage.
last cell.
I-PNNI (Integrated Private Network-Network
FC (Feedback Control)—An end-to-end network Interface)—Protocol used to exchange reachability
control mechanism to regulate the rate at which a information between routers that augment or replace
sender can inject cells into an ATM network, based on protocols such as OSPF and IPX and is compatible
feedback received from the RM-cells. with PNNI. This enables the integration of existing
router-based connectionless networks with ATM
FD (Frame Discard)—A procedure for discarding
networks.
cells related to one user frame to minimize the impact
of congestion. Can be implemented on any virtual Leaky Bucket—A flow control algorithm, where cells
circuit using AAL5 since the PTI field indicates last cell. are monitored to check whether they comply with the
established connection parameters. Non-conforming
FRF.5—Network interworking function that facilitates
cells are either tagged or dropped from the network.
the transparent transport of Frame Relay traffic over
The analogy is taken from a bucket with a hole in its
ATM, connecting two Frame Relay networks over an
bottom that allows the fluid to flow out at a certain rate.
ATM backbone.
LIS (Logical IP Subnetwork)—An IP subnetwork is
FRF.8—Service Interworking function that facilitates
a single network on which all devices have a direct com-
protocol conversion, allowing communications between
munication path to all other devices. Examples would
dissimilar Frame Relay and ATM equipment.
be a LAN or a point-to-point circuit. An LIS is a network
FUNI (Frame User Network Interface)—Frame- in which the IP protocol can operate as if all devices
based interface to ATM supporting signaling and QoS. are directly connected, even if they are not, such as a
To interoperate with a Frame Relay end system, the virtual circuit-based network.
ATM switch should support FRF.8 which is the Frame
MBS (Maximum Burst Size)—A traffic parameter
Relay/ATM Service Internetworking specification.
that specifies the maximum number of cells in a burst
Replaces the ATM-DXI.
that can be transmitted at the peak rate (PCR).
HEC (Header Error Check)—A one-byte field
MCR (Maximum Cell Rate)—An ABR traffic
within the ATM cell header providing for error detec-
parameter (in cells per second) that gives the slowest
tion. If an error is detected, the cell will be discarded
rate that the network controls the flow of the source on
before undergoing reassembly.
an ABR virtual connection (VC).
ICR (Initial Cell Rate)—The rate that a source is
MCS (Multi Cast Server)—Device for efficient
allowed to start up at following an idle period. It is
forwarding of multicast data within the classical model
established at connection set-up and is between the
and MPOA.
MCR and the PCR.

48 • ATM Interworking Access at the Customer Premises Glossary • 49

MCTD (Mean Cell Transfer Delay)—The average PNNI (Private Network-Network Interface)—
delay for ATM cells across an ATM connection. The interswitch interface within a private ATM
domain. The PNNI trunking protocol providing for
MPOA (Multi-Protocol Over ATM)—IETF
hierarchical ATM-layer routing and QoS support.
defined specifications and procedures that enable
Network Layer protocols to operate directly on top of Policing—A method to verify that the incoming VC
ATM and provide end-to-end internetworking between comply with the user’s service contract.
hosts in an ATM and non-ATM environment. PVC (Permanent Virtual Circuit)—A virtual
NHRP (Next Hop Resolution Protocol)—An connection established by the network management
IETF protocol for ATM address resolution between between an origin and a destination that can be left up
MPOA servers. permanently.
NNI (Network Node Interface (or Network-to- PVP (Permanent Virtual Path)—A term to
Network Interface)—ITU-T specified standard describe a set of grouped together permanent virtual
interface between nodes within the same network. The channels (PVCs) that exist between two cross points.
ATM forum distinguishes between two standards, one RDF (Rate Decrease Factor)—An ABR service
for private networks called P-NNI and one for public factor by which a source should decrease its transmis-
networks known as public NNI. sion rate if there is congestion.
OAM (Operation Administration and RFC 1483—Specifies the encapsulation of multiproto-
Maintenance)—Management framework defined by col data for transmission over an ATM network. RFC
the ITU. OAM cells are special purpose ATM cells 1483 make use of AAL5 in the support of PVCs and
exchanged between two ATM entities for network fault SVCs. The two methods defined in this RFC are VC
and performance management, analysis and fault isolation. muxing and LLC/SNAP encapsulation.
Payload—The 48-byte segment of the ATM cell RFC 1490—Specifies the encapsulation of multi-
containing user data. Any adaptation of user data via protocol data for transmission over Frame Relay. Frame
the AAL will take place within the payload. Relay over ATM uses this encapsulation in combination
PCR (Peak Cell Rate)—A traffic parameter (in cells with AAL5.
per second) that characterized the source and gives the RIF (Rate Interface Factor)—An ABR factor by
maximum rate at which cells can be transmitted. It is which a source can increase its transmission rate if the
calculated as the reciprocal of the minimum intercell RM cell indicates no congestion.
interval (the time between two cells) over a given virtual
RM (Resource Management cell)—A cell used
connection (VC).
for carry network resource management information,
PHY (Physical layer)—The bottom layer of the such as available buffer space, residual bandwidth and
ATM protocol reference model, it is subdivided into indicate the presence of any congestion in the network.
two sublayers: Transmission Convergence (TC) and RM cells are injected in regular data cells streams in
Physical Medium (PM). It provides ATM cell transmis- order to monitor and report the network conditions. A
sion over the physical interfaces that interconnect the value of 6 is reserved in the PTI field of an ATM cell
ATM devices. to indicate an RM cell.

50 • ATM Interworking Access at the Customer Premises Glossary • 51

RSVP (Resource reSerVation Protocol)—A proto- and video. AAL2 supports this delay sensitive,
col developed for supporting different QoS classes for connection oriented class.
IP applications.
Class 3 specifies the parameters for connection
RTT (Round Trip Time)—The round trip time it oriented data transfer. AAL3/4 and mostly AAL5
takes for a packet to travel between a source and a supports this delay tolerant class which is intended
network device. In ATM, RTT is usually measured in to provide interoperability with SMDS and IP.
numbers of cells.
Class 4 specifies the parameters for connectionless
SAAL (Signaling AAL)—Service specific parts of the data transfer. AAL3/4 or AAL5 can be used to
AAL protocol responsible for signaling. Its specifications, support this delay tolerant class which is also
were adopted from N-ISDN. intended to provide interoperability with SMDS
and IP.
SAP (Service Access Point)—Reference point
between the networking stack and applications within Class X refers to the connection oriented transport
an end system. service where the traffic type (CBR or VBR) and
timing requirements (delay sensitive or non-sensitive)
SAR (Segmentation And Reassembly)—Segments
are defined by the user. It is known as an unrestricted
the information frames into cells at the source and
service class and which is supported by AAL5.
reassembled these cells back into information frames at
the destination. These activities occur at the lower part SVC (Switched Virtual Circuit)—A logical ATM
of the AAL. Each AAL type has its own SAR format. connection established via signaling. End systems trans-
mit their UNI 3.1 or 4.0 signaling request via the
SCR (Sustainable Cell Rate)—A traffic parameter
Q.2931 signaling protocol.
that characterized a bursty source and specified the
maximum average rate at which cells can be sent over a TM (Traffic Management)—Set of actions and
given virtual connection (VC). It can be defined as the operations performed by the network to guarantee the
ratio of the MBS to the minimum burst interarrival operability of the network. TM is exercised in the form
time. of traffic control and flow control. The ATM traffic
management includes the following: CAC, FRM,
Service Categories—A group of service classes
NRM, Priority Control (PC), Traffic Shaping (TS) and
defined by the ATM forum in terms of different QoS
UPC , which the goal is to maintain the required QoS.
parameters:
Traffic policing—Mechanism whereby any traffic,
Class 0 refers to the best effort service (UBR).
which violates the traffic contract, agreed to at connec-
Class 1 specifies the parameters for circuit emulation, tion setup is detected and discarded.
and the transport of CBR uncompressed video and
Traffic shaping—A method for smoothing the bursty
for VPNs. AAL1 supports this kind of delay sensitive
traffic rate that might arrive on an access virtual circuit
connection oriented service.
so as to present a more uniform traffic rate on the net-
Class 2 specifies the parameters for the transport of work and comply with the traffic contract.
VBR (low speed or compressed packetized) audio

52 • ATM Interworking Access at the Customer Premises Glossary • 53

UBR (Unspecified Bit Rate)—One of the best VC (Virtual Connection)—A connection established
effort service types. Realistically, no traffic parameters between end users, where packets are forwarded along
are specified by the source, so , no actual quality com- the same path and bandwidth is not permanently
mitment is made by the network. allocated until it is used.
UNI (User Network Interface)—The interface, VCC (Virtual Channel Connection)—An end to
defined as a set of protocols and traffic characteristics, end connection consisting of a concatenation of two or
between the CPE and the ATM network. more Virtual Channels between two end points. VCCs
may be bundled into a VPC.
UNI 2.0—ATM forum UNI specification for the physi-
cal (PHY) and the ATM layers, the ILMI, OAM (traffic VLAN (Virtual LAN)—A network architecture which
control) and PVC support. allows geographically distributed users to communicate
as if they were on a single physical LAN by sharing a
UNI 3.0—An upgrade of UNI 2.0 that includes traffic
single broadcast and multicast domain. ATM forum
control for PCR and some recommendations regarding
LAN emulation supports VLANs.
the operation over current transmission systems.
VP (Virtual Path)—A term to describe a set of
UNI 3.1—A correct version of UNI 3.0, this specifica-
Virtual Channels (VCs) between cross points, grouped
tion also includes SSCOP standards.
together.
UNI 4.0—This UNI specification refers to signaling
VPC (Virtual Path Connection)—An end to end
issues in ABR, VP and QoS negotiations.
connection consisting of two more Virtual Path links
VBR (Variable Bit Rate)—Traffic containing bursts (VPs)
but centered around an average bandwidth. VBR,
VPN (Virtual Private Network)—Public network
divided into real-time (RT-VBR) and non-real-time
service where a customer is provided a network which
(NRT-VBR) traffic requires the same service guarantees
appears as if it were a private network. The advantage
(that is delay, cell loss and timing) provided by CBR.
of VPNs, over the dedicated private networks, is that
VBR-NRT (Variable Bit Rate Non Real Time)— the former allow a dynamic use of the network
One of the two VBR service types for transmitting resources and offer a very reliable, high speed and less
traffic where timing information is not critical. Since expensive communications.
this service type is delay tolerant, it is well suited for
bursty traffic such as data communications.
VBR-RT (Variable Bit Rate Real Time)—One
of the two VBR service types for transmitting traffic
that depends on timing and control information. It is
suitable for carrying delay sensitive traffic such as
packetized video and audio.

54 • ATM Interworking Access at the Customer Premises Glossary • 55

 ACE Family of ATM Access Products
The ACE family of ATM Network Termination Units
(NTUs) and Interworking Access Devices offers real
benefits to carriers and their ATM customers:
• Extends the ATM network up to the customer’s premises
Visit ATG’s Web Site • Reduces maintenance and operations costs
• Improved diagnostics
to read, download, and print • Ensures that service level agreement is maintained
• Wide range of interfaces to support different customer
all the Technology Guides equipment
• Faster deployment of native ATM services
in this series.
The ACE family includes:
• ACE-101 ATM Network Termination Unit
• ACE-2 ATM Single Service Interworking NTU
• ACE-101 ATM LAN and Voice Interworking NTU
• ACE-20 ATM Triple Service Interworking NTU
• ACE-2005 ATM Multiple Service Interworking
www.techguide.com Access Concentrator

ATM
Backbone

E1/T1, E3/T3, STM-1, STS-3C

E1/T1, E3/T3, STM-1, OC-3C
E3/T3, STM-1,
E1/T1 E1/T1
STS-3C

ACE-2005 ACE-20 ACE-101 ACE-2 ACE-101

FR/ATM Device PABX PABX

FR, Ethernet,
ATM DXI ATM
Frame Relay PABX Transport

“The significant problems we face cannot be solved

by the same level of thinking that created them.”
www.rad.com
Albert Einstein U.S. Headquarters International Headquarters
Tel: (201) 529-1100 Tel: 972-3-6458181
Fax: (201) 529-5777 Fax: 972-3-6498250
email: market@radmail.rad.co.il
 This Technology Guide is one in a
comprehensive series of Guides that
provide objective information and
practical guidance on technologies
related to Communications & Networking,
the Internet, Document Management,
Data Warehousing, and Enterprise
Solutions. Our team of technical editors
writes each Technology Guide to
assist IT and business professionals
in making informed decisions about
all aspects of technology application
development and strategic deployment.

techguide.com is supported by a
consortium of leading technology
providers. RAD Data Communications has
lent its support to produce this Guide.

Visit our Web Site at www.techguide.com

to view and print this Guide, as well
as all of our other Technology Guides.
This is available as a free service.