The Technology Guide Series

techguide.com

Inverse
Multiplexing
Scalable Bandwidth
Solutions for the WAN

This guide has been sponsored by

 Table of Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
The Connectivity Revolution. . . . . . . . . . . . . . . . . 2
The What and Why of Inverse Multiplexing . . . . . 5
Inverse Multiplexing Considerations . . . . . . . . . . 13
Inverse Multiplexing and Broadband Network . . . .
Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Inverse Multiplexing and ATM . . . . . . . . . . . . . . 18
Inverse Multiplexing Implementations. . . . . . . . . 27
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Glossary of Terms. . . . . . . . . . . . . . . . . . . . . . . . . . 33

About the Editor…

Jerry Ryan is a principle at ATG and Technology Editor of the
Technology Guides on Communications and Networking. He was
the president and founder of Connections Telecommunications,
Inc., a Massachusetts based company specializing in consulting, edu-
cation and software tools which address network design and billing
issues. Mr. Ryan is a member of the Networld+Interop 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 Tech-
nologies 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 contained herein may not be duplicated in
other books, databases or any other medium. Making copies of this book, or
any portion for any purpose other than your own, is a violation of United
States Copyright Laws. The information contained in this report is believed
to be reliable but cannot be guaranteed to be complete or correct.
Copyright © 1998 by The Applied Technologies Group, One Apple Hill,
Suite 216, Natick, MA 01760, Tel: (508) 651-1155, Fax: (508) 651-1171
E-mail: info@techguide.com Web Site: http://www.techguide.com
Introduction cost of bandwidth. Most LANs today operate at the 10
Mbps Ethernet speed. (In reality, the speed of a fully
optimized 802.3 network is closer to 2 Mbps as
Inverse multiplexing is truly the inverse of traditional contentions for the network increase.) New LAN tech-
multiplexing. With traditional multiplexing, multiple data nologies have increased the choice of operating speeds
streams are combined and transmitted over a single high- by orders of magnitude: to 100 Mbps and 1Gbps.
speed circuit; with inverse multiplexing, a single high- With higher LAN speeds readily available, increasingly
speed data stream is transmitted over multiple circuits. sophisticated, bandwidth-intensive LAN-based applica-
Because the term “inverse multiplexing” is a bit unwieldy, tions have become common.
the shorter terms “inverse muxing” and “imuxing” Because of the services traditionally offered by telco
(pronounced eye-muxing) are frequently used. providers, there has always been an asymmetrical rela-
A number of factors combine to make inverse tionship between LAN bandwidth and WAN
multiplexing a major technology in today’s bandwidth. Many organizations use 56/64 kbps services
internetworking environment. This booklet will for wide-area connectivity. Larger organizations may
examine those factors, discuss the advantages of employ T1 WAN services, which still require LAN
inverse muxing solutions, and show the place of inverse traffic to throttle down considerably (to 1.544 Mbps). For
muxing in an evolutionary broadband strategy. organizations needing native LAN speeds over their
WAN links, the primary alternative has been relatively
expensive T3 (45 Mbps ) circuits. Similar LAN/WAN
asymmetry exists in Europe, where the E1 (2.048 Mbps )
The Connectivity Revolution and E3 (34 Mbps ) services are frequently unavailable or
cost-prohibitive. For organizations with high-speed ATM
Not so long ago, putting computers on desktops LAN or backbone networks, the discrepancy between
was the key element of a revolution in the way business local and wide-area bandwidth can be even greater.
is done on a daily basis. Now, with desktop computing a NSPSs are taking steps to alleviate the WAN bottle-
well-established norm, the ongoing revolution is in the neck, with wide-area ATM and high-speed frame relay
way computers are interconnected. Computer users are delivering native LAN speeds. Deployment of these
continually being offered, and then demanding, access technologies is taking time, however. In the meantime,
to ever-more-powerful applications and services. The as we shall see, there are solutions that allow companies
challenge for network designers and implementers is to to extend their LANs over the wide-area at speeds
meet the rapidly-growing connectivity requirements of between T1 and T3 (or the European E1 and E3).
their sophisticated user base.
The Internet and Intranets
The LAN Effect Traditionally, the emphasis in corporate
The emergence of the local area network (LAN) as networking focused on local connectivity. In the past,
the core architecture in the corporate enterprise had a 80% of LAN traffic has remained within the local
dramatic impact on the networking industry. One of domain, with only 20% of the traffic going elsewhere.
the chief characteristics of LANs is the relatively low In today’s environment, this statistic has reversed and

2 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 3

over 80% of LAN traffic is destined for locations
outside of the local domain.
The What and Why of Inverse
Furthermore, virtually all corporate PCs are Multiplexing
connected to some type of LAN and/or network access.
It is apparent that PC connectivity is reaching saturation As shown in the diagram below, inverse
in the U.S. Most corporations deploy routers to provide multiplexing is a physical-layer technology in which a
wide-area connectivity. With these devices supporting data stream received from a router, switch, or other
higher-speed LAN infrastructures, the demand for wide- device is transmitted across multiple wide-area circuits,
area bandwidth is increasing dramatically. then reconstructed at the other end of the link into the
Today, LAN-to-WAN interconnection is driven in original data stream. The individual circuits connecting
large part by the emergence of the Internet and World the inverse multiplexers appear as a single high-speed
Wide Web as primary business tools. The Internet and data pipe to the DTE.
the Web encourage users to think of global Variable-delay T1 circuits

connectivity, of a kind of world-wide LAN, and

dramatically increases the value they place on high- Inverse T1 Inverse
Incoming Data

Outgoing Data
Multiplexer Multiplexer
T1
speed WAN connectivity. At the same time, the famil-
T1
iarity and features of Internet technology are causing Data distributed Data
bit- or cell-wise to T1 reassembled
many network designers to deploy Internet-technology T1 lines in proper order

networks, called Intranets, for wide-area connectivity

within their organizations. Inverse Muxing
Inverse multiplexing can be used to aggregate low-
Corporate Issues speed lines, as well as T1/E1 and higher-speed circuits. In
particular, inverse multiplexing provides an efficient way to
One of the biggest networking issues facing
bridge the bandwidth gap between T1 and T3 (or E1 and
corporations is the cost of transmitting information
E3). It is thus an important tool for network managers
across the wide-area. Network managers are looking
desiring to alleviate the bottleneck in the wide-area. For
for solutions that help reduce their WAN costs, which
example, assume that an organization needs to transport a
typically represent a large portion of the networking
critical application across the wide-area, and that the
budget. While there is technology available today that
application requires 6 Mbps of bandwidth. In most cases,
enables companies to optimize their WAN bandwidth
the most efficient way to achieve this 6 Mbps WAN link
(ATM, for example), the equipment to support that
will be to inverse-multiplex four T1 (or 3 E1) circuits.
can be costly.
From an application standpoint, inverse
The other important issue for network managers is
multiplexing provides a uniquely scaleable solution.
the ability to migrate to a new solution without
Through the use of aggregate bandwidth, the
rendering the current architecture obsolete. Whether it’s
technology provides high-speed applications with
a terminal-to-host environment or a client/server envi-
comparable wide-area rates, allowing network managers
ronment, a smooth migration path must be ensured.
to more closely match WAN bandwidth to application
Inverse multiplexing can provide that migration path.
bandwidth than would otherwise be possible. And, as

4 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 5

network bandwidth requirements increase over time, realistic broadband option. In general, this makes
inverse multiplexing facilitates the incremental growth multiple E1 inverse multiplexing even more attractive
of WAN links. than in the T1/T3 environment.
Several companies, including Larscom, have imple- ATM service, both high (45 Mbps, 155 Mbps) and
mented high-speed inverse multiplexing in their products low (1.544 Mbps) speed, is slowly becoming available
as a way to bring the benefits of broadband networking in limited geographic areas. Service provider pricing
to companies not yet ready to commit to the more costly for ATM has not standardized, however, and monthly
and less readily available options of T3/E3, fees vary greatly. Some providers may offer T1 ATM at
SONET/SDH, and high speed ATM. Larscom’s rates similar to a TDM-based T1 circuit, while others’
inverse multiplexing products include the Mega-T and service prices are considerably higher. Based on
Mega-E Multiple T1 and E1 Inverse Multiplexers; the recently published rates, ATM DS3 and OC3c
Orion 4000 Broadband Access Multiplexer; and the monthly charges can run in excess of $10,000, and
Orion 2000 IMA Inverse Multiplexer. vary based on the number of cells transported,
distance, PVCs/SVCs, etc. For those companies who
wish to interconnect their local ATM networks at
Who Needs Broadband Connectivity?
bandwidths higher than T1, ATM inverse multiplexing
Candidates for broadband solutions include provides an affordable and readily available solution.
companies with one application that requires more In order to decide whether a network justifies a
bandwidth than 1.544 Mbps, or a number of applica- broadband solution, a cost analysis must be conducted.
tions that require multiple T1s. Based on the most As mentioned earlier, this can be difficult because
recent AT&T pricing, the cost of multiple T1s will be pricing varies from carrier to carrier. Issues to consider
more economical than a full T3 when there are fewer in broadband pricing include usage-based pricing
than eight T1s needed. versus flat rate pricing; any mileage charges associated
When considering the cost differences between T3 with the service; and projected future pricing trends.
and T1 services, ratios in the U.S. typically range from
8:1 up to 10:1. For example, monthly charges for T3
access lines can range from $4,000 to $8,000 per The Installed Environment
circuit, compared to $400 to $800 for a T1 circuit. In Once the decision has been made to move to a
Canada, the ratio can be higher. broadband solution, how to migrate from the current
From a global perspective, broadband tariffs vary networking architecture should be the next consideration.
depending on the country, region in the country, and As companies contemplate migrating to a broadband
distance from the local central office. Additionally, many environment, an important consideration is the preserva-
providers price broadband services on a case-by-case tion of the installed infrastructure. Migration should be
basis, which makes cost comparisons very difficult. In facilitated without a “forklift upgrade”.
Europe, the ratio of intra-country E3-to-E1 costs can be Within the corporate environment, many compa-
between 10:1 and 20:1 while inter-country connection nies have established separate networks for data, voice,
ratios may be considerably lower (3:1 or 4:1). However, and video applications to ensure throughput and
the availability of E3 is quite limited in many countries, response times. However, supporting multiple networks
in which case aggregating E1 circuits is virtually the only can be both cumbersome and expensive. Implementing

6 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 7

a broadband network enables an organization to This is especially true for routers connecting small branch
consolidate network traffic, thus alleviating the cost offices. However, routers located at headquarters sites or
associated with parallel networks. central locations often support T1/E1, T3, or even high
Most enterprise networks are based on routers speed connections. While fast, T1/E1 connections are still
and/or T1/E1 multiplexers, which alone or in tandem quite a bit slower than native LAN bandwidth, and the
form the basis of the enterprise backbone. Because of higher speed connectors can be quite expensive.
the mission-critical nature of LANs, and the desire to In order to transport LAN traffic across the wide
expand corporate resources to all levels of the organi- area at high speed rates, many routers may be config-
zation, there has been a dramatic increase in the flow ured with a high-speed data interface such as V.35 or
of LAN traffic over the wide-area. In most cases, LAN the high speed serial interface (HSSI). To gain access to
applications and Internet connectivity are the drivers greater WAN bandwidth, the existing DSU/CSU
of broadband solutions. could then be replaced with an inverse multiplexer, a
broadband multiplexer, a DS3 DSU, or even an ATM
DSU. If other applications are expected to take advan-
Router Migration
tage of the same broadband network used by the LAN
Because of the increased numbers of users on LANs, traffic, a broadband multiplexer or ATM edge concen-
many corporations are moving from a shared medium to trator that can support additional existing networking
a switched environment in order to avoid potential LAN devices is the best solution.
congestion problems. Technologies such as 100 Mbps
Ethernet, 1 Gbps Ethernet, and ATM are now being
used to support the growing LAN infrastructure. T1 Multiplexer Migration
Dedicated 10 Mbps to the desktop, the increased T1 multiplexers provide consolidation of low-
processing power of PCs, and the growing importance speed traffic and voice over a single T1 link. However,
of the Internet and the World-Wide Web are driving for WAN connectivity, many organizations have
wide-area bandwidth requirements higher. Routers outgrown or will soon outgrow T1/E1 multiplexers in
supporting these higher-bandwidth LAN environments, favor of broadband solutions.
more than any other network traffic, force network Inverse multiplexing multiple T1/E1s can provide
managers to consider a broadband solution. Multi- the first step to broadband networking for a single
protocol routers are the core building blocks for inter- application or a number of applications. Full T3/E3
connecting LANs. Routers are data terminal devices services should be considered as more applications are
(DTE) that operate at layer three (the network layer) of added to the broadband network, and bandwidth
the OSI stack. For wide-area connectivity, these devices requirements cost-justify the service.
must connect to a data communications device such as Using a product like the Orion 4000 from
a DSU/CSU, which performs circuit grooming and Larscom enables a company to move to a broadband
synchronization of the data stream before it hits the solution in a gradual manner. The investment in
carrier network. broadband technology should allow the company to
With LAN traffic driving bandwidth requirements, pay as it grows. The Orion 4000 helps accomplish this
DSU/CSUs will become an important link to broadband task in three ways.
networking. Many routers today operate at 56/64 kbps.

8 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 9

 First, through its inverse multiplexing capabilities, with a problem. WAN bandwidth does not match that
the Orion 4000 can provide fractional T3/E3 services of the local area. The most common speeds for
for a single application or a number of applications. networking in the U.S. are 56 kbps and T1 (1.544
(Fractional T3 or E3 refers to bandwidths which are Mbps). In Europe, the most common speeds are 64
multiples of T1 or E1, but less than the full T3 or kbps and E1 (2.048 Mbps ). However, there are
E3.) For example, Larscom’s inverse multiplexers are services like Frame Relay which allow users to
deployed by several carriers to provide multi-megabit subscribe to specified bandwidths, from sub-T1 up to
Frame Relay service at fractional T3 rates. Second, a full 45 Mbps.
the Orion 4000 can also be configured as a T3 access T1/E1 and lower speeds may be suitable for appli-
multiplexer, when bandwidth requirements drive full cations such as file transfer, E-mail, and database access.
T3 access. Finally, the modular design of the product Today, however, the most common demand for wide-
allows a company to incrementally change and area access comes from routers and Internet/Intranet
expand their connections as their network evolves, connections. Many LAN applications require WAN
while keeping training costs to a minimum. connections at speeds greater than 1.5 Mbps or 2.048
Investing in a broadband multiplexer that Mbps. Additionally, a growing number of business
supports a variety of user interfaces and carrier networks must support bandwidth intensive applications
services allows a company to leverage its existing such as video or imaging, requiring bandwidth that
architecture, and migrate applications as needed. exceeds T1 speeds. T3/E3 is the next closest option,
Many companies are already looking for solutions but for companies who need more than T1/E1 speeds,
beyond the capacity of their existing equipment. Both the cost of T3/E3 is often difficult to justify or the
the T1 multiplexer and the router will continue to service is unavailable.
play a critical role in a broadband network as feeder In the LAN environment bandwidth is free. WAN
devices to a broadband multiplexer. The existing bandwidth, however, is an expensive resource that
carrier services can be used, and the network is now must be managed efficiently. There are two ends of
“future-proofed” to handle any changes in the appli- the spectrum when it comes to network design. At
cation environment. one end, there are companies that build extra
capacity into their network to allow for peaks and
bursts of information and disaster recovery. At the
Wide-area Services Options
other end of the spectrum, there are companies that
The next issue for broadband migration is the have not designed sufficient capacity for peaks or
ability to provide bandwidth as needed using existing traffic overflow.
carrier services. Using inverse multiplexing in a Operating at either end of the spectrum can be a
broadband solution provides incremental capacity costly situation for companies. Any extra capacity
through the use of T1/E1 circuits. Inverse built into the network must be paid for, even when it
multiplexing of multiple T1/E1 lines provides a high- is not used; and having no spare capacity can be
speed alternative to expensive or unavailable T3/E3 devastating to business operations whenever a
circuits or ATM lines. primary link has failed. Inverse multiplexing is a
LAN managers who want to extend LAN wide-area technology which allows bandwidth to be
resources across the wide-area are quickly confronted added incrementally.

10 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 11

Load Sharing vs. Inverse Muxing Routing the traffic over multiple wide-area links helps
Many router vendors will tout load sharing as a way to reduce the impact of any single link failure.
to achieve additional wide-area bandwidth. There are, Another issue with load sharing is the management
however, a number of disadvantages to load sharing. complexity associated with its use. A load sharing
Load sharing is similar to inverse multiplexing in that it router must support multiple T1/E1 access devices to
provides additional bandwidth using multiple T1/E1 provide additional wide-area bandwidth. In a managed
connections. The most common form of load sharing, in environment, using the simple network management
which an individual application is assigned to a partic- protocol (SNMP), each T1/E1 access device must have
ular T1/E1 port on the router, is called “route caching”. a separate IP address. An inverse multiplexer replaces
the need for multiple external equipment and requires
N x T1
.. N x T1
.. Inverse connection to only a single router port.
V.35 or Inverse
T1 V.35 or
When all of the important decisions are made
Router
HSSI MUX . Network
. MUX HSSI
Router

regarding management, reliability, and scalability,

A. Inverse Multiplexed T1 inverse multiplexing offers the best immediate solution
for broadband connectivity.

T1 DSU/ T1s T1s

Load- .. Serial
CSU T1
T1 DSU/
CSU Serial
.. Load-
Sharing
Router . Ports
T1 DSU/
Network
T1 DSU/
Ports
. Sharing
Router
Inverse Multiplexing
CSU CSU

B. Router Employing T1 Load Sharing Considerations

Inverse Muxing vs. Load Sharing
Inverse multiplexing of T1/E1s has proven to be
Although load sharing supports multiple T1/E1 an economical means for achieving high-throughput
connections, those connections are typically dedicated connectivity in many applications. There are several
to a specific application. Therefore, each application is points to contemplate when comparing inverse multi-
guaranteed no more than 1.5 or 2 Mbps of WAN plexing solutions.
bandwidth. With inverse multiplexing, a pool of band-
width can be allocated among many different applica-
tions-limited only by the total bandwidth available. How Much Bandwidth?
Inverse multiplexing also provides a higher level of Most vendors, including Larscom, use only one bit (8
circuit protection in case of a link failure. Load sharing, kbps) from the T1 payload to carry a channel synchro-
which uses route caching, can be a detriment when nization signal, typically called the Metaframe bit. This
there is a link failure. With applications assigned to a allows the user payload to be multiples of 1.528 Mbps.
particular T1/E1 port, a T1/E1 circuit break can Some vendors, however, require an entire DS0 (64 kbps)
cause an application to time out. However, in an for channel synchronization. This means that the rate
inverse multiplexing solution, end-to-end delivery of available to the user is N x 1.472 Mbps. In an 8 T1
network traffic would still be guaranteed. If a link fails, system, this translates to 448 kbps less throughput!
the traffic is simply spread across the remaining links.

12 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 13

 The equivalent of the Metaframe bit is also far end. Unfortunately, the FDL cannot be used where
included in the IMA specification for inverse D4 framing is required. Even when ESF (Extended
multiplexing an ATM cell stream. A special “IMA Super Frame) framing is used, the FDL usually is not
Control Protocol” (ICP) cell is inserted once per 128 guaranteed end-to-end. Larscom uses an independent
cells in a T1 or E1 for synchronization. link contained in the Metaframe channel which is
completely separate from the T1 overhead. Thus, both
Metaframe Robustness ESF and D4 framing are supported, and end-to-end
communications are guaranteed as long as any T1 line
The Larscom Metaframe algorithm is very robust.
is operational.
If the Metaframe bit is hit with a random line error,
the Mega-T or Orion 4000 reports a Metaframe Error
(MFE) but does not start a realignment process. Thus, Channel Identification
the user data is completely unaffected. The Mega-T In bit-based inverse multiplexing, it is important
and Orion 4000 quickly recognize when a T1 is really that the T1s at one end are ordered the same as the
out of alignment, and will realign the T1s to ensure T1s at the other end of the connection. Although this
minimal disruption to the user. may seem to be only an installation concern, T1s can
An IMA link is similarly robust. If two consecutive get cross-wired at the patch panels or cross-connects as
ICP cells are errored, an “Out-of-IMA Frame” event is a result of maintenance or troubleshooting.
reported. If two additional (four total) ICP errors are In the Orion 4000 and Mega-T, a unique Channel
reported, the T1 is removed from the group. ID is sent on each T1 which identifies the card slot and
port number to the remote end. Thus, an operator can
Protection from False Framing easily detect that a cross-wiring has occurred. Customers
have found that this feature saves time and trouble.
The T1 frame structure was carefully designed to
In the cell-based Orion 2000 ATM inverse multi-
protect systems from falsely locking on bit patterns
plexer, cross-wiring of T1s is permitted. However,
which resemble the framing pattern. That is one of
should the network manager wish to see if cross-wiring
the reasons a prime number (193) is used as the frame
has taken place, the ICP cell includes a Link ID similar
length in T1 systems. However, inverse multiplexing of
to the Channel ID.
T1s can actually defeat some of the design safeguards
originally used to construct the frame. Larscom has
incorporated mechanisms which catch false framing Built-In BER Testing
conditions and force a reframe operation to take place. Bit error rate (BER) testing is a proven method for
Thus, the user is protected from getting into a falsely verifying the integrity of a composite channel or an
locked situation in which the T1s look acceptable, but individual T1. In the Orion 4000, a user can test the
the user data is corrupted. composite channel or individual T1s with QRSS and
user defined test patterns. A user can also take a T1 off-
Independent Inband Data Link line and perform a bit error rate test without disturbing
the other on-line channels, or perform an absolute
Some vendors’ inverse multiplexing products use
delay calculation using a bit error rate test.
the T1 Facility Data Link (FDL) to communicate to the

14 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 15

Network Management SNMP agent supporting the standard MIBs plus a
The one remaining element that brings the broad- complete private MIB extension, providing access to all
band solution together in a cohesive reliable network is status and configuration data. The Orion 4000 and
management. Support for the simple network manage- Orion 2000 also support multiple Telnet sessions
ment protocol (SNMP), which has become the recog- concurrent with SNMP access, and inband IP
nized standard in the industry for managing LAN and forwarding, so that a manager anywhere in the network
WAN environments, is critical. can monitor or control a system from his or her desktop.
In today’s multivendor environment, there are a
number of element management platforms that
support specific devices in the network. SNMP is an
important management tool that allows companies to
Inverse Multiplexing and
monitor, control, and reconfigure their entire Broadband Network Evolution
enterprise networks (LANs and WANs) from a single
console. SNMP-compatible devices include built-in
The steps to broadband networking are clear. For
SNMP software agents, allowing a management station
those companies employing T1/E1 links today, inverse
to poll all SNMP devices in the network and obtain
multiplexing provides an affordable, higher speed solu-
operating status information. This information is then
tion when applications and/or network traffic demand
used in configuration, performance, accounting, secu-
higher bandwidth. This is true for both TDM and
rity, and diagnostics functions. Tight interworking
ATM networks. When the application environment
between an SNMP based management platform and
dictates even more bandwidth than can be supplied by
higher level network management systems such as HP’s
inverse multiplexing products, migration to T3, E3,
OpenView and IBM’s NetView will also provide a user
and OC3c should then be cost justifiable.
with extensive control and monitoring capabilities.
The move to a broadband solution can be an
In most cases, an enterprise network represents an
evolutionary process, not an overnight occurrence. The
important corporate asset. As the network evolves to
first phase might begin as simply as supporting a single
support broadband applications, it is important to have
application with a broadband solution using inverse
a management system in place that can tie the entire
multiplexing-which can still work in tandem with the
network together so that network level decisions can be
existing network architecture.
made quickly and effectively. In fact, to prevent an
As previously discussed, LAN interconnection and
NMS failure from affecting the network access system,
the Internet/Intranets are the applications that drive
there should be redundancy in the NMS-which
much of the need for broadband networks. The types
increases cost and complicates management.
of information originating from LANs will change
If a network access system is used for router traffic,
dramatically to include video to the desktop, collabora-
the same techniques and tools used to manage routers
tive computing, and imaging. Internet sites frequently
should be applied to the other components of the
include sophisticated multimedia applications. These
system. Both the Orion 4000 and Orion 2000 come
all will put heavy demands on the network.
equipped with a built-in Ethernet management port plus
The next phase in the evolution to broadband
RS232 supervisory ports. They each have an integral
will allow a company to migrate applications to full

16 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 17

T3 support. The broadband multiplexer helps accom- enough bandwidth for an application (as in the case of
plish this with very little change in the design of the T1 ATM lines, which lose 10-20% of their bandwidth
network. The last phase will implement full support for to ATM’s inherent overhead).
ATM, which will provide an even greater level of wide- Inverse multiplexing brings the same benefits to
area bandwidth optimization. Broadband multiplexers ATM as it does to the traditional TDM environment:
and edge concentrators already provide viable migra- it combines cost effective, readily available T1 lines
tion paths to ATM, in service areas where high speed into a high-speed connection that supports the trans-
ATM is available. port of multi-megabit applications. ATM imuxing is
Many companies have taken the necessary steps to also extremely versatile: depending on the available
ensure peak performance in the LAN environment. service, ATM traffic may be inverse multiplexed as
Peak performance in the WAN will require access to either cells, across an ATM service, or as bits, across a
increasingly higher bandwidths, in both the TDM and traditional TDM infrastructure. In either case, adding
router-based networks. Such change, however, need not ATM inverse multiplexing to a company’s WAN struc-
be drastic, as solutions using technology such as inverse ture increases the bandwidth available to applications,
multiplexing will provide an immediate broadband while utilizing the lower speed access lines that are
solution with minimal cost and technological risk. This affordable and available.
gradual migration to higher bandwidth allows compa-
nies to pay only as their networking needs grow.
Combining Inverse Muxing and ATM
The ATM Forum, a consortium of networking
vendors established to recommend ATM specifications,
Inverse Multiplexing and ATM has defined a standard implementation of inverse
multiplexing for ATM applications. The specification,
called inverse multiplexing for ATM (IMA), employs
ATM is a cell-based technology that supports
cell-by-cell inverse multiplexing of the ATM cell
voice, video, and data over a wide range of transmis-
stream over multiple T1 or E1 circuits.
sion speeds. The biggest benefit of ATM is its ability to
IMA is a new UNI (User-to-Network Interface)
optimize wide-area bandwidth while accommodating
which rides on top of existing T1 or ATM PHY (phys-
the characteristics of various traffic types, with multiple
ical interface). The PHY defines how ATM cells are
classes of service.
mapped onto physical layer media. Inverse
As ATM becomes a ubiquitous technology, services
multiplexing, or the aggregation of the WAN links, is
based on ATM will enable companies to transport
performed via a cell-based control protocol.
different types of applications, at native speeds, over
IMA is expected to be widely accepted in both the
the same network. However, ATM is now used
customer and equipment vendor communities. Within
primarily for LAN backbones and deep within the
the carriers’ networks, IMA can be used instead of T1
carrier infrastructure; wide-area ATM services are
ATM for point-to-point trunking between frame
available only in limited areas. When they are
relay/ATM switches, greatly increasing bandwidth
available, they can either be prohibitively expensive (as
without upgrading to DS3 or OC3c lines. On the
in the case of 45 and 155 Mbps lines) or not provide
customer premises, the IMA specification promises

18 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 19

vendor interoperability, giving customers maximum Multi-Megabit Clear Channel

flexibility in the equipment selection process. However,

as with an new technology, it will take time for IMA V.35 or Inverse
.. T1
.. Inverse V.35 or

implementations to mature and for products to become

Router
HSSI MUX . Network
. MUX HSSI
Router

readily available. For those companies who would

A. T1 Inverse Multiplexing
prefer not to deploy such a new technology, bit-based
inverse multiplexing for ATM is a proven alternative. Multi-Megabit throughput for
ATM Cells over T3,
The IMA specification is very similar in function to SONET, etc.

TDM bit-based inverse multiplexing-both take a high- ATM V.35 or ATM ATM ATM V.35 or ATM
Switch HSSI DSU Network DSU Switch
HSSI
speed data stream and send it across multiple T1 or E1
lines, resulting in a fractional T3 or E3 channel. Both
B. ATM Inverse Multiplexing
support AutoFallback, which dynamically adjusts the
imuxed channel to a new, lower bandwidth, should one T1 Imuxing Compared to ATM Imuxing
or more T1/E1 circuits fail, thereby preventing network
downtime; both are tolerant of differential delay, which
To meet the needs of both its carrier and enterprise
is the relative delay introduced into the system due to
customers, Larscom has chosen to offer both IMA and
T1/E1 lines of varying lengths; and both are bandwidth
bit-based ATM inverse multiplexing solutions.
scaleable. When bit-based inverse multiplexing is used to
The primary goal of the IMA specification is to
transport ATM traffic, network performance is indistin-
facilitate multi-vendor interoperability, thereby allowing
guishable from cell-based imuxing.
customers and carriers with IMA devices from different
One of the inherent strengths of ATM is its ability
vendors to seamlessly interconnect their networks. The
to differentiate and prioritize traffic based on defined
Orion 2000 IMA Inverse Multiplexer is fully
“Qualities of Service”. Constant Bit Rate (CBR) traffic,
conformant with the ATM Forum’s IMA 1.0 specifica-
such as voice and video, is usually considered to be of
tion, which was finalized in October, 1997. A compact
highest priority, since any interruption to the cellstream
device that is suited to both CPE and Central Office
can result in noticeable loss of quality at the receiving
(CO) applications, the Orion 2000 takes an ATM cell-
end. The same is true for Variable Bit Rate- Real Time
stream from any ATM OC3c/STM-1 UNI or NNI
(VBR-rt) traffic. Other traffic types, such as Variable Bit
and imuxes it cell-by-cell across up to eight T1s. To take
Rate-Nonreal time (VBR-nrt), Available Bit Rate
advantage of ATM’s quality-of-service mechanisms for
(ABR), and Unassigned Bit Rate (UBR) traffic are
multiple types of traffic (voice, video, data), the Orion
usually comprised of data traffic that is less sensitive to
2000 supports multiple, user-configurable priority
time-of-delivery. Both IMA and bit-based inverse multi-
queues and queue thresholds, as well as Early Packet
plexing support traffic class prioritization, allowing
Discard (EPD) for up to 16 active connections.
companies to determine what types of traffic is most
important in their particular networks.

20 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 21

 Applications for IMA devices such as the Orion Larscom’s bit-based ATM inverse multiplexers
2000 range from simple, point-to-point connections provide an alternative to IMA solutions in areas where
such as switch-to-switch or router-to-router trunks, up only TDM-based service or infrastructure is available.
to sophisticated hub-and-spoke networks. In point-to- Larscom bit-based imuxes combine a standard ATM
point applications, IMA imuxes can be used to inter- DTE interface (specifically, a DS3 or OC3c ATM
connect an existing base of ATM LAN switches, UNI) with a traditional bit-by-bit T1 or E1 inverse
routers, or edge concentrators to an ATM-based multiplexer. Cells are converted to bits via the ATM
service or network infrastructure, without needing to Forum’s “Cell-based TC (Transmission Convergence)”
make any change at all to the installed base of ATM specification, which was finalized in 1995. The
devices. In a hub-and-spoke network, NSPs can deploy resulting product provides a multimegabit channel for
IMA inverse multiplexers as the CPE end of a cost- transporting ATM cells across a standard TDM
effective ATM or transparent LAN service, in areas network. The diagram below shows a bit-based ATM
where high speed ATM links are not available or not inverse multiplexing solution as implemented in the
affordable. IMA imuxes may also be used in applica- IMUX-A45 module for Larscom’s Orion 4000
tions where the NSP’s central ATM switch does not Broadband Access Multiplexer.
support IMA or switch slots are at a premium; in this ATM
Orion 4000 DS3
case, another Orion 2000 or other IMA inverse multi- UNI ATM
IMUX-A45
plexer may be connected directly to an available UNI Router
FT3
ATM
or NNI port on the switch, thereby utilizing the ATM DS3
Orion 4000

ports that the NSP already has.

ATM
Switch
UNI
IMUX-A45
... NxT1
TDM
ATM- IMUX-A45
... NxT1 Network
Equipped ATM
DS3 ATM
OC3c Orion 2000 TDM or ATM Orion 2000 OC3c Server UNI Orion 4000 DS3
UNI/NNI Network UNI/NNI NxT1 ... IMUX-A45
UNI ATM
Switch

Orion 2000-IMA Point-to-Point Configuration ATM Inverse Muxing

Larscom’s bit-based ATM inverse multiplexers
Orion 2000
have several advantages. First, they do not require
ATM compatibility across the WAN-the T1 or E1
Orion 2000
circuits are employed in the standard TDM bit mode,
TDM or ATM
ATM
Core
rather than a cell mode. This means that ATM traffic
Access Network
Orion 2000
Network can be transported across any existing TDM network,
ATM Switch
whether or not ATM service is offered by a carrier.
Orion 2000
with IMA ATM traffic can also then be combined with non-
ATM traffic across the same link’ for example, imuxed
ATM traffic may be transported across the same chan-
Orion 2000-IMA Spoke to a 3rd Party Hub
nelized T3 line as voice and other data traffic.

22 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 23

 Orion
4000/5
Inverse Multiplexing
T1s
.. T I A
M M 4 DS3 UNI ATM Implementations
Orion
4000
. U U 5
X X
Router

I
ATM DS3 UNI M A Orion The following are examples of customer implemen-
Router 4 4000
U 5
X T
I A DS3 UNI ATM
tations of Larscom’s Orion 4000 and Mega-T products.
3
I M 4 Switch They are typical deployments of inverse multiplexing
ATM DS3 UNI M A M T3 TDM U 5
4 T X
Switch U 5 U Network
T3 3 technology, providing flexibility, scalability, and cost
X X
M
T
M .. T1s effectiveness.
T1s
..
T
M U U . PBX
X
PBX . U
X
X

Case 1: Meeting the Requirement for Increased

IMUX-A45 ATM Inverse Multiplexer Bandwidth
The State of Ohio Bureau of Workers
In addition, because the bit-based imuxes use standard Compensation (BWC) had an extensive T1 network
ATM UNIs as their DTE interfaces, the imuxes can linking LAN backbones at 28 sites throughout the state
connect to any ATM device in the installed base that of Ohio and needed to implement a high-speed WAN
supports a DS3 or OC3c UNI. A company need not between three of their major network sites. Increasing
change its ATM switch or network architecture in traffic required the BWC to upgrade the links from their
order to deploy ATM imuxing-the bit-based imuxes Main Office to their two primary data centers from T1
simply plug into an available UNI port. A costly ATM to 4.5 Mbps.
switch is not even required for connection; because all The BWC knew Larscom’s reputation for providing
rate conversion and cell buffering is handled within the high-quality inverse multiplexing solutions, and asked
inverse multiplexer, a bit-based imux may be connected the company to assist in the design of this upgrade. It
to any ATM UNI, whether on a switch, router, ATM- was determined that the two upgraded links would be
equipped server or workstation, etc. Where ATM based on inverse muxing three T1s each. As shown in
switch slots are precious or unavailable, a bit-based the illustration on the following page, the BWC chose
ATM imux can truly “plug and play”. to install an Orion 4000 with two IMUX modules at the
Using bit-based ATM imuxing does not imply that Main Office, and to deploy Mega-T stand-alone inverse
a company will someday need to upgrade to IMA. multiplexers at the primary data centers, the
Both bit-based and cell-based ATM imuxing are Administrative Center (AC) and the Strategic
applicable in today’s WAN environment. The logical Operations Center (SOC).
upgrade path from either type of inverse multiplexer,
as it is with TDM inverse multiplexing, is to a higher
speed link-in ATM’s case, to ATM DS3 or OC3c; in
TDM’s case, to T3 or E3, or even high speed ATM.

26 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 27

 Case 2: Choosing a Growth Path
Orion 4000
Cisco A major midwest power company needed to inter-
7000
connect seven divisional offices in a ring-based network.
V.35 Serial
3T
1s Link
AC Each site featured an ATM LAN, a 155 Mbps
V.35 Serial Mega-T backbone, and a Cisco Lightstream 2020 ATM switch,
Links 3T
1s
1s as illustrated below. WAN links with a bandwidth of 6
T1 3T
Network V.35
Mbps were required for interconnecting voice, data,
Serial
Link
3T
1s and video applications at all of the divisional offices.
Cisco 7000 Cisco 7000 V.35
Single T1s
Mega-T 155 Mbps
to Orion 4000
Main Office SOC ATM LAN
Other Offices

Lightstream 2020
Ohio BWC Broadband WAN Links s)
12
M
T1 bp
(8 s(
s 8
Orion 4000 bp T1 Orion 4000
M s)
The Orion 4000 was deployed at the Main Office 12

for several reasons:

12

s)
T1
M
bp
V.35 V.35

8
s(
s
• As a shelf-based platform, Orion 4000 provides

bp
(8
T1

M
s)

12
plenty of capacity for adding additional inverse- Total
Lightstream 2020 of 7 Lightstream 2020
muxed links. Divisional
155 Mbps Offices in 155 Mbps
• Its inherent redundancy features help assure ATM LAN Network ATM LAN

network availability.
Power Company Network
• IMUX modules allow scaling of individual
channels to 12 Mbps. With T3 not an available option, inverse
multiplexing of T1s was the only viable solution. After
The choice of Mega-T for the data centers was careful consideration, the power company chose to
based on the fact that bandwidth requirements were install an Orion 4000 with IMUX module at each site,
not expected to exceed 6 Mbps in the near term. When as shown in the Figure. The choice of Orion 4000 was
bandwidth greater than 6 Mbps is required at one of based on several factors:
these sites, the Mega-T can be re-deployed to another • Product reputation and quality
site in the network.
The BWC now has the WAN bandwidth it needs • Flawless performance during trial testing
to efficiently handle the traffic between its Main Office • Smooth SNMP integration with Cisco routers and
and the two primary data centers. However, video switches
and data needs are beginning to strain the remaining
• A clear and logical migration strategy to greater
T1 links in the network. The BWC in the process of
speeds and ATM
deploying Mega-T at additional network sites and
increasing the sites’ WAN links to 3.0 or 4.5 Mbps. The last factor was very significant. While the
immediate need was for 6 Mbps links between the sites,

28 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 29

the power company wanted a growth path that would Stockholm. At each university, a Cisco 7000 router
allow the WAN links to be expanded incrementally, serves as the connecting point to the LAN. High-speed
with the possibility of eventually matching the 155 serial interfaces (HSSI) connect the routers to Orion
Mbps ATM LAN backbones. A year after deploying 6 4000 IMUX/E1 modules at 8 Mbps, and the
Mbps links, the power company upgraded to 12 Mbps IMUX/E1 modules provide 8 Mbps WAN access by
links by adding more T1 connections to the existing inverse multiplexing four E1 lines.
IMUX modules. They are now testing IMUX-A45 Inverse multiplexing let NORDUNet’s developers
modules, which support direct ATM connections to the take advantage of the growing availability and relatively
LAN switches. And the Orion 4000 platform provides a low cost of E1 circuits to scale their broadband WAN
clear growth path all the way to 155 Mbps, so the links to their immediate needs. And the Orion 4000
power company has the flexibility to upgrade according platforms have allowed NORDUNet to continue to
to its needs and the available WAN services. upgrade and improve the connections between the
universities and academic communities it serves.
Case 3: An International Solution
NORDUNet, a broadband network linking four Case 4: Interconnecting ATM Islands Worldwide
major Scandinavian universities, was the first network A leading global network service provider (NSP)
in the world to employ E1 inverse multiplexing. As employs Orion 4000’s ATM inverse multiplexing capa-
shown below, NORDUNet was deployed to provide 8 bility to offer its customers cost-effective worldwide
Mbps links between universities in Stockholm, Oslo, ATM service for voice, data, and video traffic. To
Helsinki, and Copenhagen, proving that inverse multi- provide this service, the NSP needed to solve two prob-
plexing is truly an international solution to the need for lems. The first is the problem of scaleable broadband
scaleable broadband WAN options. connectivity, for which inverse multiplexing offers an
Helsinki Stockholm
ideal solution. The second problem is to connect ATM
CISCO Orion 4000 Orion 4000
applications over long distances when high-speed ATM
7000 Fiber
Terminal
links are not available or prohibitively expensive.
Redundant
Larscom allowed the NSP to solve both problems
CISCO
E1s Fiber Links with a single product: the Orion 4000 equipped with
7000
E1 ATM IMUX modules. The ATM IMUXes take an
Network
E1s E1s
ATM cell stream from a standard ATM interface
(UNI), transport it bit-by-bit across inverse-multiplexed
Orion 4000 Orion 4000
CISCO
T1 or E1 circuits, and reconstitute the cell stream at
CISCO
7000 7000 the receiving end. Thus, ATM cell streams traverse
Oslo Copenhagen
non-ATM (i.e., TDM, or time-division multiplexed)
networks using standard T1 and E1 circuits. This
NORDUNET Broadband Network
allows the NSP to offer ATM connectivity virtually
The network has a “star” architecture. Stockholm anywhere on the globe, despite the fact that actual
connects directly to each of the other three universities, ATM WAN circuits are rarely available.
which can then communicate with each other via

30 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Technology Guide • 31

 The illustration below shows a typical international
ATM connection provided by the NSP.
Glossary of Terms
Orion 4000 Orion 4000
Inverse-Multiplexed E1s
802.3—An IEEE standard for the Ethernet LAN
protocol.
ATM UNI ATM UNI

AAL 5—See ATM.

ATM Switch ATM Switch
Adaptation LayerAccess Rate—The transmission
Customer’s Customer’s speed, in bits per second, of the physical access circuit
ATM Cell Stream ATM Cell Stream
New York London
between the end user and the network.
Alignment—The process of identifying the different
NSP’s Typical International ATM Connection
parts (framing, channels, etc.) of a synchronous serial
data signal; synchronization.
The client wants to link ATM “islands” in New York
and London, with a bandwidth of 4 Mbps between the Architecture—The manner or structure in which
two locations. The NSP employs Orion 4000s with ATM hardware or software is constructed. Defines how a
E1 IMUX modules to transport the customer’s ATM cell system or program is structured, how various compo-
stream between ATM switches in New York and London. nents and parts interact, and what protocols and inter-
The Orion 4000 platform allows the NSP to scale indi- faces are used for communication and cooperation
vidual links by adding E1 (or T1) circuits, and to deploy between system modules and components.
additional wide-area ATM links by installing additional
ATM—Asynchronous Transfer Mode. A cell-based,
ATM IMUX modules. This flexibility makes it easy for
fast-packet technology that provides a protocol for
the NSP to respond quickly and cost-effectively to the
transmitting voice and data over high-speed networks.
growing ATM WAN requirements of its customers.
ATM is a connection oriented technology used in both
LAN and WAN environments. It is asynchronous in
that the recurrence of cells depends on the required or
Conclusion instantaneous bit rate.
ATM Adaptation Layer (AAL)—The uppermost
Inverse multiplexing of T1 and E1 facilities is today layer of the ATM protocol. There are five AAL types for
a proven network technology that is recognized by different types of bit streams (constant vs. varial rates,
companies worldwide. It should be considered by anyone connection-oriented vs. connectionless, large packet size).
responsible for network design, implementation, and ATM Forum—An industry group which develops and
growth, particularly given today’s increasing demands for codifies standards for ATM.
higher bandwidth and flexibility. Inverse multiplexing
ATM Islands—Local implementations of ATM
serves as an excellent means of providing high-speed
equipment and applications which are not
connectivity between T1/E1 and T3/E3, as well as
interconnected by ATM wide-area services.
between traditional TDM and evolving ATM networks.

32 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Glossary • 33

Backbone—The top (highest-bandwidth) level in a Channelization—Division of a signal into multiple
hierarchical network, to which subnetworks and LANs data channels (or T3/E3 into T1/E1), which may or
are connected and over which they intercommunicate. may not follow time slot boundaries depending on
the application.
Bandwidth—The maximum amount of data that can
flow across a network segment. Bandwidth is generally Circuit Traffic—Communication between two
measured in kilobits (thousands of bits), Megabits network sites based on a temporary or permanent
(million of bits), or Gigabits (billions of bits) per second, circuit being established between the two sites. As
which are abbreviated kbps, Mbps, and Gbps, respec- opposed to cell traffic.
tively. Bandwidth should not be confused with
Clear Channel—A transmission link which transports
throughput, which is the amount of information actu-
a single traffic stream.
ally transmitted across a network segment.
Clear Channel ATM—Also CCA. A technology,
Bandwidth Gap—The large difference in bandwidth
using the ATM Forum’s Cell-Based TC specification,
between T1 and T3, or between E1 and E3; the lack of
for transmitting ATM cells bit-by-bit over standard
intermediate-rate carrier services.
TDM circuits, such as inverse-multipled T1s.
BER—Bit error rate. The number of logical bit errors
Connectivity—The ability of PCs and other network
divided by the number of bits transmitted, usually
equipment to interconnect and communicate.
represented as a negative exponent of 10 (e.g., 10 3, or
1 error in each 1,000 bits). Determining an absolute CPE—Customer Premises Equipment. Network equip-
Bit Error Rate requires bit by bit comparison of the ment (such as modems, multiplexers, CSUs, and
received signal with a known transmitted signal; routers) located at the customer’s site.
approximations can be achieved using CRC errors. CSU—Channel Service Unit. A device that interfaces
Cell Traffic—Communication between two network customer T1 (or other signal format) equipment to a
sites based on individually-addressed packets or cells of carrier facility of the same type. At its most basic level,
data which may be independently routed through the a CSU performs certain line conditioning and equaliza-
network based on the destination address or broadcast to tion functions, and responds to loopback commands
all network sites and accepted only by the addressed site. sent from the central office.
Broadband—Transmission at greater than 2 Mbps, DCE—Data Circuit-terminating Equipment. Devices
i.e., greater than the T1 or E1 rate. (such as modems, CSUs, packet switches, etc.) which
terminate circuits in a communication system.
CCA—See Clear Channel ATM.
Delay—Basically, the time the information takes to
Channel Service Unit (CSU)—A device that interfaces
transit a network or network segment. Differential delay
customer T1 (or E1) equipment to a carrier’s T1 (or E1)
is the difference in transit time between data taking sepa-
service. At its most basic level, a CSU performs certain line
rate transmission paths-for example, inverse-multiplexed
conditioning and equalization functions, and responds to
T1s employing different routes through T1 networks.
loopback commands sent from the central office. In an E1
environment, the term CSU does apply; terms such as DS1, DS3—See T1, T3.
PMU (Performance Monitoring Unit) are used.
34 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Glossary • 35
DS3 UNI—A particular ATM interface in which ATM popular way of providing variable bandwidth WAN
cells are mapped onto a standard DS3 signal. services. Based on the LAPD protocol of ISDN that
performs routing and transfer with less processing
DSU—Digital Service Unit. A device that interfaces
than X.25.
RS232, V.35, or other terminal interface protocols to
T1, T3, or other digital telephone signals. High-speed Data (HSD)—A type of data interface,
normally employing industry standard V.35 or EIA530
DSU/CSU—A device which combines the functions
electrical connections and typically used for speeds up
of a DSU and a CSU.
to 6 Mbps.
DTE—Data Terminal Equipment. Devices (such as
HSSI—High-speed Serial Interface. Standard for a
multiplexers, PBXs, front end processors, computers,
serial interface at high-speeds (up to 52 Mbps ) between
terminals, and LAN routers) which function as the
DTE and DCE equipment over very short distances.
source of a transmitted digital signal and the destina-
tion of a received digital signal. IMA—Inverse Multiplexing for ATM. The cell-based
inverse multiplexing standard currently under develop-
DXI—Data Exchange Interface. An interface protocol
ment by the ATM Forum.
for connecting routers to ATM DSUs.
Interexchange Carrier (IXC or IEC)—A long-
El, E3—The primary digital transmission systems
distance telephone company offering circuit-switched,
defined by the CCITT and used today in Europe, parts
leased-line or packet-switched service or some
of the Pacific Rim, and Latin America. Roughly analo-
combination.
gous to the North American T1 and T3 transmission
systems, E1 and E3 operate at 2.048 Mbps and 34.368 Internet—Capitalized, Internet refers to the special
Mbps, respectively. collection of hosts world wide that can communicate
with each other using the TCP/IP protocol suite and
Enterprise Network—A geographically dispersed
functions as a single, cooperative virtual network. Not
network under the auspices of one organization.
capitalized, internet is a generic term describing multiple
Ethernet—The most commonly-used LAN protocol, networks connected together, using a common protocol.
typically operating at 10 Mbps over twisted-pair wiring
Internetworking—General term used to refer to
referred to as 10Base-T. Newer iterations of Ethernet
the industry that has arisen around the problem of
use speeds of 100 Mbps and approximately 1 Gigabit
connecting networks together The term can refer to
as well as switching between LAN segments.
products, procedures, and technologies.
Fractional Services—Communications channels
Intranet—A private network that uses Internet soft-
which are subrates of standard digital service offerings
ware and standards.
such as T1, E1, T3, and E3 (designated FT1, FE1,
FT3, and FE3, respectively). Inverse Multiplexing—Also Inverse Muxing and
Imuxing. Combining the bandwidth of two or more
Frame Relay—A multiplexing data service based on
communications circuits into a single communication
fast transport of formatted information frames, without
channel.
flow control or error correction. Currently the most

36 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Glossary • 37

ISDN—Integrated Services Digital Network. An Orion 2000—Larscom’s IMA Inverse Multiplexer,
evolving set of standards for a digital network carrying fully compliant with the ATM Forum IMA 1.0 specifi-
both voice and data communications. cation, as finalized in October, 1997.
LAN—Local Area Network. A privately owned data Orion 4000—Larscom’s flagship product, a modular
communications system that provides reliable high- broadband access multiplexer platform with a unique
speed, switched connections between devices in a single WAN access architecture that handles both cell-based
building, campus, or complex. (ATM) and circuit-based (TDM) traffic.
Leased Line—A transmission line reserved by a OSI (Open Systems Interconnection) Reference
communications carrier for the private use of a customer. Model—Also OSI Stack. A seven-layer logical struc-
ture and architecture for network operation that defines
Legacy—A term denoting “holdover” equipment,
protocol standards by which compliant devices can
applications, structures, etc., from a an earlier network
exchange information. The reference model divides the
or other implementation which is to be integrated into
complex set of functions into seven manageable, self-
a current implementation.
contained and functional layers that are, from the
Load Sharing—The use by a router or bridge of lowest layer.
multiple parallel transmission circuits.
Layer 1, Physical Layer: Concerned with the mechan-
MAN—Metropolitan Area Network. A technology ical and electrical means by which devices are physically
evolved from local-area network designs that is connected and data is transmitted.
optimized for network distances over 50 km, speeds
Layer 2, Link Layer: Concerned with how to move data
more than 100 Mbps, and diverse forms of information
reliably across the physical data link.
(voice, data, image, video).
Layer 3, Network Layer: Provides the means to establish,
Mega-T—Larscom’s multiple T1 inverse multiplexer
maintain and terminate connections between systems;
provides a multi-megabit communication channel
concerned with switching and routing information.
equivalent to fractional T3 (FT3) channel.
Layer 4, Transport Layer: Concerned with end-to-end
Metaframe—A pattern of framing bits that allows for
data integrity and quality of service.
alignment of an inverse multiplexed signal.
Layer 5, Session Layer: Standardizes the task of setting
Multiplexer (Mux)—A device that combines input
up and terminating a session; coordinates interaction
from two or more terminals, computer ports or other
between end application processes.
multiplexers, and transmits the combined data stream
over one high-speed channel. At the receiving end, the Layer 6, Presentation Layer: Relates to the character set
high-speed channel is demultiplexed, either by another and data code that are used, and to the way data is
multiplexer or by software. displayed, on a screen or printer.
Multiplexing—Combining multiple information Layer 7, Application Layer: Concerned with the higher-
channels to create a higher-speed channel, typically to level functions that support the application or system
take advantage of bandwidth economy of scale activities.
inherent in most carrier services.

38 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Glossary • 39

PBX—Private Branch Exchange. A telephone T1, T3—Digital transmission systems with data capaci-
exchange serving an individual organization and having ties of 1.544 and 44.736 Mbps, respectively; also known
connections to a public telephone exchange. as Digital Signal Levels 1 and 3 (DS1 and DS3). T1 and
T3 are part of the digital telephone network hierarchy
Physical Layer—See OSI.
used in North America. A T1 incorporates 24 standard
QRSS—Quasi-Random Signal Source. An industry- voice channels (DS0s). A T3 incorporates 28 T1s.
standard test pattern employing a fixed bit sequence to
TDM—Time Division Multiplexing. The technique
simulate random data.
used in the currently-dominant circuit-based networks
Redundancy—A relationship of dual network to channelize transmission links by dedicating time slots
devices or facilities in which one functions as a backup within the data stream to the individual channels. For
for the other. example, the 24 DS0s that make up a T1 (DS1) are
Route Caching—A type of load sharing in a which created by dividing the data stream into eight bit time
an application is assigned to a particular one of several slots and dedicating specific time slots to specific DS0s.
parallel transmission circuits. Telnet—The TCP/IP standard protocol for remote
Router—A combination of hardware and software terminal connection service. TELNET allows a user at
that allows the connection of two or more subnetworks. one site to interact with a remote timesharing system at
another site as if the user’s terminal were connected
Scalability—The ability of a transmission technology directly to the remote machine.
to operate at a range of bandwidths.
UNI—User-to-Network Interface. Any of a class of
SDH—Synchronous Digital Hierarchy. The international interfaces applicable to user equipment connecting to
structure of optical transmission standards, corresponding ATM networks. As opposed to NNI (Network-to-
to the U.S. SONET standards. SDH starts at the 155 Network Interface).
Mbps level and increases in 155- Mbps multiples.
V.35—A commonly-used hardware interface for high-
SNMP—Simple Network Management Protocol. A de speed synchronous serial interfaces.
facto network management standard established by LAN
vendors. This standard has widespread acceptance in the WAN—Wide-area Network. A network covering a large
U.S., and growing acceptance abroad. geographic area, generally employing carrier-owned
facilities (e.g., long-haul telephone lines) between sites.
SONET—Synchronous Optical Network. A standard for
the multiplexing and transmission of broadband informa-
tion in optical form, so that low-speed components of the
optical channel can be easily extracted. Operates from
51.84 Mbps (OC-1), 155 Mbps (OC-3), to 622 Mbps
(OC-12) and beyond.
Synchronization—Locking onto a framing pattern in
a synchronous digital signal so that the signal can be
properly interpreted or processed.

40 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Glossary • 41

 NOTES NOTES

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

42 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Notes • 43

 NOTES NOTES

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

44 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Notes • 45

 NOTES NOTES

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

__________________________________________ __________________________________________

46 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN Notes • 47

 NOTES

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

__________________________________________

48 • Inverse Multiplexing—Scalable Bandwidth Solutions for the WAN

 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. Larscom 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.
Version 4, 1998