Sonet speed limits. If Sonet is to survive next-generation demand, it will need a little help from some new friends.
americasnetwork.com
Annie Lindstrom, May 15, 1998
There's a tidal wave of data heading for the public switched telephone
network (PSTN), and carriers are looking for ways to keep their heads above
water or actually start to swim when it hits.
For many years, carriers' synchronous optical network (Sonet) rings have done
a fine job of reliably transmitting tremendous volumes of voice traffic and a
growing amount of data traffic. However, the Sonet optical interface standard
was not designed with data in mind, so the growing amount of data traffic calls
for some adjustments in Sonet's role as a transmission technology.
The problem is that bandwidth is wasted when carriers try to fit today's data
services into Sonet pipes. That's because Sonet gear uses time division
multiplexing (TDM) to originate and terminate optical signals.
In the TDM world, transmission rates are based on multiples of 64 kbps, or
one voice channel. Once traffic is multiplexed onto the Sonet backbone, optical
signals are transmitted at rates that are equal to multiples of 51.84 Mbps, or
Optical Carrier 1 (OC-1); or up to 10 Gbps, or OC-192. An OC-1 carries
the equivalent of 28 1.5 Mbps T1 links, or one 45 Mbps T3 link.
Unfortunately, more and more of the carriers' customers want only the
bandwidth they need-not the chunks of bandwidth solidified by some standards
body many years ago.
In addition to defining the speeds at which optical signals are transmitted, the
Sonet standard defines a frame format and operations protocol that enable
carriers to monitor and control the flow of traffic across a Sonet ring.
TDM lets carriers transmit multiple parties' signals across the same fiber by
interleaving the disparate signals onto a single wavelength for transmission to
their destination, where they are demultiplexed off the Sonet backbone and
delivered to the intended recipient. Bandwidth to be transported across the
Sonet ring is assigned to each customer in T1, T3 or OC-n-sized blocks
according to specific time slots. Those chunks of bandwidth are set aside for
each customer, whether they are used or not.
Many of today's data services do not conform to the TDM/Sonet rate
structure. For example, a 10 Mbps transparent local area network (LAN)
service would require a T3-sized chunk of bandwidth to traverse a Sonet ring,
wasting 35 Mbps of bandwidth.
Nevertheless, Sonet has the momentum from years of deployment and a
well-deserved reputation as a reliable transmission technology. Both
interexchange carriers (IXCs) and local exchange carriers (LECs) have
deployed and are continuing to deploy Sonet rings in their networks, and they
are not about to dig them up and replace them. However, it is more than likely
that these carriers eventually will enhance their Sonet infrastructures in
significant ways so they can surf atop the approaching data wave, rather than
drown in it.
With help from two emerging technologies, Sonet's role will change, but it will
survive for years to come in long distance networks and local access and
interoffice networks.
Abracadabra
Dense wave division multiplexing (DWDM), which has been in the IXC
domain for nearly a decade, has proven its ability to increase the capacity of
optical fiber dramatically. DWDM systems allow a single fiber to carry multiple
wavelengths-from 16 to 40 in commercial networks today, rather than just one
or two wavelengths.
DWDM systems work in tandem with the existing Sonet infrastructure.
This means that carriers can expand the capacity of their Sonet networks while
maintaining the reliability provided by them without having to replace or
upgrade their Sonet transmission equipment or add/drop multiplexers (ADMs).
More channels mean more bandwidth, and that's good news.
Although DWDM is used primarily by IXCs, other carriers [competitive LECs
(CLECs) and incumbent LECs (ILECs)] are beginning or planning to use it to
accommodate growing demand for bandwidth in light of fiber exhaust
situations. Bell Atlantic, for example, plans to begin installing 16- and
24-channel DWDM systems in its interoffice network. LECs using DWDM to
solve capacity problems in their interoffice networks first leaves their access
networks still in need of an efficiency overhaul.
Virtual reality
While DWDM gives carriers more pipes to fill, it does not help carriers use
those pipes more efficiently. Carriers can opt to place traffic directly onto
DWDM and bypass the Sonet layer altogether, but doing so can be risky.
Although some DWDM vendors offer Sonet-like protection in their gear, most
LECs will be wary of 'Sonet-like' until it's proven as trustworthy as Sonet.
However, all is not lost in the meantime.
About four years ago, Bellcore (Morristown, N.J.) introduced the idea of
statistically multiplexing asynchronous transfer mode (ATM) traffic onto virtual
paths (VPs) that travel across a Sonet ring. Rather than assigning bandwidth in
64 kbps chunks, ATM VP allows carriers to fill their Sonet pipes with actual
traffic on an as-needed basis, rather than with empty time slots. Customers are
given a guaranteed rate and a peak rate, which they can obtain when the
bandwidth is available. The network delivers and manages bandwidth on the
ring according to the quality of service the customer has specified.
"ATM was initially viewed as a transport and switching capability, but early on
the focus was on building ATM switches," says Bill Rubin, Sonet/Synchronous
Digital Hierarchy (SDH) project manager for Bellcore. "Virtual paths are used
in ATM and rings are used in Sonet, so it was just a matter of realizing that you
could use the concepts of ATM VP allocation of bandwidth on a Sonet ring."
Bellcore's GR-2837, refined two months ago, defines three ways to place
ATM traffic onto Sonet rings, according to Chad Dunn, product manager for
Omnia Communications (Marlboro, Mass.), a company that is developing an
ATM-over-Sonet product for metropolitan-area Sonet rings:
The first way is to channelize the ring, and take one or more tributaries
and assign them to carry traffic that is multiplexed at the ATM level while
the others carry traffic multiplexed at the Sonet level.
The second way is to channelize the ring and concatenate the channels
that will carry ATM. A carrier could channelize an OC-12 ring and use
an OC-3c on it for ATM traffic, and let the rest of the ring carry
traditional synchronous transfer mode (STM) traffic, according to Dunn.
The third way, which Omnia finds "very attractive, is to channelize the
entire ring and concatenate all of the channels, so you have a full OC-12
worth of bandwidth to play with. Any channelization that takes place is
done on the virtual circuit level," Dunn says. "That's a much more
efficient way to divide up the ring, because those pipes are a lot more
flexible. You get the benefit of statistically muxing your entire traffic
payload over ATM."
With ATM VP, carriers don't waste bandwidth, because bandwidth units can
be defined according to their customers' needs, Rubin says. ATM VP also
allows carriers to accommodate asymmetrically apportioned bandwidth on
their Sonet rings; this feature will be important to carriers that offer
asymmetrical digital subscriber line (ADSL) service, which delivers more
bandwidth downstream than it does upstream.
Other players in this space include ADC Telecommunications (Minneapolis)
and startup company Atmosphere Networks (Cupertino, Calif.). ADC's
Cellworx Service Transport Node is available in limited quantity and will be
generally available in the July/August time frame, according to Karl Rookstool,
vice president of marketing for the ATM transport division of ADC's
Broadband Business Group. In addition to operating as an ADM on the Sonet
ring, the node also can function as a standalone integrated access multiplexer
extended off of a ring at OC-3c or DS3 rates, he adds.
This service access multiplexer (SAM) function, defined in Bellcore GR-2842,
allows carriers to terminate a variety of different service interfaces to
customers, such as frame relay, native LAN, private branch exchange (PBX)
and T1, Rubin says.
Atmosphere's Cirrus Full Service Node (FSN) and Cirrus Network
Termination Unit (NTU) will be commercially available this quarter, according
to Alex Dobrushin, vice president of marketing for Atmosphere. The FSN can
act as a ring multiplexer, linear add/drop multiplexer or a SAM. The NTU
works as a SAM. Cirrus FSN for an OC-12 ring is priced at $16,000, while
the NTU costs $5,000.
"Essentially, you get a much higher performance box for the same price you'd
pay for a conventional OC-3 Sonet ADM," Dobrushin says. "Our function is to
concentrate lower speed traffic and deliver it to gigabit backbones. It's a more
efficient access layer to the optical network."
Although products are just becoming available, ILECs and CLECs are
showing an interest in ATM VP.
"We think ILECs will be our biggest customers and IXCs will become
interested in it as they move into the local market," Rookstool says. "For
CLECs, ATM VP is an easy decision, because it's the most cost-effective way
to handle multiple services."
"What I'd like to do is layer ATM VP on DWDM. That would create a perfect world," says Mike Viren, senior vice president of regulatory and strategic planning for Intermedia Communications Inc. (Tampa).
The idea is not far-fetched. Atmosphere's products can be linked to any device
that has a Sonet-compatible optical interface, Dobrushin says, adding that most
DWDM systems indeed have such interfaces. "Our solution is complementary
to offerings from companies like Cambrian [Kanata, Ontario] or Ciena
[Baltimore], which are focused on ring-based DWDM systems," Dobrushin
says. "Those systems only provide optical interfaces, but the world at large is
not optical. Business comes into the world on copper or coaxial cable, and all
that traffic needs to be converted from a wire-based environment to an optical
environment and concentrated before it goes into a DWDM system."
At least one ILEC has proclaimed its interest in ATM VP. LoraineBeyer,
manager of infrastructure planning and strategies for BellSouth (Atlanta),
showed that the technology is on the company's drawing boards during a
presentation at the International Engineering Consortium's Broadband Access
Communications Forum.
The catch
Even though ATM VP leverages Sonet's reliability, there is an Achille's Heel, of
sorts. Atmosphere's products, for example, cannot be managed by Sonet
network management systems. They must be monitored by simple network
management protocol (SNMP)-capable systems, such as those carriers use to
monitor their ATM networks.
"Those SNMP interfaces will have to have some kind of mediation device that
will enable them to speak to embedded Sonet management systems," Rubin
says.
Standard SNMP management platforms such as HP OpenView provide
uplinks from the ATM VP products to carrier operations support systems
(OSSs), Dobrushin notes.
Although Sonet will remain a staple of the public network for some years to come, it will have to co-exist with ATM VP and DWDM.
"There's no question that Sonet is going to be around for the long haul. The
industry will retain the good elements of Sonet, such as self healing. But
because there is more data coming into the network than there has been
historically, the multiplexing hierarchy has to change,"Dobrushin says. "That's
why ATM VP rings are so attractive to carriers now. It allows them to deliver
a larger portfolio of competitive services using their existing infrastructures. It is
part of the grand scheme of the 21st century optical network deployment."
LECs will deploy DWDM in their interoffice networks. They will deploy ATM
VP in their access networks, where DWDM may be more cost-prohibitive,
initially, and remain that way if ATM VP is a success. The two may be used in
tandem, but if doing so proves to be to difficult to manage, carriers may opt for
DWDM alone in their access networks. ATM VP may be used in the
interoffice network where there is a need for more efficient bandwidth use,
Rubin notes.
The emergence of DWDM may put the brakes on the Sonet speed limit at
OC-192. There's a good chance that OC-768 will never see the light of day.
That's because it will be easier for carriers to multiplex wavelengths from
OC-192 onto DWDM than it will be to build a higher speed Sonet multiplexer,
Dobrushin notes.
"Right now, ATM VP, DWDM and Sonet are complementary, but five years from now, we will start to see network simplification, and there may not be separate Sonet network elements," Rubin says. Companies like Qwest [Denver] that deal with very high-capacity networks will be looking to put their ATM or IP traffic directly over DWDM."
May 15, 1998 table of contents
Copyright 1998 Advanstar Communications. Please send any technical comments or
questions to the America's Network webmaster.
|
