Must-read. The ride of the century. Sonet stays firmly ensconced even as it prepares for a next generation optical networking adventure
internettelephony.com
Cover Story, August 24, 1998
SUSAN BIAGI
Is the Sonet cycle ending? Has it made the slow climb to the apex of the Ferris
wheel of optical networking?
Vendors and service providers are seeking alternatives to the technology that has
defined reliability for optical networking. It appears that Sonet might be on a slow
decline, but it won't relinquish its seat without a replacement. Carriers still need its
functionality, and Sonet's ubiquity secures its position in the network for several
years.
Change is assured, however. The all-optical buzz is gaining momentum. Carriers
want their networks to be bigger, faster, better--demands that have brought about
wavelength division multiplexing.
Many carriers are banking on dense WDM and asynchronous transfer mode for
the future, and these technologies are sure to change networks. As vendors
develop products designed for DWDM and ATM, Sonet's role within the network
will change, too.
CLECs and vendors are discussing the merits and pitfalls of all-optical networking.
But is it feasible? Some analysts predict that all-optical networks are at least a
decade away. Yet the tradeoffs could be too great. One key reason is Sonet--and
carriers' reliance upon it.
Why do public networks depend so heavily on Sonet? Sonet grooms and routes
traffic, provides performance monitoring and, perhaps most important, handles
restoration. That function alone is the network equivalent of job security.
Bellcore developed the Sonet specification about a decade ago. Transport gear
manufacturers embraced the specification because it enabled products to
internetwork and interoperate, thus commoditizing the products and cutting costs.
Sonet leveled the playing field for vendors, allowing them to offer products with
similar functionality at lower prices.
The Sonet specification defines a frame format, a physical interface, optical carrier
line rates, and an operations, administration, maintenance and provisioning
protocol. Sonet also introduced ring operations, says Joe Savage, vice president of
Ryan Hankin Kent.
"Instead of having all point-to-point circuits, now you have the ability to route
working traffic in one direction and protection traffic in the other ring," he says.
Traffic can travel through different facilities and routes to reach the same
destination, enhancing the Sonet network's overall reliability. This was the advent of
the self-healing network. If a line is cut, Sonet can reroute traffic so it reaches its
destination.
Sonet dims in light of data
This scheme has been perfected for voice. But today's networks are shifting toward
data. Studies predict that by the turn of the century, data will comprise 40% to
90% of the traffic flowing over glass. Under this new network model, Sonet begins
to lose its luster.
Data is bursty in nature, and Sonet is an inefficient means of transporting data.
Private data networks use a time division multiplexing (TDM) scheme, while Sonet
uses optical carrier rates. When data traffic is placed on the Sonet ring, the speeds
don't match up, resulting in wasted bandwidth. A typical 10 Mb/s LAN would
require an expensive T-3 bandwidth pipe, nearly 35 Mb/s of which would be
unused.
Adding to the mix is ATM. Heavily relied upon to ensure quality of service in data
networks, ATM brings another set of complications to the carrier network. Most
important, ATM can only be transmitted on Sonet using virtual paths, or virtual
circuits, and Sonet network management systems cannot manage those circuits.
To combat this, carriers are looking at pure optical networking and re-evaluating
Sonet as a transport mechanism.
Figure 1 depicts the transport infrastructure from the customer premises to the
interexchange carrier. Today, most networks have special service overlays to
handle fiber distributed data interface (FDDI), 10BaseT, gigabit Ethernet and other
services. This overlay is separate from the main network and can hog bandwidth
unnecessarily. On the access side, incoming signals are fed into the access rings.
Those rings connect to the interoffice facility via a TDM 3/1 cross-connect. The
signal is then forwarded through a 3/3 TDM cross-connect to the point-to-point
WDM network.
First metro DWDM access rings will be deployed, says Solomon Wong, assistant
vice president of marketing at Cambrian Systems. Next, WDM will come into the
interoffice ring, and finally WDM will reach the interexchange carrier. "Then it's
becoming networking instead of point-to-point connections," he says. When that
occurs, Sonet will be out of everything beyond the access ring.
The Sonet market grew from about $3.1 billion in 1996 to $3.7 billion in 1997,
says Mathew Steinberg, director of optical networking at Ryan Hankin Kent.
Growth comes from second tier interexchange carriers, competitive LECs and
utility companies, and is driven by competition for business services and on-line
data services.
Change will come from the edge and work its way back into the network. New
companies are developing non-Sonet optical metro devices, but acceptance could
be slow.
"If we keep things in pure light, total optical networks, we don't have to look at
Sonet," predicts Michael Vent, executive vice president of network engineering and
operations at IXC Communications. The all-optical network will emerge at the
edge in distributed access and metropolitan networks. Several companies are
banking on that model.
"The optical layer doesn't have to be fragmented into the Sonet format," Vent says.
"Companies like Ciena, Cambridge and Cambrian are saying, 'All you need is a
light pump and a frequency.' You don't need to format [the payload] into Sonet. As
light becomes more native and switchable and you can [cross-connect] it, Sonet
loses its advantage." Transition is the issue, he says, and he wonders if the industry
"is ready to make the transition [to non-Sonet networking]."
Cambrian thinks so. The data crunch will force the issue, says Wong. Sonet is
ideally suited for DS-1 (1.54 Mb/s) and DS-3 (45 Mb/s) transport, which it does
cost effectively. ATM maps well, too. But gigabit Ethernet, Internet protocol (IP)
or an enterprise systems connection (ESCON) is problematic.
To wrap a Sonet signal around a DS-3 means that 86% of the envelope is used for
the payload. The other 14% is used for overhead. Carriers don't complain,
however, because the 14% Sonet consumes is worth the tradeoff.
Other solutions center around increasing payload use. Companies such as Omnia
Communications and Atmosphere Networks are working to cut the Sonet
container from 1.5 Mb/s to 53 bytes, the size of an ATM cell (see sidebar).
In addition, introducing new interface points can push the all-optical network
forward. DWDM systems need an interface point, usually a digital cross-connect,
but optical cross-connects are generating interest.
"Digital cross-connects as we know them today aren't going to be around in the
future." Wong says. "As the network evolves, digital cross-connects will become a
hybrid of a digital cross-connect with optical cross-connect capabilities." To work,
the optical cross-connect has got to be bit-rate-protocol independent and
bit-rate-protocol dependent, he says. Without that flexibility, the device won't be
able to handle both ESCON and TDM-mapped signals.
The evolution of the network is illustrated in Figure 2. All access methods are fed
into the access ring, which also supports ESCON, OC-3, OC-12, gigabit Ethernet
and FDDI. The metro access ring links to the interoffice facilities through a
TDM/ATM optical cross-connect. The signal stays in the optical realm, moving
through an optical cross-connect to the carrier's DWDM point-to-point ring.
Terabit switches, ATM switches, IP routers and new digital loop carriers can also
provide Sonet interfaces. If these are adopted in large-scale fashion, stand-alone
Sonet devices will fade away. [Who is the best supplier of these 4 products now and projected future? To my knowledge, ATM switches is ASND/Cascade and digital loop carriers is AFCI. What about the other 2 technologies?]
"Sonet was initially blurry on the edges," Wong says. "Now with ATM and
DWDM, it is coming into focus. Carriers are focusing in on where Sonet has
value."
One place where Sonet still has value with DWDM is in electrical/optical
conversion. "WDM will never eliminate Sonet," Wong says. "You will never take
electrical signals and carry that on a wavelength. WDM is the lowest level common
denominator. It still needs Sonet for DS-1 and DS-3, for electrical signals."
However, Cambrian's OPTera DWDM device seeks to avoid conversion as often
as possible. Keeping the transport optical will open up service opportunities for
carriers, Wong says. DWDM doesn't care about the traffic type. It transports data,
voice, gigabit Ethernet and ESCON traffic.
"WDM is bit-rate-protocol independent, which gets rid of that hurdle," Wong says.
"The theme is that there are new alternatives [to Sonet] and new possibilities. We
are not trying to eliminate Sonet. But [technology] that wasn't in focus before is
now coming into play."
Challenges ahead
To decrease reliance on Sonet in the optical network, DWDM must provide ring
reliability equal to Sonet, says Wong. "At a minimum, DWDM must have the same
survivability as Sonet. If it doesn't, service providers won't maintain their service
state. It has to be a real ring with real ring protection switching."
Savage points out that Sonet's main functions will have to be incorporated in other
network devices before Sonet can exit the scene. He cites five areas to consider:
link-based performance monitoring, fully redundant routing capabilities,
management, provisioning and multiplexing, and operations support systems (OSS)
compatibility. When those items are incorporated elsewhere in the network, Sonet
will give way to the next generation of networking.
The challenge is that DWDM has to have the same survivability characteristics as
Sonet on each wavelength, not on all wavelengths at once. Sonet is cumbersome in
a DWDM environment, says Dan Taylor, managing director of telecommunications
at The Aberdeen Group. DWDM creates multiple virtual rings, increasing the
amount of overhead. "Even on WDM, you need some way to format the traffic.
WDM is using Sonet as an overlay," he says.
Some of the obvious inhibitors to non-Sonet networking aren't related to
technology at all. Making the shift requires a huge commitment from manufacturers
and carriers. Manufacturers are set up to build Sonet equipment, notes IXC's Vent.
"They have to retool their manufacturing [operations] to transition," he says.
Most vendors and carriers agree that Sonet is here to stay for the next several
years. The current (and continuing) investments in Sonet equipment will keep it
within the network structure for at least 10 or 15 years. As non-Sonet equipment is
being developed and tested--before it is proved viable, that is--carriers will choose
tried-and-true Sonet. And the cycle continues.
Carriers will continue investing in Sonet equipment, pushing back the date
non-Sonet equipment takes hold in the market.
"Sonet is just like any standard," Vent says. "It's hard to change when all the
equipment, education and empirical knowledge is based around it. The whole
industry is looking for better ways to get throughput. They are experimenting. My
concern is that we go away from Sonet into an ambiguous standard or no standard.
If Sonet goes away, hopefully, we all can push to a new, improved standard, not an
arbitrary one."
Contact Susan Biagi.
Data's hostile takeover
The industry is rumbling about moving to the coveted all-optical network. Vendors
claim it will be more efficient than today's electro-optical networks--especially as
data traffic usurps voice on the public switched network. The challenge, then, is to
create a more efficient traffic-agnostic network that upholds the telecom industry's
99.999% reliability standards.
That's no small task. And it's made tougher because public and private networks
operate at different speeds and use different technologies. Data packets travel over
Ethernet (10 or 100 Mb/s) or token ring (4 or 16 Mb/s). To transport a 10 Mb/s
LAN on the public network, carriers have to connect those time division
multiplexing topologies to the optical network. A DS-1 (at 1.54 Mb/s) is not
enough bandwidth, and a DS-3 (at 45 Mb/s) provides way too much bandwidth.
"The last bastion of TDM in the wired network has been Sonet," says Mike
Champa, president and CEO of Omnia Communications, Marlboro, Mass. "That is
going away, and it's being replaced with virtual path technology. If TDM is not
appropriate for T-1, it's certainly not appropriate for OC-3, OC-12 or OC-48
links."
The solution appears to be ATM. "With ATM, you get a finer granularity of
bandwidth," says Alex Dobrushin, marketing vice president at Atmosphere
Networks, Cupertino, Calif. Wavelength division multiplexing can be added on top
of the ATM network to boost bandwidth. Instead of the traditional circuit-based
Sonet virtual tributary, ATM cells will be transmitted using Sonet virtual paths. The
net result is a more adaptable network with more bandwidth capacity on the
existing fiber. With that flexibility, carriers can add and vary services.
Atmosphere and Omnia are both developing access ring devices for this purpose.
"Today, carriers are very limited in terms of the services they can provide
[because] of the transmission structure," Dobrushin says. "We are removing that
constraint. We allow the transmission infrastructure to be optimized for all the new
[data] transmission technologies. We preserve the infrastructure, and give [carriers
the capability] to create new services at any speed that a customer may want."
Atmosphere's Cirrus Full Service Node 1200 uses a "thin layer of ATM," a
stripped-down version of today's ATM technology, which is transported on top of
existing fiber and copper infrastructures. Actually, Dobrushin acknowledges, the
thin layer technology is essentially ATM in its original form, before higher-level
service protocols are added. Atmosphere's version merges transmission, access
and termination.
In addition, Atmosphere is developing the distributed bandwidth management
protocol, which allows all the nodes on a Sonet ring to communicate as traffic
enters the ring. It also provides add/drop multiplexing functionality (see figure).
"Our Sonet ring behaves like a single switched entity," Dobrushin says. "We have
to match the traditional characteristics of Sonet [such as through latency and ring
healing]. We meet all those parameters."
Omnia is taking a similar approach. "We are building a next generation ADM that
will map ATM onto Sonet rings in place of existing TDM technology and improve
access technology to provide high-speed data interfaces as well as voice
interfaces," Champa says.
Omnia is eliminating the TDM component of the network, making the transport
technology on the access side consistent with the backbone. "The Sonet
environment is still there, [but it's] relatively small. It still has the performance
monitoring and control channels you need," says Chief Technology Officer Jeffrey
Weiss. It also eliminates the need for a digital cross-connect.
Atmosphere's FSN 1200 is priced at $16,000, and will be available this summer.
Omnia's product is under development, and will be announced in
September.--Susan Biagi
RETURN TO STORY
RETURN TO TOP
From our Archives
July 27, 1998 Switching & Transmission
New room for growth
Access networks expected to provide billions of dollars for WDM vendors
WAYNE CARTER
July 6, 1998
The future is now
Supercomm brings out the DWDM in vendors
SUSAN BIAGI
June 22, 1998 Switching & Transmission
A kinder, gentler network
Ascend and Positron aim to simplify networks by converging functionality
SUSAN BIAGI
June 8, 1998
If you build it,they will come
With the potential capacity of fiber reaching new highs, carriers need
solutions to help them manage these meganetworks
BILL GARTNER and BRUCE NELSON
May 11, 1998
Bandwidth for tomorrow
The next wave of data drives carriers toward ultradense WDM systems
WAYNE CARTER
April 27, 1998 InFocus
WDM is the answer
Recent developments in wave division multiplexing technologies and the
availability of low-cost overlay transmitters will make possible low-cost,
highly flexible broadband networks
JASON SHREERAM and DON SIPES
April 27, 1998 A.M. Report
DIRECT CONNECTION
Vendors link switching, routing equipment directly to WDM systems
WAYNE CARTER
April 20, 1998 Switching & Transmission
New frontier for DWDM
Some RHCs hot on DWDM for their networks, others more wary
WAYNE CARTER
March 23, 1998 Undercurrents
Networks: The next generation
SANDRA GUY
December 8, 1997 Perspectives
When IP met Sonet
STEVEN TITCH
September 15, 1997
Toward an optical layer
Photonic switching will be key to developing an all-optical network
STEVEN TITCH
September 15, 1997
United they stand
Industry groups are making progress in developing an all-optical network
WAYNE CARTER
RETURN TO TOP
Additional resources on the Net
SONET INTEROPERABILITY FORUM (SIF)
Optical Networks tutorial
An IEC Web ProForum sponsored by Alcatel
A Sonet tutorial
An IEC Web ProForum sponsored by Tektronix
A DWDM tutorial
An IEC Web ProForum sponsored by Lucent Technologies
RETURN TO TOP
Any Comments?
Send them to Karen Murphy at msblues@earthlink.net.
www.internettelephony.com
Telephony August 24
c1998 Primedia Intertec
All Rights Reserved.
|
