Separate and incompatible transmission protocols SDH and SONET dominate different regions of the world. However, a form of integration is at hand with the development of multi-protocol label switching and its recent 'generalised' offshoot.
By Antony Savvas
The wider world of networking protocols often sees the latest "new kid on the block" standard being proffered as the solution to everyone's problems.
But when it comes to optical transport protocols some industry watchers could be left with the distinct impression that there haven't been many new kids since the last graduation.
A few years ago, ATM was being billed as a networking solution that no carrier or big business could do without. Then along came Gigabit Ethernet and every business was asked to consider working with that too. Now we have the appearance of 10 Gigabit Ethernet and carriers and firms are being encouraged to adopt this solution in local area, wide area and metropolitan networks.At the optical transport layer though there hasn't been much change since SONET was adopted by the US, Canada and Japan, and SDH started operating around Europe at the end of the 1980s.
All change
This state of affairs could now change with the development of some new solutions that give the tried and trusted SONET and SDH a much-needed facelift to cope with QoS (quality of service) and SLA (service level agreement) issues.
Some may be tempted to ask why there isn't one standard, which would make it easier for networking suppliers for a start, but such an obvious beneficial development has not been the priority for the various standards bodies and carriers alike.
With broadband access increasing and multimedia and bandwidth-hungry applications becoming standard in the home and office, the carriers need something to enhance either SONET or SDH as soon as possible to be able to take advantage of niche data markets like video conferencing, webcasts, and broadband TV.
But for those still looking towards a single optical standard, perhaps we should address the issue, even though others haven't quite accepted that with networking, the solution is usually developed by making a "patchwork quilt" through mixing and matching protocols.Steve Harbour, chief executive officer at Native Networks, says: "SDH has 200,000 rings deployed and SONET 100,000, with a combined expenditure since 1985 of over USD300bn.
"There is no incentive to re-engineer this because there would be little real benefit to anyone, but there is every incentive to harmonise what flows over it, especially considering next-generation data and integrated voice and data solutions."
Integrated protocols
What Harbour and many others are waiting for are protocols that can be integrated with SONET and SDH to make it easier and more efficient to manage niche data. This is where protocols such as MPLS (multi-protocol label switching) and GMPLS (generalised multi-protocol label switching) come in.
The fact that existing SONET and SDH solutions still have the confidence of the carriers is illustrated by Deutsche Telekom's decision to expand its optical network backbone down the eastern side of the US, using SONET as a multi-million dollar solution to connect its worldwide customers.
What Deutsche Telekom will now be considering is how it can improve its overall performance and profit margins with increased quality of service (QoS) and tighter service level agreements (SLAs) featuring in customer contracts.
As Harbour succinctly puts it, MPLS and GMPLS provide the "intelligent glue" to cost effectively map packet traffic onto an SDH or SONET bearer.
"MPLS allows you to put any traffic on SDH or SONET in a highly scalable way. Our own solution for instance enables the equivalent of a 50Gbit/s payload being put onto a 2.5Gbit/s fibre and still delivering on SLAs for hundreds of MPLS tunnels of specific bandwidth."
Roberto Sabella, research manager at Ericsson Lab Italy (see panel on p.22), expands on the theme of getting more out of existing SONET and SDH rings. He says: "Retrospectively, it would probably have made sense for there to have been a common synchronous network standard from the beginning.
"But SDH and SONET will have less importance in the future as intelligent GMPLS over DWDM will offer a universal transport network management solution."
Sabella says: "As telecom networks evolve into a layered architecture - with separate connectivity, control and service/application layers - the choice of transmission technology becomes less critical to delivery and management of services."This layered approach permits the creation of a converged connectivity layer, where all services can potentially be carried over one unified infrastructure. In optical networking, convergence is already happening and will be further enhanced by the general shift to IP."
Sabella explains that this shift will give rise to an intelligent optical network that will not only deliver the high bandwidths required by the expected growth of IP-based multimedia services, but will also offer the necessary management capabilities to allocate bandwidth flexibly where and when required.
The rise of GMPLS
This new network infrastructure will provide fast switching of aggregated IP flows and efficient traffic engineering. This is something that the Internet Engineering Task Force (IETF) standards body is promoting as it pushes GMPLS as a core solution for delivering a distributed IP control plane for all switching layers including the optical transport layer.
GMPLS does not restrict the way that the layers work together, extending the MPLS concept to multiple switching layers.
The potential of GMPLS is supported by Urband, which is building an extensive MAN in London using the city's sewer system, among other projects.
Stephen Chivers, Urband chief technical officer, says: "GMPLS extends the use of MPLS from purely packet-based networks to time division systems such as SDH, SONET, and Plesiochronous Digital Hierarchy (PDH) networks running at different rates.
"Urband will be watching developments closely as we view GMPLS as the preferred method for obtaining a degree of integration between SDH and SONET."
Optical transport layer
Regarding the development of such standards, Gzim Ocakoglu, EMEA technical marketing director at Centrepoint, says there are other standards to consider.
For instance there is the emerging ITU-T G.709 digital wrapper standard which, like GMPLS, is also designed to improve management control in the optical transport layer.
Ocakoglu says G.709 is an extension to the SDH layer that allows operators to enjoy easier performance monitoring, and bit error rate (BER) threshold-based protection.
There is also quality of service protocol ITU-T G.ASON, which takes its inspiration from the ATM solution P-NNI (Private Network-to-Network Interface).
Of progress towards achieving better results in the optical transport layer, Ocakoglu says: "As to when carriers will see a unified and ready-to-use set of standards is a difficult question, but reasonable progress will have been made by 2005."
From circuits to packets
Dr Andrew Wheen, principal consultant at telecoms network builder Mentor, agrees that the roll-out of new and better QoS solutions will be very gradual.
Wheen says: "Everyone sees a gradual migration from circuits to packets, but there are several different views as to how this will be achieved.
"Some carriers have committed themselves firmly to carrying all forms of traffic over MPLS-enabled IP routers. However, other carriers, particularly the incumbents, are more cautious about this approach. They believe that further work remains to be done in areas such as QoS guarantees, equipment reliability, standardisation and vendor inter-working."He predicts: "This second group of carriers generally have significant investments in ATM equipment, and it is likely that they will follow the ATM/IP multi-service route for the next few years."
Wheen says the protocols chosen by operators will depend on the type of SLAs demanded by customers. He says MPLS will certainly enable an SLA to be provided, but that many feel that the hard and reliable QoS currently provided by the circuit switched network for voice traffic can only be matched by ATM.
Wheen points out that some IP-based carriers have attempted to avoid the QoS problem by seriously over-provisioning their networks - usually by a factor of five - which is obviously expensive and which is rapidly becoming an unsustainable solution.
The end of ATM?
The last word goes to another carrier though, which is taking MPLS very seriously. Richard Collins, MPLS VPN product manager at pan-European carrier Interoute, says the company expects ATM to disappear within five years.
Collins says, "We can expect MPLS to gradually replace ATM and Frame Relay as telecoms managers and heads of IT within businesses become more knowledgeable of its benefits. MPLS offers a credible migration path from Layer 2 solutions providing what could be described as a 'bridge' between traditional Layer 2 technologies and full IP at Layer 3."
Collins says MPLS could supersede ATM in five years time as a result of its superior scalability, flexibility and cost efficiencies.The industry will have to wait and see if MPLS and its big brother GMPLS can achieve more than simply turning up and turning the heads of the rest of the class. But with the school of protocols seeing a number of industry drop-outs as a result of financial difficulties, anything which can deliver greater operator efficiencies has a chance of graduating with honours.
Antony Savvas
Networks and Telecoms writer
Antony Savvas is a freelance networks & telecoms writer: [email protected]
| Glossary | |
| SONET | Synchronous Optical Network |
| SDH | Synchronous Digital Hierarchy |
| ATM | Asynchronous Transfer Mode |
| MPLS | Multi-Protocol Label Switching |
| GMPLS | Generalised Multi-Protocol |
Sabella's research interests have covered the fields of optical device technology, high-speed optical communication systems, and WDM optical networks. In May 1997 he became the technical coordinator of the research consortium named CoRiTeL.
Since 1999 he has been the manager of the research department on new technologies in Ericsson Lab Italy. He holds two patents on optical cross-connects, is co-author of a book on high-speed optical communications, and author and co-author of more than 80 papers in either international scientific/technical journals or for international conferences. He has been lecturer (professore a contratto) at the University of Rome "Tor Vergata", and at the Polytechnic of Bari, both in Italy. He is member of the IEEE/LEOS Technical Committee on Optical Networks and Systems, was chair of the Optical Networks & Systems session during its annual meeting in San Diego, October 2001, and member of the editorial board of the journal Photonic Network Communications (New York: Kluwer Academic).
For the same journal he has operated as guest editor for a special issue on WDM transport networks. He was also guest editor of a special issue on optical networks for the journal Computer Network, and guest editor for the special issue 'Optical Networking Solutions for Next Generation Internet Networks'.