DTM - Riding to the rescue of IP?

April 2002

While Ethernet has become the latest in a long line of technologies capable of 'saving' broadband, Net Insight's Tomas Duffy offers an alternative view. In his opinion, it is DTM (Dynamic synchronous Transfer Mode) that truly gives operators the commercial tools they are looking for

A year ago DTM (Dynamic synchronous Transfer Mode) was standardised by ETSI as a technology that provides dynamic channel establishment and dynamic bandwidth adjustment over isochronous channels using simple and predictable switching mechanisms. Furthermore, it offers high bandwidth, low arrival jitter and strict Quality of Service at a low cost. ETSI officially recognised DTM as capable of delivering a multi-service transport service with scalability into the future.

The question is, thus: 'why is it a year on that everybody's talking about Ethernet and DTM is still lurking in its shadow?' While Ethernet has begun to be regarded as the saviour for broadband services, many proponents are missing some important factors in their enthusiasm. Moreover, over the past year many companies - especially those in the media industry - have realised that DTM is perhaps the only technology that capable of delivering quality broadband services that people will be happy to pay for.

To understand this, let us review how IP works in conjunction with other protocols including Ethernet, ATM and SDH/Sonet and then see why DTM can be viewed as an attractive complement to these protocols or as a complete replacement.

IP network glue

IP is a packet-switched technology fully optimised for data transport that has been identified as the common interface for all network services in the future: IP telephony, IP TV, IP storage, video conferencing, Internet et cetera. The idea is to simplify network management by gathering all services under a common ground. People often talk about pure IP networks. But IP packets are not sent directly through the fibre without an underlying transport technology. Most common protocols for IP traffic today are Ethernet, ATM and SDH/Sonet. They all have their specific benefits and drawbacks.

IP over Ethernet -- Ethernet was developed during the 1970s to make data transport in local networks as cost effective as possible. The technical principle involves full sharing of the network capacity with IP packets being blended together with packets from other users. With high network load queues are generated, causing the same type of delay we experience driving through rush hour traffic.

IP over Ethernet is therefore often called 'best effort' or 'send and pray'. It is never possible to predict exactly how high the load is going to be. In fact, the only way to assure the service quality is to heavily over-dimension the network capacity. But Ethernet is still a good choice for services that are not delay-critical, such as e-mail, file transfer and surfing on the Internet.

Today, Ethernet is commonly installed at the edge of networks, in LANs or for access to operator backbones and during recent years Ethernet vendors have developed technologies with higher capacity. For example, Gigabit Ethernet and 10-Gigabit Ethernet address the backbone market - but the field of application is still transport of 'best effort' data.

IP over SDH and Sonet -- In contrast to Ethernet, SDH/Sonet is optimised for voice traffic, but the technology is also widely deployed for data transport. SDH/Sonet fully utilises the TDM (Time Division Multiplexing) principle, reserving a separate circuit between sender and receiver in the network - as a delay of only a fraction of a second makes a phone call very annoying. IP transport is handled the same way as telephone calls. Instead of sharing the network links with other packets, IP packets are sent in an undisturbed and even flow through the reserved channel.

The drawback is that a large portion of the network capacity is left unused when transporting IP traffic. Since SDH/Sonet is developed with telephony in mind, it is ill fitted for the more unpredictable data traffic. SDH and Sonet networks are built with bandwidth in large increments: 155Mbps, 622Mbps and 2.5Gbps.

Bandwidth cannot be adapted to individual users or service requirements. IP over SDH/Sonet is expensive, and is therefore primarily used to transport larger traffic streams in high-capacity backbones. Further out in the edges of the network, the IP traffic is normally shifted over to a more flexible Ethernet infrastructure.

IP over ATM -- ATM is positioned somewhere between Ethernet and SDH/Sonet. The IP packets are loaded onto so-called 'cells' that can be granted free paths before other cells on their way through the network. The general idea is that certain content should be prioritised to guarantee the quality of the transport.

But the complexity has made ATM a very expensive solution, hard to scale into large networks and difficult to manage. Additionally, prioritisation has not solved the problem of delay - when a growing part of the traffic is streaming media, needing a free path. In practise it is like letting more traffic into the bus lane. Eventually it will be as crowded as the other lanes.

Best of all worlds

Step-by-step integration of DTM equipment into existing infrastructure makes it possible to offer all the new services we talk about: bandwidth-on-demand, video-on-demand, cable TV, IP-telephony, Virtual Private Networks, and fast Internet -- all through the same plug in the wall, and all at the price of Ethernet. Transporting IP, DTM is the only technology that has successfully managed to combine the Quality of Service, inherent in SDH/Sonet, with the flexibility and economy of Ethernet.

Built on the same founding technical principle as SDH/Sonet, DTM can carry demanding IP traffic such as MPEG video without delays or stops. The ability to build bandwidth with small channels, with different amount of channels in each direction, makes it possible to build datacom services at the price level of Ethernet.

DTM also complements both Ethernet and SDH transport. It improves their ability to manage IP. If you do not want to throw your entire current infrastructure away, you can leave Ethernet as the physical interface for IP, but let DTM guarantee the transport. This way it is possible to promise 100 per cent Quality of Service. On the other hand, DTM can function as the physical interface for IP with SDH/Sonet for underlying transport. With this configuration, IP traffic can be sent over smaller channels that easily can be adapted to various needs.

A significant benefit of DTM technology is that it enables building of large networks, covering both access network and backbone. A network operator can operate the same technology both for aggregating large traffic flows and for reaching end-users.
In tomorrow's broadband networks, operators will need to charge for these services. The qualities of DTM make it possible to guarantee services and charge different end-users for the exact amount of bandwidth they use. This is a great advantage over Ethernet, where all users share network resources -- often unequally distributed.

The right blend

The transparency of DTM technology means that it can be easily integrated into infrastructure already installed. Network operators can make a step-by-step transition towards next generation optical networks, and choose to build out their networks in a pace set by customer demand.

A general perception is that Ethernet and Gigabit Ethernet will strengthen its position as physical interface towards end-users in the access networks of tomorrow. It is therefore important for network vendors to offer Ethernet as the common data interface.

DTM takes it all one-step further and improves the Ethernet transport by introducing Quality of Service. Also, SDH/Sonet infrastructure can be modernised to squeeze out more performance from the installed base. DTM can also be used to channel the capacity, enabling better resource utilisation and brings support for a new set of revenue-generating services.