CONVERGING READY OR NOT
Concern about uniting voice and data is being fired out from all angles. Is it any wonder then the technology is lagging behind an impossible dream? Too much time has been spent talking instead of doing, as Niall Magennis finds out.
Whether you like it or not, for the rest of the year you're going to have convergence rammed down your throat. Vendors have decided there's rich pickings to be had from the voice market and they're going to use resellers to sell it.
But the question is, are the technologies ready for market and are users going to have enough faith in data networks to trust them with their voice traffic? Unfortunately, the answer is a definite maybe in both cases.
So far, despite the hype, there are practically no Lan telephony systems being bought in the UK. Most of the vendors are still operating product trials and those that are out of that stage are not really seeing much take up of their products. Lan telephony is not so much a market as a potentiality.
Simon Wallis, sales manager at indirect channel research company Compubase, says: "We did some research back in June and what became apparent very quickly was that there are very few resellers that are declaring they have specific technical skills in the voice and data convergence sector. Although much has been made of the potential, in terms of revenues, there are very few players in the indirect channel."
He attributes this to a lack of solid products available from the key manufacturers. "It's this shortfall that's holding the process back. But I'm sure the moment the leading vendors bring the products out, which they're planning to do this year, then we will see that change very quickly," Wallis adds.
Guy Koster, sales manager for Nortel products at Landis, agrees that it's a bit early to start talking about a market for convergence products.
I don't think anyone's going to be diving into the convergence pool for probably another two or three months at least, and certainly not on the Lan," he says. "On the Wan, products have been available for a while. Nortel's Micom Marathon, now known as Passport, is a Wan convergence box. It hasn't had a huge take up in terms of market penetration, but it's getting more visibility with the Nortel merger."
But others believe it may take even longer. Colin Gibbs, systems engineer manager at Nortel Networks, says the industry might have to wait until next year for convergence on the Lan to really take off.
"I would think that by the middle of next year, the early adopters and leading edge players will get into it," he predicts. "By the middle of the following year, there will be a lot of mainstream companies testing it and putting it in their branch offices. And parallel with those, there will be the more traditional firms evolving their standard PBXs to run over IP infrastructures using cards."
Many companies already use hardware that combines voice and data on Wan links, but these aren't necessarily looked upon as convergence systems.
The real crunch comes when people decide to use convergence on the Lan.
This is because many Wan technologies, such as frame relay and ATM, have quality of service (QoS) mechanisms built in, to guarantee that packetised voice traffic won't suffer from data congestion. In contrast, ethernet on the Lan doesn't have any inherent ability to guarantee that voice traffic won't be affected by data overload.
This has led ethernet manufacturers to try to map some of ATM's support for different classes of traffic onto ethernet networks. 3Com is one firm that is attempting to deliver not just QoS, but also class of service (CoS) across Lans.
Neil Kinder, technical director at 3Com, says: "CoS is a way of labelling a particular flow of information, for example, a SAP application being labelled one way and email traffic being labelled another, so one is given priority over the other. QoS is about guaranteeing certain delivery timescales or bandwidth availability for a particular flow. The technology that lends itself best to that is ATM, but we are certainly able to use very good technologies for implementing CoS on gigabit ethernet as well."
Being able to offer QoS and CoS is crucial to the success of convergence systems. 3Com is pushing hard to convince users that ethernet networks really can be used for voice services. The vendor has just introduced a set of Lan to ATM systems that allow the mapping of CoS from the packet world onto QoS in the ATM world. On top of this, it has a special set of drivers, called Dynamic Access, for its interface cards that allow traffic to be prioritised on a per application basis. 3Com is also using IEEE 802.1p to offer CoS throughout the network.
Other vendors are following a similar path. Nortel is also using prioritisation to offer QoS guarantees. "We use IP QoS mechanisms, such as differentiated services, which allows the different QoSs to be set depending on the policies applied to particular elements. At the edge of the network, there is the IEEE 802.1q and p standards - layer 2 at the edge," says Gibbs
Lucent is taking a slightly different approach. While it uses all the same QoS and CoS standards as the others, it is dependent heavily on directory services for the overall management of convergence traffic flows.
John Collins, EMEA internal offer manager at Lucent, says: "The idea is that we can use different prioritisation schemes in different parts of the network, based on what is right for that part of the network. The customer doesn't have to worry about it because the policy management tools set the priorities and manage all the different capabilities of the switches."
Obviously, policy management is a complex process to set up and it's not necessarily something that customers are going to want to do themselves.
The outcome of this could potentially be another revenue stream for resellers.
Collins says: "There are significant issues involved with implementing policy, creating the directory infrastructure, understanding what the packet formats look like and how to prioritise the queues in a particular infrastructure product. That's the service a reseller can provide. What they can turn over to the network manager is a very simple GUI."
Lucent's idea is that resellers will move from supplying boxes to a situation in which they are involved in the specifications of the network and services to support the customers' use of that product.
"We think that's where resellers can do more than just pushing boxes," Collins argues. "They'll become business process consultants, tying the business process and the business objectives to the infrastructure using policy as the tool to get there. Resellers can do a lot with policies to make it easy for their customers to implement them, but the revenue is in the set-up and the process analysis."
Services could be a key revenue generating sector for resellers of convergence systems. "The margins resellers are making in tin are diminishing and those that aren't investing in their staff in terms of training, technical skills and their ability to deliver service to their customers, are going to be in an underclass," says Koster.
Wallis believes convergence products could signal a significant shift in the channel away from pure box-shifting. "There should be good margins on offer and the opportunity to sell services on the back of it is enormous.
If IT and telecoms are linked, then resellers have a stranglehold in terms of maintenance over the user.
"I would guess the whole business of integrating these things is going to be complex, which means there will be a lot of services built in on the back of the products. There's a good opportunity there - no question about that," he adds.
But before this becomes a reality, there's a long way to go to convince customers that Lan systems really can offer the same reliability as traditional telephony switches. "When you think about what a PBX is, it's only a very sophisticated computer with a telephony interface in it," says Gibbs.
"How you differentiate between an NT server and a PBX almost comes down to semantics. The biggest difference is that the PBX tends to be more reliable."
But Kinder believes that data networks can be equally as reliable as voice networks - it's just that customers tend to under-specify the fault tolerance capabilities of data networks in comparison with the voice networks.
"To some extent there is going to be some hesitancy to overcome, not because the technology is unreliable, but because it's unfamiliar," he says. "When people talk about low reliability expectations with data networks, they are usually referring to the fact that in many, the actual infrastructure hasn't really been designed for high levels of resiliency.
"While the voice network will generally have back-up power supplies, switching modules and links between all the PBXs, that's not always the case with the data network, which results in a reduced level of availability."
This brings up the whole issue of whether data vendors really understand the levels of reliability that their customers demand of their latest voice networks. Naturally, Gibbs doesn't think that they do know this.
"I would worry about companies that only supply data systems, trying overnight to recognise and invent everything they need to in their product lines," he explains. "We've got a lot of experience in voice and in the reliability mechanisms that underpin it and that's what many vendors don't have. In this case, there is absolutely no substitute for experience."
It seems likely therefore that there will be a similar amount of mud-slinging in the channel during the next couple of years over who is experienced enough to integrate voice and data systems.
"What's happening is that there are a lot of telecoms installation companies moving little by little into the IT sector," says Wallis. "But they don't really have the IT experience, while the IT guys don't have the telecoms experience. So there are two lots of resellers moving towards this sector, but for the moment it's a very small trickle. In the UK, there are probably less than 200 companies that say they have convergence skills."
Resellers and integrators are obviously taking a cautious approach to this emerging convergence market, and with products at such an early stage in their development, that might be a very clever move.
But happy customers are worth more in the long term than a quick profit on immature products now.
OVER THE RAINBOW - A HISTORY OF CONVERGENCE
True convergence has had vendors chasing rainbows since the early 90s. The extra revenue stream the big vendors could generate if convergence products were to take off could potentially be measured in billions of dollars worldwide. This is the crock of gold that the leading players have been dreaming of.
Unfortunately, the technology has always lagged behind expectations.
One of the early forays into convergence was Isoethernet. This bolted basic rate ISDN onto standard 10Mbps ethernet, but ultimately proved to be a failure. The problem was that the system was expensive, didn't scale and was cumbersome to use. Its proponents argued it could be used to deliver time sensitive traffic to the desktop, such as video and standard telephony but, in the end, users decided it didn't offer enough advantages over standard 10Mbps ethernet.
Next came ATM. Originally developed by carrier organisations and funded in part by media companies such as Blockbuster Video because of the possibility of using it for video-on-demand services, the technology was later adapted for the Lan.
Led by Fore Systems and after that IBM, ATM was taken onto the Lan as a proprietary technology which borrowed much from the telco standards, but tweaked them for lower speeds.
Fore introduced routers and switches to allow ATM to be used as a fast Lan backbone technology. It delivered speeds of 155Mbit/s. IBM went to work on the desktop version and adapted much of the technology from Token Ring to produce ATM25, which brought 25Mbit/s to the desktop. Later, both these implementations were tweaked and ratified by ATM.
But both technologies were expensive and lacked applications that could truly take advantage of ATM's undoubtedly superior technology. While ATM is still used in many Lan backbones, it has all but disappeared as a viable desktop package. Users have been persuaded by the cheap bandwidth of 100Mbps ethernet and, more recently, gigabit ethernet.
Most of the technologies being pushed now as convergence systems use a hybrid of ATM and ethernet. Bandwidth management and quality of service (QoS) technology is used to map some of ATM's characteristics onto ethernet networks.
ALL ABOARD THE CONVERGENCE TRAIN
One of the early problems facing vendors promoting convergence products was that nobody fully understood what convergence was supposed to offer.
While this has changed over the past year, mainly due to the hype vendors have managed to spin around the their ranges, a lot of education is still needed. Both resellers and customers need to get their expectations in line with the realities of what converged products can deliver.
All the main vendors are running seminars for resellers and integrators. While these can often be mere product announcements, they do offer an opportunity to question them over their strategies. Some companies, such as 3Com and Lucent, run hands-on days as part of their education policy, but these may only be offered after resellers sign up to those particular products.
With vendors so desperate to sign up partners for their convergence strategies, no one should be afraid to insist on some vendor-assisted training for their sales support engineers or network designers.
"I'm sure vendors will bring out heavy technical support programmes with their products," says Simon Wallis, sales manager at Compubase. "A lot of the problem is that there isn't enough expertise in the channel. The crossover between telecoms and IT is one that people are very nervous about and they're not making moves in terms of training and so on to get on top of that. Once they see there's real potential there then they will do it - it's a true chicken and egg situation."
He believes most resellers are waiting for serious products to turn up before they commit to training for staff. "While there are no serious products from the leading vendors, they won't move," Wallis adds. "Once the products emerge and they can see they can make money out of them, then they'll start moving."