CONSOLIDATION WARNING
The world of data networking will see plenty of breakthroughs and breakdowns as the industry giants, backed by millions in their kitty, muscle in on their much weaker counterparts. Convergence is the war cry and the titans of telecommunication are ringing the changes.
As every dealer knows, the history of networking is littered with theakdowns as the industry giants, backed by millions in their kitty, muscle in on their much weaker counterparts. Convergence is the war cry and the titans of telecommunication are ringing the changes. ghosts of companies and technologies abandoned by the wayside. These spectres are invariably accompanied by the wailing of network managers and their channel partners who backed the wrong technology and were left with marginalised and inadequately supported systems.
The days when you bought into a system and felt confident your vendor would be there to provide maintenance and support are a distant memory.
Data networking industry stalwarts are falling like flies in the mother of all market consolidations. And it's usually the channel that is left to carry the can when manufacturers shift technological focus or, worse still, go out of business altogether. As the rate at which firms are being swallowed up quickens, the danger of being left with a technological turkey is more real than ever before.
We are witnessing an unprecedented rate of industry change that is largely driven by the trend - real or perceived - towards the convergence of voice and data traffic over a single infrastructure. The traditional networking market is being attacked from above by titanic telecoms equipment vendors.
The new kids on the block include Lucent, created from the demerger of AT&T, with its estimated war chest for corporate acquisitions of at least £625 million. It is busy moving in on smaller data companies, most recently Ascend. Elsewhere, Bay Networks has been gobbled up by Nortel.
Formerly Northern Telecom, the latter has changed its name to Nortel Networks.
Siemens is another telecoms giant with huge resources targeted at the data networking market. Neil Rickard, research director for networking at the Gartner Group, believes that although the networking market is expanding rapidly, ultimately it will be dominated by just three or four players.
'We will see the telecoms companies moving in hard,' he claims. 'Lucent and Nortel will start to make significant inroads. Siemens will be up there with some kind of tie-up with Newbridge or 3Com. But whether this link-up will be acquisition or alliance, it's still too early to predict.
'Alcatel is also in a positive position after buying Packet Engines and Xylan, but has a long way to go before it's deemed a leading force,' says Rickard. 'Cisco won't go away, but it will feel the heat,' he adds, noting that the market muscle of this new breed of industry giant dwarfs even the mighty Cisco.
But it has always been an illusion to view Cisco as the Microsoft of the networking market. The vendor is alarmed to see companies such as Lucent moving in on its patch. 'With the high-end carrier equipment and data technology it already has, combined with the expertise Lucent has picked up from acquisitions such as Ascend, the ingredients are there for a real fight,' says Rickard.
Convergence is the name of the game and the biggest players have upped the ante. There will be more acquisitions before the dust settles over the networking market.
'We'd expect to see corporate convergence happening on a number of levels,' says Clive Longbottom, analyst at Strategy Partners International. 'There will be the top-tier players, notably the telecoms networking players, buying up small and medium-sized firms,' he predicts. 'Almost always, it is cheaper, in terms of time and money to take over a company that has a technology seen as strategic, than to develop it in-house.'
Longbottom foresees no large companies emerging to take on the existing top-tier players. He attributes this to a belief that successful medium-sized companies will be bought before they can challenge the industry big guns.
Still, he outlines a rapidly growing sector, one that will develop as firms move to outsource elements of their IT systems. This sector will comprise small service-oriented organisations, reliant on the internet to establish market presence and to deliver outsourced IT services to large corporates.
In terms of actual technology convergence, the main players are rushing to bring IP-powered PBXs to market and a plethora of such devices has been released by companies such as Alcatel, 3Com and Cisco. But despite the hype, the jury is still out among the network managers who would actually implement converged systems.
'No matter what the vendors say, IT managers are not preparing for the expense and upheaval of throwing out their PBXs,' says Rickard. 'We predict it will be at least two or three years before the convergence trend becomes mainstream and about five years before it is established. We can't yet advise users to install an IP-based Lan PBX as there are still many issues.
Lack of reliability and quality of service in data systems, compared with voice systems, remains the biggest hurdle.'
But Cisco remains bullish, predicting that all communications will converge by the year 2002. Peter Alexander, vice president of enterprise marketing at Cisco, says user concerns about the reliability of voice over data and the desire to retain investment in voice technology will not inhibit migrations.
'Customers won't throw away PBXs. The role of the PBX is changing in the way the role of the mainframe changed,' he argues. 'Standardisation will help drive out complexity and cost.'
But these claims are dismissed by Longbottom as no more than marketing hyperbole. 'There is a mantra about voice/data convergence coming from firms such as Cisco,' he says. 'But the screaming about VoIP originates from the vendors and not the users, who don't really buy into the sales hype that tells them they have to rip out proven systems and start again.
It will happen, but not yet, and certainly not as quickly as the vendors would have us believe.'
On the main Lan networking technology, the future of Gigabit Ethernet is good, or at least safe. Although the kit is confined to the top end of the market, the trend towards Gigabit Ethernet that is ubiquitous is unstoppable.
For all the competing arguments about technologies such as ATM and Token Ring, there is no doubt that the desktop battle has been won by Ethernet.
Even those firms which have long championed Token Ring and ATM admit that Ethernet is here to stay.
Just a few years ago, ATM was going to be the bandwidth panacea that promised to bring broadband to our desktops. The argument seemed convincing.
Massive bandwidth scalability was combined with guaranteed quality of service. But the prophets of doom - chiefly led by the Gigabit Ethernet crowd - have now begun to sharpen their knives, claiming that ATM stands for 'A Terrible Mistake'.
The truth is somewhere in the middle. ATM is being widely introduced by carriers and as a Lan backbone technology. There are still some high profile exceptions - British Airways is one and specialist niche players such as the video post-production industry - that really need the bandwidth.
But they're exceptions, not the rule.
Fore Systems has turned from its religious dedication to ATM and grudgingly embraced Ethernet, following its acquisition of Gigabit Ethernet vendor Berkeley Networks. Andy Williams, enterprise marketing director at Newbridge, says the move was inevitable: 'Fore had backed itself into a corner. Even if ATM was free it couldn't break Ethernet as a default standard.'
But John Wanklyn, sales manager at Fore, denied the company had backtracked: 'There'll be a continuation of ATM implementations in large backbones but Ethernet upgrades at the network edges will be huge.'
Token Ring stalwart Olicom has added Fast and Gigabit Ethernet to its product range too. Niels Christian Furu, president and chief executive of Olicom, believes the delays in bringing High Speed Token Ring (HSTR) to market effectively drove the final nail into the coffin of a system that could have given Ethernet a run for its money.
To take a lesson from the networking history books, one only has to remember HP's 100VG AnyLan technology which offered better data-throughput rates and more efficient use of bandwidth because of reduced contention. But it is only a memory now - and one that's fading fast.
Still, there are some very strong, compelling arguments for adopting a number of technologies in preference to Ethernet. Longbottom predicts that Gigabit Ethernet will be slow to achieve mainstream acceptance. 'It needs to learn to walk before it starts running. At the moment, it's still crawling,' he says. 'The migration from 10Mbps to Fast Ethernet was painless, but the move to Gigabit will need a lot of wiring and hubs and switches. Most people will wait until they have a real need for the bandwidth.'
One of the most hyped events in the world of networking resellers will be the arrival of Microsoft's Windows 2000, due later this year. Microsoft is tight-lipped about the precise shipping date for the operating system, which reportedly weighs in at about 40 million lines of code - more than twice the size of NT4. The software giant says it will ship only after it has been comprehensively beta tested and its customers say it is ready.
'When Microsoft finally gets Windows 2000 out of the door and, more importantly, gets it stable, it will be a massive event,' says Longbottom. 'It won't become the standard OS, but there will be real benefits to be seen in terms of improved scalability, speed, provision of directory services and support for quality of service. The fact that it will support native IP and remove spoofing is also important.'
But the Gartner Group predicts trouble ahead and is advising network managers and the channel not to hold their breath.
'We always said Windows 2000 would ship later than promised,' says Rickard.
'We expect it out in the first quarter of 2000. It seems that Microsoft's timescales are slipping to match our predictions. It is important to realise that this is one of the biggest pieces of monolithic software ever written and should be treated with caution. We advise users not to think about deployment until the second service pack has dealt with the main bugs.'
While network managers continue to wait for Windows 2000, there will be consolidation in the Unix market. 'The market's still growing and Unix will continue to expand,' says Longbottom. 'But the smaller players will hitch up with the bigger firms and the number of types will drop. Solaris will definitely be there. Compaq is making aggressive noises about Digital Unix, but still controls only about five per cent of the market. There will be a really hard-fought battle between IBM/SCO AIX and HP-UX.'
Despite the growing popularity of the open-source free Unix variant Linux, with its upgraded kernel, Longbottom does not consider it a contender in the corporate market. 'Linux is still a bit of a red herring. It's a great little tool for those that want it, but there are serious usability issues with its various front ends. The biggest problem is lack of accountability.
Who do you scream at when something goes wrong?' he says.
After much hype, 1999 looks set to be the year when directory services finally take off. The benefits of the technology - mainly reduced administration of network users and objects from a central point - is well understood.
But the market remains relatively immature, with only Novell Directory Services (NDS) having a significant installed base, with about 45 million seats. But Microsoft Active Directory (Mad) is on its way and Oracle has recently released its Oracle Internet Directory (OID). Netscape remains a wild card as its plans for a directory are still unclear in the wake of the AOL merger. But the main battle will be between Microsoft and Novell.
Stephen Davies, director of strategy at Novell, joined the war of words after NDS had secured the support of industry heavyweights Lucent, Cisco and Nortel. 'Active Directory is like putting lipstick on a pig,' he claims.
'We are going to see a pig with red lipstick on trying to run domains. We can increase the scalability within NT5, but we are going to do better by increasing the problems within the domain as well as the inter-domain communications.'
Microsoft hit back at Novell's jibes, denying that NDS was too established in the market to be overtaken by Mad. 'I believe people will get benefits from Active Directory to make the change compelling,' says Frances Reay, server product manager at Microsoft Windows NT. 'The main killer for Active Directory will be application integration. We already have SAP and PeopleSoft working to integrate it.'
Something for all network managers to look forward to will be cheaper Wan bandwidth, at least on common routes. Rickard predicts an explosion in the amount of fibre that is being laid down. 'Deregulation means we're getting massive amounts of bandwidth on the popular routes. The linear pricing model - where traffic is charged according to distance - is breaking up. On the information motorways, such as London to New York, the prices will come right down. The problem will be on the less popular routes, where even if the distance is relatively short, charges will increase,' he says.
One of the most talked-about technologies this year will be Intel's IA-64 Merced processor. The chip giant claims the CPU's 64-bit construction will revolutionise the server market by offering network managers low-cost systems running NT, NetWare or SCO, for essential, line-of-business applications. There is no doubt that the fledgling chip architecture has received huge levels of advance support from both software and hardware companies. The industry heavyweights Microsoft, Novell, Sun Microsystems, HP, Compaq and SCO are all jumping on the bandwagon and developing servers or server operating systems to support the platform.
But Intel was over-optimistic with its shipping dates and Merced will not hit mainstream markets until at least next year.
'As usual, we were very aggressive with our original plans,' says Orietta Sutherberry, a representative for Intel. 'Then we did a reality check. OEMs, ISVs and other partners will still get Merced towards the end of 1999 and we don't expect the six-month delay to have a big impact on customers.'
But this delay means that Risc systems such as IBM's PowerPC, Sun's Sparc and Compaq's Alpha, will be offered a respite in the mid-range server market. Such systems will continue to be the work horses for at least another two more years, according to IDC, a leading US analyst.
You can advise customers to take the risk with Risc, go Mad with directory services, throw out their PBXs and converge, or just sit tight and wait.
But whatever you suggest, when navigating the slippery, technological minefield of today's networking market there is only one rule: tread carefully.
A version of this article appeared in Network Solutions, April 1999.
CISCO MOVE SEES BREAK WITH TRADITION
Beyond calling for convergence, Cisco is moving away from its traditional role in the routing market. Paul Mountford, managing director of Cisco UK, says the networking giant will shift its attention to the SME and service provider markets, due to a slow down in the enterprise routing sector. He also puts much emphasis on the importance of the integration of voice and data, together with virtual private network (VPN) products.
Mountford also says Cisco's future will focus on alliances with companies such as Sony, producing set-top boxes and other consumer goods. Cisco is also thought to be interested in acquiring firewall, network management and caching technology.
Its decision to licence its IOS software Web Cache Communication Protocol specification to other cache vendors is a clear indication that Cisco is keen to bolster its products in this area.
Neil Rickard, research director for networking at analyst Gartner Group, predicts that Cisco will see its existing markets erode as enterprise companies increasingly outsource their communications. The service provider market will become increasingly important to Cisco as a result.
SANS FOR THE MEMORY
Storage Area Network (San) systems are one of the hottest technological trends. And Peter Coleman, technical manager at Gadzoox, claims many corporates will be starting on the bottom rung of the San ladder. 'Scalable storage on a single server will come first,' he says. 'By the end of the year, we will have multiple servers with shared storage. At least three finance houses already have their own San systems on trial.
'We expect mainstream companies to look at the technology by the end of the year.'
But Andrew Rowney, marketing manager at K-Net, is sceptical that Sans will crack the enormous installed SCSI base. 'It might well be the year of the San, but I really think it won't live up to the hype,' he warns. 'A lot of the strategies are based on fibre channel, which still has a number of inter-operability issues.'
SUN RISES OVER JAVA
Consolidation is the name of the game for Sun Microsystems. The vendor cites Java, Jini and the exploitation of its alliance with Netscape and AOL as key elements of its expansive strategy. Sun's launch of the latest generation of Java and the liberalised licensing regime - allowing developers to alter the sourcecode without seeking permission - was described as pivotal.
'The Java platform is set to take off after the launch of Java 2. Increased functionality, performance and security are significant improvements,' says Guy Martin, product marketing manager for software and technology at Sun. 'It is no longer about delivering pipes over a TCP/IP connection.
We now need to provide complete services and this is where portals and AOL come in. All Sun's acquisitions and technologies are heading towards providing end-to-end services.'
NORTEL HELD AT BAY
The branch of Nortel that was formerly Bay Networks has admitted its biggest short-term challenge will come from integrating itself with the telecoms giant. Paul Trowbridge, communications manager for Europe at Nortel, says: 'It won't be easy to bring the Bay brand value into the Nortel Networks name because businesses are most familiar with Bay.'
Jean Louis Seguineau, analyst at Meta Group, says: 'Bay has not been performing well as a networking company since the merger. It is losing some financial momentum to 3Com, which has the price advantage.'
Nortel's main corporate message is convergence but Trowbridge is realistic about the timescale before the technology will become common. 'VoIP is the biggest piece of hype at the moment because it is not there yet. VoIP is easy, but the challenge is to offer telephony as we know it today.'
Trowbridge predicts remote access will be a significant growth area: 'We expected a lot of companies to have outsourced their remote access to ISPs by now, but it has not happened as quickly as we expected,' he admits.