To print: Click here or Select File and then Print from your browser's menu

This story was printed from silicon.com, located at http://www.silicon.com/

Story URL: http://comment.silicon.com/0,39024711,11036129,00.htm

The Bloor Perspective: web services reality, strong CA and big iron AMD
This week Robin Bloor and his colleagues consider realistic adoption of web services, progress at a top three software vendor, and a big chips win...

By Bloor Research

Published: Monday 28 October 2002

There is no disputing basic web services on offer today can offer value. But these are undoubtedly limited by the multitude of problems - security and vendor hype being the biggest ones.

This is further hindered by the reclassification of projects - legacy applications integration for instance is nowadays considered a web services project.

However, everybody knows the real value of web services comes from the ability to deliver software as a service. And that is where IDC takes the stage.

"Most of the web services vision is just pure speculation, with no real consideration of what is achievable and it will cost to actually build out the vision for full use on the open Internet," says Rikki Kirzner, an IDC research director.

Worryingly, he's absolutely right. Depending on who you talk to, and who you believe, web services should provide a seamless, secure infrastructure for all kinds of services that can be deployed to anybody you want to deploy them to. That means web services are going to be built around software and deployed on a rules basis. However, as IDC rightly notes, the tech industry is currently in no position to deliver this kind of offering.

For starters, true web services deployments are going to require the deployment of software components that can be rehashed and integrated into a multitude of front-ends across a multitude of businesses. That simply isn't possible right now, not cost effectively anyway. IDC notes that major technology hurdles stand in the way of such deployments before anyone can consider rolling out software components from third parties.

Similarly, how many businesses today would be prepared to even consider this route given the way organisations currently view their software, intellectual property and development assets? Not many. And that's before you even start delving into the current, clear problems with a lack of standards, end to end security and the marketing baloney touted by the vendors. As IDC says, web services in their true, much hyped form will not be available for at least 10 years. You've been warned.

*CA bucks trend*

Reporting its second quarter results Computer Associates said it had grown revenues by 5 per cent year on year to $772m. This, along with some presumably rigorous cost cutting measures, helped CA deliver earnings per share of $0.04, outdoing the analyst consensus of $0.02.

The cost cutting measures at the firm appear to be working to good effect. Net losses showed some dramatic improvement too falling from $291m in the second quarter of last year to just $52m in the Q2 2002.

Better still, of course, sales are on the up. CA some time ago switched to a subscription payments policy for much of its offerings. It was criticised by many but looking at how it is panning out for the firm those critics must be biting their tongues. With $394m in deferred subscription revenues for the second quarter, an increase of 14 per cent year on year, CA is now sitting on a future subscription revenue stream of $3.3bn.

Best of all, subscription revenues now equate to more than 45 per cent of CA's total revenues coming through the door, which they can predict to the very day. That's some nice financial information to have to hand - and a considerable boost to any sales force.

That doesn't mean CA is having an easy time out there because it isn't. CA's professional services arm is said to have seen its revenues fall by more than 18 per cent. Sanjay Kumar, president and CEO, said: "Customers are buying, but they continue to purchase technology on an 'as-needed' basis, and they are seeking greater returns on their investments."

This is true and would appear CA is having to work pretty hard to keep things moving along so effectively. It's one of few tech firms today growing revenues, which makes Kumar believe they are stealing market share from the competition.

But don't expect miracles. The outlook for the end of the fiscal year at CA remains the same, an estimated $3.1bn in revenues.

*AMD's super Cray contract*

AMD received a welcome winter warmer this week from Cray, the maker of supercomputers, which has just appointed the firm as chip supplier for its latest monster mainframe. The computer, which has been commissioned by the Sandia National Laboratories in the US, will be a colossal machine, giving AMD a further boost. But it must come as a heady blow to Intel which likes to think it has the chip market sewn up.

The chips of choice in this case are AMD's Opteron, a chip based around a 64-bit architecture. With more than 10,000 Opteron's juiced up and ready to deliver, it is expected the system will generate some 40 teraflops. That's 40 trillion calculations per second to you and me.

Naturally, this isn't what you would call a cheap project. The supercomputer will cost somewhere in the region of $90m. It's all part of the Accelerated Strategic Computing Initiative (ASCI) which is the US government's nuclear stockpile simulation project that has seen a raft of supercomputers delivered to leading research institutes, including the famous Los Alamos laboratories.

This latest machine is being dubbed Red Storm and is very much being seen as a follow-up to the previous ASCI machine that Cray delivered for Sandia some years ago. The difference this time around is one of size.

Over the past few years supercomputer development has radically changed. In the past mainframes were large scale single machine, MPP systems. Over recent years, however, this has changed to encompass a mass of clustered systems - where arguably cheaper Intel architectures are strapped together to deliver a high performance.

A prime example of this is found in the www.top500.org list of the most powerful supercomputers. It currently has more than 30 supercomputers which are, essentially, clustered IBM Netfinity machines.

In this case Sandia has opted out of the clustering scenario preferring instead the performance efficiencies that can only be delivered through a single system. Interestingly for AMD, Sandia has also chosen its chips as the power base. This isn't the first time AMD has scooped a supercomputer contract. It has some seven supercomputers to its name, but Intel is already delivering chips with this kind of power. AMD won't have them until next year. Sandia, however, found that the AMD architecture was more suited to its needs.

**Bloor Research is a leading independent analyst organisation in Europe. You can find out more at http://www.bloor-research.com or by emailing mail@bloor-research.com .