Microsoft revamps datacentre

Microsoft Corp.’s recently revamped and renamed Datacenter High Availability Program will bring users additional options for support and a quicker path to swap out minor components from their pretested configurations.

But while the changes represent a positive step, it’s unclear how much they will affect the existing user base or whether they will spark an uptick in the sluggish adoption of Microsoft’s challenger to high-end Unix systems. The new program aspires to the same lofty goals that its predecessor – the more plainly named Windows Datacenter Program – did when it was launched in September 2000.

“The Datacenter product is great. I love that I can vertically scale Intel chips,” said Morris Koeneke, database services manager at early adopter Mary Kay Inc. in Dallas. “But the Datacenter program is in desperate need of repair, and I’m not optimistic that they’ve really fixed the problem.”

Koeneke said he would like more choice with the systems that attach to his Datacenter server, since he would like to do more mixing and matching of products. But he doesn’t want to have to pay the OEM to test the configuration.

Under the current Datacenter program, OEMs are responsible for rigorously testing a complete hardware and software configuration – which typically includes a server, operating system, attached storage, storage backup utility and antivirus software — to ensure that it will provide the sort of reliability and performance customers expect from high-end systems. The test configuration must run for 14 days without a failure.

Under the new program, Microsoft has streamlined the process so low-level component changes can be tested more quickly and reliably. Drivers or application components such as antivirus programs and backup utilities can be tested in as little as one day, once they have been certified through VeriTest, the testing arm of Lionbridge Technologies Inc. in Waltham, Mass.

Bob Crownhart, an IT director at Premera Blue Cross in Mountlake Terrace, Wash., said the new, streamlined process could help his company. He recalled a case in which Premera wanted to shift to a new driver level, only to learn that Unisys Corp. hadn’t certified it yet. Consequently, Premera had to make sure Unisys and the other vendor hooked up to work out the issue. “It was aggravation and delays,” Crownhart said.

But the new certification policy won’t help every customer. Koeneke said he was considering five storage vendors, but of those, only EMC Corp.’s had been included in a configuration that Unisys had tested. He recalled that he was once given a price tag of more than US$100,000 to test a configuration that differed from the one Unisys had tested, though he was able to negotiate his way out of paying it.

Bob Ellsworth, director of Microsoft’s Windows server product management group, acknowledged that substitutions of major components like storage subsystems in a Datacenter configuration will still require the 14-day retest. To address that, Koeneke wants Microsoft to come up with a set of predefined standards for vendors to achieve Datacenter certification.

Ellsworth said Microsoft has heard that suggestion from some customers but isn’t yet ready to support that kind of change. “It’s something that we’re investigating,” he said.

Tony Iams, an analyst at D.H. Brown Associates Inc. in Port Chester, N.Y., predicted that the program changes will make it easier to certify products on Datacenter, thereby making more systems available to customers. That, in turn, might spur more Datacenter usage, he said.

Another change that’s catching the attention of some customers is the set of new support options that will take effect April 24, with the release of Windows Server 2003. OEMs will no longer be the only choice for support. Instead, users will gain the option to contract for support from resellers, systems integrators or Microsoft.

Customers that have premier support contracts with Microsoft will gain the benefit of having Datacenter as an included product. But they will need a separate registration if they want access to Microsoft’s High Availability Resolution Queue, a 24-hour direct hot line to engineers who can solve their problems.

Pricing hasn’t been announced for the high-availability resolution program, which is an expansion of the existing Joint Support Queue.

Under the current system, OEMs are supposed to serve as the single point of contact, coordinating support among the various vendors that touch the Datacenter system. It remains to be seen how the new system will play out, since some early adopters don’t appear to be following the existing program’s procedure for requesting support.

Larry Godec, CIO at First American Title Insurance Co. in Santa Ana, Calif., said he already gets high-availability resolution. When his company has problems, it calls Microsoft, which has “jumped through hoops” to resolve them, he said. Godec said his company’s OEM, Unisys, was mentioned as the single point of contact during a couple of conversations, but Microsoft didn’t balk in helping out when called first.

Koeneke, too, said he isn’t shy about going directly to Unisys, Microsoft or his company’s storage vendor, Hewlett-Packard Co., with problems. He said he can’t imagine going with a consulting organization for support.

“They aren’t in the engineering team at Unisys, Microsoft or HP,” he said. “Microsoft is going to get bogged down with certifying a bunch of resellers of support, and I personally have yet to see delivery on repair of the existing OEM support provision.”