IBM's New Customer Design Center Focuses on High Availability

by Alex Woodie

In late October, IBM began planning for its first customer visit at the new Availability Development Center in Poughkeepsie, New York. Company representatives aren't saying who the iSeries user is, but they are eager to talk about what will go on at the Availability Development Center, and why it was created.

"Everybody needs high availability," says John Reed, IBM Rochester's new director of client availability solutions, who formerly held the title of iSeries product line manager. "Nobody is ever satisfied with their level of availability--obviously, since I got this new assignment. If there's a company that doesn't think they need high availability, maybe they're not taking advantage of what IT can do to make them more efficient."

To help customers maximize their availability, IBM created the Availability Development Center. The goal of this new center is to help users get a better understanding of what business processes factor into their availability, what new skills they need, and how they can use technology to meet their availability requirements. Poughkeepsie is the starting point, and, depending on how things go, more Availability Development Centers could spring up around the globe.

The Availability Development Center was created because of the growing complexity of companies' IT environments and the difficulty in managing the different parts as a cohesive whole. "We know the environment has gotten more complex; a lot of things have been added to the stack," Reed says. "The idea behind the Availability Development Center is to have a physical place, for a person to walk through the set of steps, to look at the business requirements and how it flows into IT requirements."

The first step is to assist customers in determining their availability requirements, such as having no more than 30 minutes of downtime per week, or needing 24/7 availability. After that, IBM helps the customer translate that business requirement into a technology requirement. "Then we get an opportunity to architect this thing; then you have to do some coding," Reed says. "The idea is to have a place where you have skilled people, but also the technology--the servers, the storage, the network--to have all that stuff come together, so you don't have to tear apart a development box."

The Availability Development Center is quite new, and IBM is open to what it might become. There are definitely certain things that it won't be, however. It will not be strictly about exploring "traditional" high availability technology, such as replicating data and objects between two servers using mirroring, remote journaling, or clustering technology. It won't be all about the iSeries, either; after all, this is Poughkeepsie, IBM's mainframe lab and factory, and there will be xSeries and pSeries servers, Shark disk arrays, automated tape libraries, Fibre Channel networks, and compression technology there, too.

The new center is not dedicated to creating new products, but that's not to say it won't happen. "The light bulb might turn on," Reed says. Business partners will be invited to participate, but it's not about setting up new contracts. "It's about focusing on the client," Reed says repeatedly. Some things at the Availability Development Center will be free, and some will be for a fee.

SHORT TERM, LONG TERM

There are both short-term plans and long-term hopes for IBM's Availability Development Center. In the short run, IBM wants to help real-world clients in specific industries to improve the availability levels of their particular computer programs. For example, most data centers have a mix of computer platforms, says Steve Finnes, IBM Rochester's high availability product manager. If a bank added a WebSphere-on-Linux front-end to its core OS/400-based transaction processing system, how do you get those two systems talking to each other and communicating for things like roll-overs? "All the pieces in the chain need to be continuously available," Finnes says. "It has become an order of magnitude more complex."

At the same time it helps customers with their specific problems, IBM is looking to expand its knowledge base of availability techniques and standards for the different industries it serves. "We're looking to improve our overall methodology," Reed says. Even though each client engagement will be unique, the meetings could help IBM establish more standardized approaches for achieving high availability, which will end up benefiting more customers in the long run.

This leads into the more intriguing, long-term potential of the Availability Development Center: providing a real-world laboratory environment and testing ground for out-of-the-box thinking and the incubation of new approaches to high availability.

"New approaches" to high availability doesn't necessarily mean new technology. One of the examples that Reed brought up was subcapacity pricing for WebSphere on the iSeries. While there may be some new lines of code written for subcapacity pricing, it's more about a new licensing and delivery method for existing technology, which can increase the level of availability by giving users more granularity in pricing and control over the consumption of technology.

Complexity is an enemy of availability, and one of Reed's new tasks is to find ways to purge complexity from the iSeries and its software stack. Reed points to the wizard-based installation scripts for WebSphere Application Server Express and the WebSphere Portal Express products as good examples of how IBM can cleanse its products of the demons of complexity. "When we looked at Portal, without the wizard it took 138 steps to set up," he says. "It was almost daunting to set up."

With the new wizard guiding the WebSphere Portal Express set up, the customer only needs to answer 17 questions, and the product completes the rest of the set up. "If we could do that in terms of overall availability and setting up your mirroring and independent-ASP-type of technology . . . wouldn't it be nice, to answer 20 questions to set it up?" (To read an IT Jungle essay on exactly this sort of solution, see "iSeries High Availability Should Be Integrated and Invisible," in last week's issue of this newsletter.)

IBM will also be inviting its high availability business partners to Rochester for a series of "deep dives," or in-depth discussions on high availability topics. These meetings will feature subject matter experts on the box and will involve discussions of high availability "best practices," such as determining baseline availability levels, setting high availability business objectives, and architecting and testing implementations. "All of this is about achieving business goals," Reed says. "It's about cooperating to meet goals, broadening the aperture, and making this whole ecosystem more resilient."