Hitachi Data Systems is developing a unified compute platform that the vendor says will consolidate server, storage and network management into one system automatically orchestrated through software based on an OEM of Microsoft Corp.'s System Center management software.
The new systems won't be released before 2011, but Hitachi Data Systems disclosed details today while announcing an OEM deal to distribute Microsoft Windows Server 2008 R2, System Center suite and SQL Server 2008.
The integrated system will consist of Hitachi Data Systems' USP V storage, Hitachi Ltd. blade servers, and Fibre Channel (FC) and Ethernet switching, all integrated with Microsoft System Center Operations Manager, Windows Server 2008 R2 and Microsoft System Center Virtual Machine Manager. Hitachi Data Systems' USP V will provide storage virtualization, and the systems will support Microsoft Hyper-V and VMware hypervisors.
IT infrastructure stack alliances to emerge among enterprise vendors in the last six months. Other large vendors have also launched unified computing platforms, such as Cisco Systems Inc.'s Unified Computing System (UCS) and Hewlett-Packard Co.'s BladeSystem Matrix.
Miki Sandorfi, chief strategist for file and content services at Hitachi Data Systems, said the unique part of this deal is the orchestration software — a middleware layer being developed by Hitachi Data Systems and Microsoft.
"This software will treat server, storage and networking infrastructure resources as pools to be assigned as needed to business services," which will be defined according to user-determined service-level agreements (SLAs), Sandorfi said. "The orchestrator will be able to pull in resources to meet SLAs, monitor and manage the infrastructure, and -- if SLAs aren't met -- can remediate the problem."
Hitachi Data Systems is calling the unified compute platform hardware packages Scale Units, and plans to sell them in small, medium and large sizes. The company hasn't named its networking partner for the planned platform yet.
While storage vendors are embracing pre-integrated stacks of enterprise IT products, it remains to be seen where exactly they'll find a home among customers. Some data storage pros have said they fear lock-in with a preconfigured end-to-end stack, while others have said such pre-integration is becoming necessary to cope with data growth, complexity introduced by server virtualization and shrinking operational budgets.
Analysts said they see some enterprise customers already making the transition to converged IT, at least anecdotally. "I talk to customers all the time who embrace stacks," said Wikibon.org analyst David Vellante. "Mainframe customers are a good example. VMware is headed there. Oracle is heading there, too. "
"Our conversations with large enterprises and resellers indicate customers are undertaking major virtual machine consolidation efforts, and have been describing what they want [from vendors] in terms of how many virtual machines can be supported in a self-contained pod," said Rick Villars, vice president, storage systems & executive strategies at Framingham, Mass.-based IDC.
Mark Bowker, a senior analyst at Milford, Mass.-based Enterprise Strategy Group, agrees that the proliferation of virtual machines is driving these stacks.
"This is all being driven by [server] virtualization," he said. "Some users have gotten to a certain percentage of virtualization in their environment but have stalled in neutral. The skill set of a virtual server admin might not go far enough to integrate complex applications in mature virtualization environments."
Will customers deploy a unified compute platform?
Not every Hitachi Data Systems' user is keen to deploy the unified compute platform, though. "The sweet spot for offerings like this is SMB and enterprise customers who want turnkey service and one throat to choke over squeezing every nickel out of their capital budget," said Jeffrey Papen, CEO at Peak Web Hosting and Peak Web Consulting, which does consulting and managed services for MySpace, eHarmony and Netflix, and hosts customers' data on Hitachi Data Systems' AMS arrays in its data centers.
Papen said in his experience, pre-integrated products tend to have lower total cost of ownership (TCO) but higher initial price tags. "It's a godsend for a lot of small operating environments, but it's also an expensive luxury because it removes some element of competition," he said.
Greg Knieriemen, vice president of marketing at Chi Corp., a Hitachi Data Systems' VAR, said it's good for his customers to know what's on the Hitachi Data Systems' roadmap, but he has a number of questions about how the scale units will be configured and stacked together. Hitachi Data Systems has said its orchestration software will be able to manage multiple Scale Units at once, but "as customers scale, if you max out the RAM in one [scale unit] but still have storage capacity left over, can you add RAM without purchasing another whole additional block of storage?" Knieriemen said. "For the Unified Compute Platform to be successful, you have to be able to scale while still maintaining full utilization."
"The vision looks good," Wikibon.org's Vellante said. "The keys will be that orchestration layer, a DR and backup story, and customers willing to [deploy] Hitachi servers and storage."