Analysis As El Reg previously reported, HP this morning announced a private cloud stack of integrated servers, storage, switches, systems management tools, and virtualization hypervisors called CloudSystem. The CloudSystem setup is comprised of the BladeSystem Matrix blade servers, announced in April 2009, ironically on the same day that Oracle announced its $5.6bn takeover of Sun Microsystems after IBM declined a deal.
The BladeSystem Matrix brought together HP's ProLiant Xeon-based blade servers and an integrated set of management and provisioning tools (derived from its Insight system administration and Opsware system provisioning and patching tools) called the Matrix Operating Environment. The Matrix software also included application deployment templates that turned the Matrix boxes into a "push button data center" for deploying applications. The Matrix gear is interesting and useful, and was primarily used to provision software on bare-metal servers to support n-tier applications (database, application, and Web tiers, usually).
New to the server market, Cisco Systems has no installed base of incompatible iron to worry about, and a clean slate on which it can design cloudy infrastructure that it peddles in conjunction with EMC and VMware through the Acadia partnership announced in November 2009. Not so for Oracle, IBM, and Hewlett-Packard, which are all struggling with how to configure something akin to Acadia's Vblocks without alienating portions of their customer bases.
In November 2009, Cisco, EMC, and VMware announced their Acadia partnership in November 2009, which has set up a free-standing company to integrate and support virtualized server stacks called Vblocks. There were originally three different references architectures being certified by the Acadia Three, which is now known as the Virtual Computing Environment Company, or VCE for short. But as El Reg reported last week, VCE is no longer just doing certification of reference architectures on the Vblocks, but actually manufacturing and supporting them. (Cisco, EMC, and VMware do all the sales of Vblocks, however.)
HP's response the day after the Vblocks were announced that November was to allow for Itanium-based Integrity blade servers to be slid into the chassis next to Xeon blades, thereby enabling HP-UX, OpenVMS, and NonStop workloads to be controlled by the Matrix software as well as Windows and Linux running on the ProLiant blades. HP was not inclined at the time to create a Matrix uber-system out of ProLiant or Integrity rack servers, and it renamed the secret software sauce on the Matrix box Infrastructure Operating Environment, which included the Insight Dynamics, Insight Control, and Business Technology Optimization tools woven together.
Last August, HP punted a new product called CloudStart, which was a private cloud based on the Matrix iron and a new tool called Cloud Service Automation, which leveraged the Matrix management tools but added in a self-service portal, resource metering, chargeback, and reporting features that virtualized systems require to make sure the right people get billed for the capacity they use.
The CloudStart appliances were based on ProLiant blades and support VMware's vSphere and Microsoft's Hyper-V hypervisors. A starter CloudStart with eight blade servers, an EVA storage array, and all the goodies cost around $190,000; the system could expand to around 1,000 server nodes, which is a pretty big cloud. CloudStarts were available in Asia/Pacific first, and were supposed to be expanded to the rest of the globe last December and eventually extended to support Integrity blades and HP's own Integrity VMs for virtualizing HP-UX, Windows, and Linux on Itanium-based blades.
With the CloudSystem launched today, HP takes BladeSystem Matrix hardware and layers on a slightly different stack of software called Cloud Service Automation Suite. According to Matt Zanner, vice president of cloud infrastructure solutions at HP's mammoth Enterprise Business group, which makes and sells systems and services to IT shops, the CSA tool stack is used to manage, provision, and optimize clouds, and that means managing hypervisors and their virtual servers in addition to bare-metal machines and their operating systems and applications.