Cloud-in-a-box from Oracle
Oracle has delivered its promised "end-to-end" stack of hardware and software. The Exalogic Elastic Cloud is a version of Oracle's new-ish Exadata appliances, which are custom-painted
It is Oracle's "cloud-in-a-box" offering, a ready-made platform for virtualized computing. Plug one in, Larry Ellison says, and you've got a place to run your virtual machines and tap all sorts of fun Oracle products "transparently."
"Any Oracle Applications, including the Oracle E-Business Suite, Oracle's Siebel CRM, Oracle's PeopleSoft Enterprise, Oracle's JD Edwards and Oracle's industry specific business applications, will run on Oracle Exalogic Elastic Cloud transparently, without needing any modifications," the press release said.
Oracle is in good company. IBM has CloudBurst, HP touts its BladeSystem Matrix and Cisco sells the Unified Computing System: all of these do what Oracle is claiming, to various degrees and in various platforms.
Larry Ellison said that two of the 360-CPU devices could handle Facebook's Web request loads. He also said that pricing would start at $1 million each. SearchCloudComputing.com fully expects Facebook to immediately buy two Exalogic appliances for $2 million dollars and prove him right.
Microsoft HPC will support Azure
Microsoft's Windows HPC Server 2008 R2 will be updated to include spill-over capacity to Windows Azure. Redmond says that users of its high-performance computing (HPC) operating system will soon have an automatic way to extend workloads on to its cloud computing service if their own capacities are insufficient or inconvenient.
According to a demonstration, HPC nodes will be provisioned and managed through existing management tools; users will presumably have to calculate their own bandwidth overhead and the process degradation that comes with it.
Amazon now has actual HPC customers
Amazon recently released its Cluster Compute instance types, designed specifically for number crunching and functional computing. Now, the cloud provider is showing off some of the fun. The MATLAB team at MathWorks hit 1.3 teraflops doing simple math problems, and NASA Jet Propulsion Lab scientists used it for processing some really, really big pictures to help robots find their way around. Wow, so it really is rocket science.