While grid computing vendors Sun and IBM are taking the reins of the utility computing market, the competition is collaborating to provide proof-of-concept projects to improve and showcase their services and offerings.
This week brought us the announcement of Project MegaGrid, a collaborative effort between Dell, EMC, Intel and Oracle that is focused on developing a standardized approach to designing, deploying and maintaining an enterprise grid architecture.
Each of the four companies is bringing in the technologies for which it is best known. Dell is providing the complete networking architecture, including the server hardware. EMC is providing the storage networking architecture. Intel brings its processors and expertise in processor and server management. And Oracle provides its database software and management tools designed to run in a grid environment. Each of the participants is contributing cutting-edge products to the test environment.
The project is being done in phases. Phase 1, now complete, compared two different grid models, one based on dual-processor Xeon servers and the other based on quad-processor Itanium 2 servers, versus one of the latest generation large SMP Unix servers currently on the market. The goal was to run a real-world telco application at a rate of 550,000 transactions an hour in order to simulate an environment that could benefit from the advantages that grid computing offers.
The project claims to have demonstrated that Oracle Real Application clusters are capable of handling the scale-out approach of grid computing without requiring any application changes, as well as showing comparable performance to the single large, high-priced Unix server.
The actual deliverable from this test project is not something that a customer can pick up the phone and purchase; instead, the goal of this part of the project was to determine best practices for the deployment of large grids with the participating vendors hardware and software, guides to handling large cluster performance management, as well as detailed information on capacity planning in these grid environments.
Phase 2 of Project MegaGrid will include not only the continued expansion of the grid itself, but testing of the key technology of resource provisioning. Without a good, reliable provisioning solution, the management of large-scale grids is simply impossible. In conjunction with the testing of workload management, additional Phase 2 testing will include handling single-point failures and determining the impact of such failures on grid application availability.
While there are no products coming straight out of Project MegaGrid, the testing process will give the participating vendors good real-use data on how to design and configure the combinations of hardware and software grid technologies offered by these vendors. Even though it would be limited to the participating vendors, there is the opportunity for these vendors to offer standardized combinations of products that are configured to solve specific business needs.
Frankly, IBM, Sun and even HP have quite a jump on these four vendors, with large deployed grid installations, established hardware platforms, and grid and utility technologies that are already in use with end-user customers. While theres no question that Dell, EMC, Intel and Oracle all have a place in the grid market space, it remains to be seen if this showcase project is enough to raise the profile of the vendors in the eyes of the business community currently implementing or considering grid projects.