Yahoo, Intel and HP join to create distributed computer 'cloud' for research

Calling it the Cloud Computing Test Bed, HP, Intel and Yahoo today announced a multi-datacenter testing environment to promote internet-scale open source collaboration.

Building upon the success of Yahoo and the Apache Software Foundation's M45 project with Carnegie Mellon University, this collaboration will go several steps further and include six test beds, with each facility offering between 1,000 and 4,000 processor cores.

In a conference call this afternoon, the companies admitted to not having tallied how much power this would yield in terms of Flops, and also that power should not be the focus, but rather the scalability.

The M45 cluster promised 3 terabytes of memory, 1.5 petabytes of storage, and a peak perfomance of 27 trillion calculations per second. Multiply this by six for a rough idea of the potential magnitude of this project.

The project will be similar to Google and IBM's cloud computing initiative which also ran an open source computing infrastructure from Apache's Hadoop project.

Andrew Chien, Director of Intel Research and VP of the Corporate Technology Group at Intel pointed out that this project does not compete with Google and IBM's collaboration, but rather complements it. Those companies have concentrated upon the application level, allowing users to harness the cloud's power for specialized applications. This collaboration will provide "a greater breadth of technological space that can be explored," according to Chien, on an infrastructural level.

HP's senior Vice President of Research Prith Banerjee said his group believes this is the dawn of the "everything as a service" era, where the cloud is the platform and internet-scale computing will be available equally to enterprises and individuals.

Intelligent Infrastructure and dynamic cloud services are the two main focuses for HP in the test bed project, with major research in datacenter structuring that could ultimately result in cheaper and more energy efficient datacenters.

While some test bed locations are up and running, the "cloud" is expected to be fully operational later this year. Researchers will be given access to the system through a selection process that will take place later this year.