X
Business

Fujitsu prototypes resource pool architecture cloud servers

Fujitsu looks to bring detailed on-demand resources into datacenter provisioning
Written by David Chernicoff, Contributor

Referring to Fujitsu pool architecture cloud services as a "server" is a bit of a misnomer in the traditional sense. Fujitsu Laboratories is looking to develop a resource pool based cloud serve technology that can be sized and configured to fit any demand by adding the appropriate resource to meet the needs of the customer's business, or the fluctuating demands of the cloud.

The goal of this technology is to move away from the need for traditional servers that are explicitly tailored to specific applications, such as web services. With the pool technology, resources such as processors and storage are assigned as needed. The technology uses individual pools of specific resources, such as CPUs and hard disks, that are tied together via high-speed interconnects. The performance of the interconnects is such that the disks perform as if local to the CPUs that require their resources. In the prototype version, the CPU/HDD interconnects ran at a speed of 6 Gbps.

But this type of architecture is far more than just the hardware; it requires a couple type of software that can cope with the potential changes to the hardware backend in a seamless fashion that minimizes any impact on users of the supported services. Fujitsu has broken this part into pool management and middleware.

The middleware handles the storage. It controls the disk and data management functions and can be used to configure storage resources for the most appropriate type and performance level of data delivery. The pool management allocates the necessary pool resources (CPUs, HDDs, etc.) as demanded by the application being provisioned. It handles the deployment of the selected operating systems and the resources controlled by the middleware. It is the critical component of the on-demand provisioning process.

Fujitsu has been able to deliver a flexible environment in the lab and hopes to be able to bring this technology forward. They believe that this resource pool model will enable their services business to deliver the most cost-0effective and flexible datacenter infrastructure in the industry.

Editorial standards