|
Dynamically provisioning processing resources according to application requirements and load peaks can slash power bills.
That is the promise of Platform Computing which says its software can manage heterogenuous IT as if it was a mainframe environment delivering true data center automation and cutting costs in the process.
The software layer, originally developed for High Performance Computing (typically mainframe) systems is now being deployed across Unix and Windows operating systems running on industry standard servers.
Songian Zhou, CEO of Platform told DatacenterDynamics Focus that colocation players looking to move into the systems management side of the business have been “loosely speaking” with Platform about using its software to dynamically provision resources on an as needed basis.
“We are talking to carriers and colo companies very loosely because they are at a very early stage – the enterprises have been doing this and those who are expanding out of High Performance Computing are leading the way. Colocation companies are evolving out of the colo – you wanted 10 servers, I gave you 10 servers - hopefully they are finding more sophisticated services, such as expanding into the cloud services business. Enterprises have been doing this for years in the mainframe environment and they are ahead.”
In Platform’s vision the future of data center operations will be between the internal cloud and the external cloud. The external cloud will be supplied by service providers, while the internal cloud will be driven by capital and operational expenditure cuts, mission critcal service level agreements, and high security and enterprise specific requirements. Key to this will be automation of processes and provisioning which will match supply with demand. "“It sounds complicated but the whole idea is to simplify it. The interface between IT and business is SLAs. “These are the applications I want to run, this is the performance I want, the throughput I need.” Once the SLA is provided, IT is there to marshall and deliver services and deliver the infrastructure as a service. Then it is up IT how to do that. IT needs to change the way to do that. Systems administrators are not creating a manual way of doing that. Systems management software enables IT data center operations to set up policies which are needed to translate SLAs to the business base around the sharing of resources.”"
Platform’s software sits between the application and base software that drives the devices but is application agnostic.
“What we did is move data center architecture into the third generation, called grid computing, utility computing or dynamic data center etc, bringing a mainframe type management to heterogenuous environment – across all devices, server network and storage all are viewed as resources.”
In power and cooling terms Platform says that a data center running 4,000 siloed CPUs could benefit cut its power bill by almost one third by dynamically provisioning resources using its software, cutting its energy bill from $1.8 million to $1.1 million. Taking into account a total refit involving hardware, OS support and systems admin Platform claims a cost in data center costs of $9 million ($8 million in hardware). “The electricity bill cannot be paid separate from IT because IT is causing the problem,” he said.
Platform works closely with Vmware and other industry partners but says that moving into the provisioning capabilities into a heterogenuous environment is a young market which in some ways is being held back by intertia.
Zhou says he believes sharing resources is the next step in data center automation and will herald a new revolution in data center design. |
