Intel: How to Tackle A Major Consolidation

An overview of Intel's process in its massive data center consolidation, which will reduce 133 data centers to just eight.

Rich Miller

November 16, 2007

3 Min Read
DataCenterKnowledge logo in a gray background | DataCenterKnowledge

Planning and logistics are critical to the success of a major data consolidation. And consolidation projects don't get much bigger than the one currently underway at Intel (INTC), which is consolidating 133 data centers worldwide into just eight high-density facilities.

The process is driven by challenges in power and cooling in legacy data centers. More than 60 percent of Intel's data centers are at least 10 years old, with designs that make it difficult to capture the benefits of virtualization and energy-saving technologies. The consolidation process will take eight years to complete, and will overhaul many aspects of Intel's infrastructure.

"The transition is pretty major for all our business units," said Uttam Shetty, the Director of Data Center Efficiency at Intel. "All of the businesses (at Intel) have been open to these changes. It took a quarter or so to align the rest of the corporation, and there are some technical challenges within the design process. What we have done is show the benefits as we go to various business units."

Intel's task is made somewhat easier by the fact that it has space in existing facilities for the new data center space, so there's no new construction required. When the consolidation is completed, Intel will have added about 300,000 square feet of new space in those eight sites, which will be strategically located in the US, Europe and Asia. Intel is not identifying the location of their new facilities. It has existing state-of-the-art data centers in Oregon and New Mexico.


So how do you design a process for consolidating 133 data centers? "We had a significant amount of data we collected, and our team sat down together to analyze that data," said Shetty. "We went through an extensive selection criteria. We looked at existing infrastructure that is available, and where Intel business units are concentrated in locations across the world. Another criteria was how we could do this with the lowest use of energy."

Hardware Refreshes Drive Timeline
New hardware is a central component of the plan, and upgrade schedules are driving other decisions about what to do first. "As we consolidate the servers, we are aggressively moving to dual core and multicore machines with much more power," said Shetty. "Our strategy is that we will consolidate as the assets are ready to be refreshed. It reduces the cost of consolidation significantly. That's how we are managing our timeline."

The data center expansions will be built out using a modular structure, with each module designed to host 7,000 to 10,000 servers. "Some of our newer modules are designed with better airflow," said Shetty, who added that Intel is also doing a pilot installation to test the use of outside air in cooling the data center.

Intel is also segmenting its equipment into modules based on their mission. Intel found that its storage servers require a cooler environment to operate well, and will be separated into separate areas where the data center can be run at a lower temperature. That then allows the remainder of the data center space to run at a higher temperature set point, saving energy.

So what happens to the old data center facilities? "We look at it one data center at a time," said Shetty. "In some cases, we shut it down, in some cases, we handed over to corporate services. Some of them are leased facilities where we can evaluate the least."

Subscribe to the Data Center Knowledge Newsletter
Get analysis and expert insight on the latest in data center business and technology delivered to your inbox daily.

You May Also Like