This article is more than 1 year old

Uncle Sam has a datacenter waste problem

Abysmal server utilization and other problems uncovered

US government auditors want to save taxpayers' money by bolstering the capability and efficiency of Uncle Sam's far-flung stable of datacenters. Each federal agency's sites have a host of problems, unsurprisingly.

The US Government Accountability Office (GAO) kicked off its Datacenter Optimization Initiative (DCOI) in 2016 with the goal of reducing the massive financial and energy waste being generated by these warehouses of computers. We're not talking small change, either: in 2007, the Environmental Protection Agency estimated the annual electricity bill for federal government datacenters and servers was $450 million, and the Dept of Energy said each facility can use 100 to 200 times as much energy as a commercial building. 

Despite all that energy and monetary expenditure, two years later the US Office of Management and Budget (OMB) found that government server utilization rates were as low as five percent.

"The high cost and low utilization led OMB to establish a coordinated, government-wide effort to improve the efficiency, performance, and environmental footprint of federal datacenter activities," the GAO said.

The GAO's latest report [PDF] on the state of DCOI progress indicates "mixed progress" at best. Of the 24 agencies that fall under the DCOI's purview, only half met virtualization metrics, 16 met availability requirements, half were properly monitoring energy consumption, and 13 were properly reporting underutilized servers. 

To make matters worse, the OMB reports that the number of federal government datacenters has actually increased since the DCOI was enacted, growing from 5,607 in 2016 to 5,916 by 2018.

All of this has led to closing datacenters completely: 96 were shuttered in fiscal year 2020, 51 had already happened when the report was written and 29 more were in the planning phase.

So, why did US government agencies not meet their savings goals in 2021? The reasons vary, and some indicate that the GAO's measurement methods may not be accounting for all potential paths to savings.

The Nuclear Regulatory Commission, for example, didn't meet its virtualization goals because it implemented a higher density of virtual servers on smaller, higher-performance hardware. This means it fell short in 2021, but in 2022 "targets have been adjusted to align its new infrastructure with the metric."

The Small Business Administration wasn't able to meet its virtualization goals due to technical gremlins – it couldn't fit virtualized servers on existing hardware; and budgetary holdups – it had no money to move servers to the cloud. Its plans for the future involve applying for a grant and hoping for modernization funds.

Heath and Human Services was unable to meet its energy monitoring goals, and that may not have even been their direct fault: "it did not meet advance energy metering due to a change in the number of valid datacenters counted to establish its energy metering target," the report said. 

Despite shortcomings in meeting particular metrics, the overall intent of the DCOI, to reduce datacenter costs, is being fulfilled to a degree. In FY 2020, the GAO reported that 14 agencies exceeded their cost-savings goals, four met their goals, and six met their goals on a technicality (they were aiming for zero-dollar savings and got there).

The data from 2021 so far isn't as cut and dry, though there's still hope to be found. Eight agencies exceeded their savings goals by August, and eight had and met the same zero-dollar savings goal as in 2020. Seven agencies said they hadn't met their goals as of August, but still planned to save a total of $267.92 million. ®

More about

TIP US OFF

Send us news


Other stories you might like