This week the Uptime Institute announced the winners of the inaugural “Server Roundup Contest”. AOL, who came first, have retired and recycled nearly 9500 servers in less than six months. NBCUniversal were runner up with its infrastructure team having removed 284 servers or approximately 7.4% of the total enterprise. Only about 4800 physical servers remain in NBCUniversal’s infrastructure.
The contest was designed to highlight the energy savings available through the decommissioning of older servers. Participants were required to document the decommissioning of the machines and provide data on power savings and photos of the servers.
AOL decommissioned 9484 servers, representing 26% turnover in the company’s server assets, resulting in estimated cost savings of $5.05 million USD from reduced utility costs, maintenance and licensing costs, and $1.2 million USD in case raised from asset sales and reclamation. The environmental benefits were seen in the reduction of almost 20 tons of carbon emissions.
According to Uptime, decommissioning a single 1U rack server can result in $500 USD per year in energy savings, and additional $500 USD in operating system licenses, and $1500 USD in hardware maintenance costs. Matt Stansberry, Director of Content and Publications at the Uptime Institute, said “So much of the past several years’ efforts around data center efficiency has focused on the facilities infrastructure and PUE. This content was specifically designed to engage the IT community to get serious about asset utilisation and energy efficiency.”
Many organisations are also in a situation where they could realise major cost savings and improvement in environmental performance through the decommissioning of older and more energy hungry equipment. This is not only within the data centre but across the organisation including desktop computers, printers, monitors etc.
Crucial to the effective undertaking of such an activity will be a comprehensive and accurate configuration management system (CMS) supported by effective and efficient processes such as service asset and configuration management, change management, release and deployment management.
The accurate data will allow informed decision making in order to realise the return on investment from any refresh or decommissioning activity. The data will inform IT in regards to the utilisation of equipment, identification of unused or underused equipment, the energy efficiency of the equipment and its age. This can then drive a cost effective and profitable refresh programme.
Many organisations routinely decommission equipment as par of their everyday IT practices but this is often based purely on the age of equipment. Often a one-off audit is undertaken to identify equipment over a certain age and this is marked for decommission. An up-to-date CMS would negate the need for this one-off audit but would also allow the refresh to be driven by other criteria than just age. Equipment that could be replaced by more energy efficient machines can be identified and a cost benefit analysis undertaken. It may be more cost effective to replace this equipment before it’s designated end-of-life due to the energy efficiency of its replacement.
An up-to-date CMS will also allow identification and removal of redundant equipment in a timely manner. A supporting change management process and release and deployment process will ensure that equipment that will become unused or under utilised as the result of a change or release, can be removed from the environment. This keeps the IT environment lean and mean.
In 2009, following a large 2 year data centre consolidation exercise, Sun Microsystems recognised that despite those efforts, there was more than could be done and initiated a project called “Bring Out Your Dead” (inspired by the Monty Python’s Holy Grail film). This project was an effort to hunt down and remove orphaned and unused hardware at the company.
In just three months, at four of their major campuses they pulled out 440+ pallets of equipment, 6,199 devices in all with 4,100 of them being servers.
The best was yet to come. In their words: “the icing on the cake was that 64% of the equipment we pulled was still powered on! It was just sitting there burning energy”. The picture below shows 50% of the equipment that was removed which filled one of Sun’s warehouses.
This was like pulling 6000 cars off the road. This was a great win for sustainability and realisation of some massive cost savings but the situation should never have existed in the first place. It is accepted that reorganisations, mergers and acquisitions result in additional hardware assets but these should be soon identified through service asset and configuration management and removed from the environment before they have the adverse cost and environmental impact that we are all being asked to avoid.
The cost savings realised by AOL and NBC through participation in the Uptime competition should more than justify the investment in a CMS and supporting processes. Time to speak to the CIO.