Our latest research on comatose servers
I’ve been working with Jon Taylor of Anthesis Group and Aaron Rallo of TSO Logic to compile data on servers in enterprises that are using electricity but generating no useful computing output (we call these comatose servers). Until now, it has been difficult to compile data on idle servers over the network, but recent developments in measurement of server utilization and network data flows allow us finally to identify these servers in an automated way.
The Uptime Institute and McKinsey and Company had earlier estimated that up to 30% of servers in many data centers were comatose, and new data from TSO Logic confirms these estimates. Our initial sample size is small (4000 servers) but the data show that 30% of the servers in this sample hadn’t been used in more than six months.
If this finding holds up for larger sample sizes (and we expect it will) then about 10 million servers in the world are comatose, stranding tens of billions of dollars of data center capital and wasting billions every year in operating cost and software license costs.
In the twenty first century, every company is an IT company, but too many enterprises settle for vast inefficiencies in their IT infrastructure. The existence of so many comatose servers is a clear indication that the ways IT resources in enterprises are designed, built, provisioned, and operated need to change. The needed changes are not primarily technical, but revolve instead around management practices, information flows, and incentives. To learn how to implement such changes, see my Fall 2015 online class titled Management Essentials for Transforming Enterprise IT.
We will update the analysis as the data set grows, with the next update due in Fall 2015.
Updates
Forbes posted a nice summary of our work, giving some important context.
TSO logic did a blog post with more information.
Data Center Knowledge posted an article summarizing the management implications of our findings.
Computer Business Review did a summary article on our work.
Silicon Angle, a technology business publication, wrote a summary June 5, 2015.
Tech Republic summarized our findings and brought in other related efforts.
eWeek (June 15, 2015) summarized our work.
I had a nice chat with Patrick Thibodeau of Computer World, who wrote it all up here on June 19, 2015.
Useful discussion here from readers of The Hacker News.com
Information Week weighed in on WHY such inefficiencies persist in data centers after all these years, using our study as a jumping off point.
CIO magazine also summarized the research on August 17th, 2015, and explained what you can do about it in your data center.
