Avatar

Last week I introduced this topic, the pervasive problem of “comatose” servers in data centers, based upon an interesting recent eWeek article entitled “30 Percent of Servers Worldwide Sit Idle”, which in turn was based upon the research report by Stanford University in conjunction with the Anthesis Group.  In my blog, I described the costs of this problem, ranging from the obvious (e.g. power and facilities) to the hidden (e.g. un-used software licenses).  This week I’ll discuss why this happens and what you can do about this problem.

The Problem of Idle Servers - What Can You Do About It
The Problem of Idle Servers – What Can You Do About It

Why Does this Happen?

There are various reasons.  It is usually management priorities, lack of process (including lack of diligent inventory management) and lack of incentives that results in such IT waste. Let’s face it, we are all often too busy with developing next year’s plan and executing on priority initiates for today, that we don’t get time to do all those “tidy up” actions we know should be done … at some point.  And it’s fair to say that “Nobody gets promoted for going around in the data center and unplugging servers”!!

Additionally, I’m sure we’ve all be in projects that were cancelled at some time, sometimes with hardware CAPEX that isn’t successfully re-purposed.  Sometimes the servers – bought with one use case in mind – are not up to the job required in the next project – and hence lie idle.  Pre-Cisco UCS servers, old generation architectures, have been side-lined in the rush to virtualization, and haven’t been decommissioned. And sometimes it is the server being just plain “old”. The continued “idle” state, at the end of the day, is not usually down to the server technology.

Mergers and acquisitions are also a major source of idle servers.  Mergers and acquisitions inevitably result in consolidation projects, with the old servers being left in-situ … and then forgotten about.  This reflects on a management issue, that of poor inventory tracking: even today, despite the advances in adoption of CMDBs (Configuration Management Database), data center inventory tracking in many organizations is still challenged.

In summary, there are many reasons why servers can simply become “lost”!  They remain housed in the data center, inevitably switched on, consuming power, and hosting licensed software apps that are no longer used.

Effecting Change: Identifying and Removing Idle Servers

Hopefully by now I’ve convinced you that idle servers are a genuine problem affecting many if not all data centers across the world in varying degrees.  You could make significant savings if you can drive change in your organization to tackle this problem.  Barclays, a UK financial services company, for example, eliminated 9,000 servers back in 2013 and the US Uptime Institute has an annual competition, the “Server Roundup Competition” dedicated to identifying and eliminating comatose servers!

In our experience, the most sure-fire approach to solving this problem is to exploit advanced discovery tools to find these idle servers and un-used software licenses in your data center. It’s not just about tools though. You also need the (i) skills and (ii) time (!) – to enable you to exploit these toolsets.   Let’s face it – it’s lack of time in many cases that has resulted in these servers going missing!     Even finding previous owners and establishing the access credentials for “missing” servers can be a time consuming task once you uncover them.  I am willing to bet that you have way too many projects already filling up your days, evenings and weekends. How then can you solve this “catch 22”?

This is where Cisco Services can help you.  With services such as the Cisco Application Dependency Mapping Service (which I blogged on here back in May), and the advanced network, server and application discovery technologies we use as part of this service, our consultants can take the weight off you.  We’ll find those idle servers and the unused but paid for software licenses.  We’ll do a lot more, and help you understand the application-server-network architecture issues that are impacting your end user performance.  We’ll not only help you save power costs, floor space and software licensing costs, we’ll help you deliver better application performance and network utilization.  Contact me on Twitter if you need to start saving!



Authors

Stephen Speirs

SP Product Management

Cisco Customer Experience (CX)