If you’re like most enterprise IT departments, chances are high you’re sitting on an abundance of excess capacity in your server estate. There are all sorts of valid reasons for this to happen – mergers and acquisitions, requests for additional server capacity that far exceed any notifications that capacity is no longer needed – the list goes on.
Back in the not-too-distant past, this wasn’t a big problem; but escalating power, cooling and space costs have pushed rightsizing and shrinking this excess capacity to the top of most enterprise IT agendas.
But rightsizing can be a risky undertaking when the performance of your business depends on those servers. How close to the bone do you dare go? IT managers charged with rightsizing struggle to balance the need to drive saving costs with ensuring there is enough spare capacity to cope with often unexpected and fluctuating demand.
As a result, most companies don’t rightsize as aggressively as they could – maintaining extra headroom “just in case”.
What if you knew exactly how much capacity you were going to need?
Conventional capacity planning tools do a good job of showing you how much capacity is currently being used, and how much was used in the past. But, in today’s complex and dynamic IT environments, when it comes to figuring out exactly how much capacity you really need for the future, it’s not so easy to be precise.
Predictive analytics changes everything.
Predictive IT analytics can help you forecast future capacity requirements with accuracy – giving you the confidence to really optimize. You can try modeling different scenarios based on the level of risk you’re comfortable with and then decide which one to go with.
It’s amazing how much more aggressive enterprises can be shrinking their IT estate if they apply predictive analytics.
Here’s a real example of how it works.
Predictive analytics helps a global bank shrink part of its server estate by 75%
A leading Tier 1 global bank believed its IT capacity far exceeded its current and future needs and wanted to look at options for rightsizing its global server estate – an estate of many thousands of servers.
The bank wanted to come up with a detailed, validated, actionable plan to start right sizing. They called in Sumerian and our Forward Thinking analytics. We started a program of analytics, taking each technology platform in turn.
Our first step was to work with the bank to understand the relevant application architectures and constraints, so that a clear set of migration rules could be established – for example, which applications could co-exist and which could not. Once these rules had been defined, we ran a baseline modeling exercise on the first technology platform in the UK, building up an accurate profile of current server hardware and resource utilization. Detail captured included the age of servers, their depreciation cycle, power consumption, CPU, memory and disk I/O.
Once the baseline was established, we used our Forward Thinking predictive analytics to model several potential consolidation scenarios. Depending on the age of servers, where they were in terms of depreciation, and application end-of-life plans, we modeled migration across existing and new hardware.
The output was an actionable migration plan combining technology refresh at a hardware level and re-balancing of server workloads. As a result the bank could reduce the number of physical servers deployed to support this platform by 75%.
And this is just the start of their consolidation journey – one that will shrink their estate far more aggressively than they would have had the confidence to aim for without the quantified and auditable evidence provided by predictive IT analytics.
To find out more about how predictive analytics can help you to shrink your IT read our white paper on Datacenter Consolidation.