You get what you measure – but how do you measure a cloud?

The world loves a business case. It makes investment decisions clearer, and also congeals in our collective minds what we’re actually shooting for with a project. It’s important to understand your goals and metrics – and thus make your decisions based on the outcomes you’re anticipating, or even banking on.

 

But in cloud – what SHOULD be your metrics? The following are a few options – and measuring them depends on having a ruler by which they can be tracked. I couldn’t possibly advocate for choosing “All”  — that invokes the age old “Fast, Cheap, Good – Pick 2” adage… but it’s nice to have a selection from which to .. well.. select.

ruler.JPG

OpEx and CapEx — The classic metric. In theory, particularly if you’re moving a great deal to the public cloud, the cost of buying and maintaining physical infrastructure will go down, and the monthly costs of renting public cloud will go up. If your company is interested in motivating this move, tracking the baseline CapEx – and how it drops and OpEx rises – is a good proxy for understanding the success of your policies.

 

Time to Provision – This may be the most quoted metric in cloud. Everyone has a “before, it took 3 weeks/4 days/7 months to get a server” story, and in a private or hybrid cloud, that drops to hours or minutes. It’s a fabulously impactful metric but requires a touch of refinement. First, there are plenty of companies with very fuzzy ideas on the “before” metric. Second, it begs the question “what is a service?”

 

You want to compare apples to apples. If you had to wait 3 weeks to get a completely configured application before vs. 2 hours to get a bare copy of Linux afterwards – that’s not a fair comparison. Defining success here will help set the requirements for your provisioning system.

 

Users Served – I have rarely seen this metric used, but I love it. In theory, as your cloud grows and develops, it addresses the needs of more and more of the users of IT in your company. So, it would be ideal to see a cloud growing from 2% to 35% of users served. Certainly, not everyone in the company will have need for some cloud – but understand what number will, and seek to serve them through your platform.

 

Uptime Metrics – These are classic, and required in any production platform. There are dozens to choose from, and perhaps the best ones are ones that account for the dynamic nature of the cloud, as well as providing some continuity with the non-cloud part of the infrastructure. In an early cloud, it may be acceptable to have lower uptime numbers – if you’re trading agility for quality. But, as always, you have to know what you’re trying to accomplish.

 

The Elusive Agility Metric – The biggest stated goal of clouds the world over is to increase agility – usually business agility. So, similarly, the holy grail of metrics would be one that tracked how much faster a product could get to market, how much more effective a campaign could be, or how many more visitors a web site could support – on the back of this new technology.  Capturing the “before” state of this is often challenging, and requires a somewhat harsh mirror to the organization’s current state. Capturing the “after” state requires a lot of support from the business units involved.  Still, success would be incredible – and well worth it.

 

What metrics do you use to measure your cloud?

These postings are my own and do not necessarily represent BMC's position, strategies, or opinion.

Share This Post