You May Be Wasting 90% of Your Cloud Bill
Cost optimization (read cost reduction) is one of the biggest promises of Cloud, in close competition with the promise of agility (as achieved by flexibility, scalability, simplicity and speed). We see numbers reported every month and every quarter that an increasing number of organizations are being serious about their cloud strategy.
However, as more and more companies adopt the Cloud, we are also realizing that we need to be careful about our Cloud usage. If unmanaged, Cloud costs can spiral up very fast. I equate on-premises infrastructure to a debit card (you can only spend what you already have) and Cloud infrastructure to a credit card (the billing statement at the end of the month can hit you like a train). People have started getting a fair understanding of the challenges involved. Not only are organizations trying to adopt a systematic and planned approach to leverage Cloud potential, but Cloud providers themselves are continuously enhancing their platforms to provide you better control and visibility. In addition, several ISVs are creating tools to help you better manage your Cloud infrastructure.
I have the good fortune of experiencing this space very closely and have come to realize that the savings we currently achieve from Cloud is not even the tip of the iceberg (in a positive sense). This is especially true for non-production workloads. I believe that, currently, we may be wasting as much as 90% of our Cloud cost.
Let me illustrate this with an example:
Last month, one of our customers asked us to implement Enterprise Single Sign-on between their SaaS application (an education technology-based product) and their SaaS instance of Canvas (a learning management system from Instructure). We started the project with a proof of concept, which shouldn’t have taken more than few days. The team requested for a VM on our Azure account for a period of four weeks (contingency planning, you know!).
Our Azure administrator (yes, we control it via a single owner) promptly serviced the request. A Standard_A2 VM instance was created on 22nd May. The POC was completed and demoed to client on 5th June. As of yesterday, the VM was still running. And its CPU utilization never crossed double digits.
Let’s do some cost analysis for this small situation:
Cost of Standard A2 Windows VM on Azure for 1 month
(minus) Cost of VM provisioned beyond requirement:
Out of 33 days, the VM was required only for 14 days. 19 days extra cost meant wastage of (19 days / 33 days) = 57.58% of USD 134.00. Actual cost could have been USD 56.85.
(minus) Cost of VM running when not in use:
The machine was being used for a maximum of 40 hrs. a week, instead of full 168 hrs. This 128 hrs. per week meant extra cost of (128 hrs / 168 hrs) = 76.19% of USD 56.85. Actual cost could have been further reduced to USD 13.54.
(minus) Cost of unnecessary higher configuration:
A Basic A0 machine, costing just USD 13 per month, was sufficient to service the requirement. Standard A2 machine costed additional USD 121 per month. This meant extra cost of (USD 121 / USD 134) = 90.30% of USD 13.54. Cost corresponding to the real business value of the assignment was just USD 1.31.
Wastage of USD 132.69, a Whopping 99.02%!!!
We can argue that the project team was irresponsible or the Azure administrator was incompetent. That’s not the point.
The point is, such leakages are happening in almost all organizations. And these leakages are controllable to a great extent. Maybe not all of them. But, we can (and should) start pushing Cloud to such extremes. Right processes and right technology will follow.
Thanks for sparing your time to read this. I want to leave you with a question – In your experience and opinion, what is coming in the way of organizations in achieving such extreme level of cost optimization?
Image Credit: www.freephotosbank.com