The Cloud Cost Management Dilemma
Launching new infrastructure in the cloud is easier when compared to the traditional path of purchasing hardware, imaging that server, connecting to the network, getting security policies setup, IP address assignments and it’s subsequent addition into various inventory management systems.
With the unlimited scale we see some of the same issues of managing our own infrastructure, namely under-utilization and cost overruns.
You can overbuy server instances and while most apps autoscale up nicely, they don’t always give computing resources back to the public cloud when you no longer need extra compute.
Do you Use Any of these methods to manage cost?
- the public cloud interface
- an elaborate spreadsheet or
- your monthly bill from the cloud vendor
If you’ve answered yes to any of these questions, chances are that you are drastically overspending.
For workloads that are between $10,000 and $40,000 monthly, we usually see a 15-25% overspend. Once an environment exceeds $50,000 per month we can sometimes find up to 40% in overspend. It’s akin to running your business without accounting software.
This overspend is usually a result of:
- Server Instance Families that are greater than the workload
- Not using Reserved Instances in certain aspects
- Billing anomalies that aren’t caught until the end of the month
- Orphaned server or storage instances that are idle but being billed.
If you are managing your public cloud spend manually, you are missing out on massive cost saving opportunities. You can also check out our ever evolving What is Cloud Computing reference section which outlines the challenges behind cloud and the history of cloud.