Cloud spend optimization is always top of mind for public cloud users. It’s usually up there with Security, Governance, and Compliance – and now in 2020, 73% of respondents to Flexera’s State of the Cloud report said that ‘Optimize existing use of cloud (cost savings)’ was their #1 initiative this year.
So – what the heck does that mean? There are many ways to spin it, and while “cost optimization” is broadly applicable, the strategies and tactics to get there will vary widely based on your organization and the maturity of your cloud use.
Having this discussion within enterprises can be challenging, and perspectives change depending on who you talk to within an organization – FinOps? CloudOps? ITOps? DevOps?. And outside of operations, what about the Line of Business (LoB) or the Application owners? Maybe they don’t care about optimization in terms of cost but in terms of performance, so in reality optimization can mean something different to cloud owners and users based on your role and responsibility.
Ultimately though, there are a number of steps that are common no matter who you are. In order to facilitate this discussion and understand where enterprises are in their cloud cost optimization journey, we created a framework called the Cloud Cost Optimization Maturity Curve to identify these common steps.
Cloud Spend Optimization Maturity Curve
While cloud users could be doing any combination of these actions, this is a representation of actions you can take to control cloud spend in order of complexity. For example, Visibility in and of itself does not necessarily save you money but can help identify areas ripe for optimization based on data. And taking scaling actions on IaaS may or may not save you money, but may help you improve application performance through better resource allocation, scaling either up (more $$$) or down (less $$$).
Let’s dig into each in a little more detail:
- Visibility – visibility of all costs across clouds, accounts, and applications. This is cloud cost management 1.0, the ability to see cost data better through budgeting, chargeback, and showback.
- Schedule suspend – turn off idle resources like virtual machines, databases, scale groups, and container services when not being used, such as nights and weekends based on usage data. This is most common for non-production resources but can have a big bang in terms of savings – 65% savings is a good target that many ParkMyCloud customers achieve even during a free trial.
- Delete unused resources – this includes identifying orphaned resources and volumes and then deleting them. Even though you may not be using them, your cloud provider is still charging you for them.
- Sizing IaaS (non-production) – many enterprises overprovision their non-production resources and are using only 5-10% of the capacity of a given resource, meaning 90% is unused (really!) so by leveraging usage data you can get recommendations to resize those under utilized resources to save 50% or more.
- RI / Savings Plan Management – AWS, Azure, and Google provide the ability to pre-buy capacity and get discounts ranging from 20-60% based on your commitments in both spend and terms. While the savings make it worthwhile, this is not a simple process (though it’s improved with AWS’s savings plans) and requires a very good understanding of the services you will need 12-36 months out.
- Scaling IaaS (prod) – this requires collecting data and understanding both the infrastructure and application layers and taking sizing actions up or down to improve both performance and cost. Taking these actions on production resources requires strong communication between Operations and LoB.
- Optimizing PaaS – virtual machines, databases, and storage are all physical in nature and can be turned off and resized, but these top the maturity curve since many PaaS services have to be optimized in other ways like scaling the service up/down based on usage or rearchitecting parts of your application.
For more ways to reduce costs, check out the cloud waste checklist for 26 steps to take to eliminate wasted spend at a more granular level.