In the previous post we outlined and defined the three categories of cost management:
- Quick Wins
- Cost Optimisation
In this post we'll take a closer look at 'Quick Wins', and outline some of the different categories of savings you might make.
We focus here on AWS services, but similar techniques and services exist for all the main cloud service providers (CSPs).
Quick Wins Defined
A cost optimisation 'quick win' can mean many things to many different people, but at Container Solutions we define it as:
- Requiring little or no knowledge of the business or technology to identify, and
- Little or no disruption to the business or technology to implement
The first item means that any consultant like us can suggest these things to any cloud-consuming business without knowing much about them. Suggesting these quick wins, therefore, can unfortunately result in a bit of eye-rolling from the client, as many will tell us that they have of course thought of these 'obvious' cost savings themselves.
However, we find it is still worth going through our list of 'quick wins', as some may have been missed or misunderstood. In any case, since they are relatively cheap and easy to implement it's worth a little awkwardness just to make sure they've been covered off.
Quick Win Categories
Within the 'Quick Win' category of cost management, we have identified these classes of optimisations:
- Resource Review
- Resource Removal
- Resource Usage
- Finops Foundations
- Service Level Optimisation
These techniques are the most basic and well-known steps to cloud cost optimisation, since they are those that are offered by the cloud service providers themselves.
To take just AWS as an example, you can use:
- AWS Trusted Advisor to generate advice on cost optimisation
- AWS Cost Explorer to generate reports on consumption history and forecasting
- AWS CloudWatch to examine which resources might be being under-utilised
Using these tools effectively will give you a quick insight into the client's resource usage, and quite possibly point towards some ways in which money could be saved.
One of these ways could be simply to remove unused resources. This might sound obvious, but it's surprising how frequently resources are left unused by businesses to accumulate unnecessary cost. Either people aren't aware of their existence, or are unsure who or what 'owns' them, and therefore are unwilling to delete them in case doing so might cause disruption.
But it's not just 'forgotten' resources that can incur costs. There are other common patterns of usage which might result in superfluous resources being allocated. For example, EBS snapshots allow you to restore the state of a disk at the snapshotted point in time. These snapshots are often observed to be left untouched for longer than they are needed as 'backups'. There can also be confusion about these 'incremental' snapshots: you can delete any incremental snapshot you don't need without affecting any other restoration point.
Another very common candidate for resource removal springs from the separate lifecycles that virtual machines have with EBS volumes: if you don't tick the 'delete on termination' option when adding an EBS volume to a EC2 instance, then the volume will persist after the EC2 instance has been terminated.
There is even a how-to on Amazon's blog to help you set up a Lambda function to check for unused EBS volumes.
Another way to save money without interrupting business might be to change the way the resources are used. The most well-known of these is to switch off non-production machines outside office hours, and there are various scripts and services that can help you automate that. However, you can go further: you can use AWS Compute Optimizer to seek savings based on machine learning analysis of your patterns of usage.
Another angle to consider is moving to newer resource types. On AWS, Graviton instances offer significant cost savings across many services if your workloads can be moved to them.
Without knowing anything about a business's operations or cloud spend, there are still techniques you can apply which lay the foundation for more thoroughgoing change associated with the 'FinOps' stage of cost optimisation.
- Setting up budgets and cost alerts
- Setting up service metrics
- Tagging resources based on organisational boundaries
Probably the least disruptive way to save money with a CSP is to promise to use their services beyond the standard 'pay as you go' model.
There are many ways to use commercial agreements with CSPs to reduce your overall cost. Commitment-based discounts allow you to pay for 'reserved instances' (ie compute that you promise to use for a given period) over a year or multi-year terms. There also exist 'sustained use' discounts which apply to resources used against a committed level of spend. Similar options exist for storage in addition to compute.
Finally, if you are on an Enterprise support plan with AWS you can get consultancy from AWS directly on your cost management.
A Bewildering Menu
As with all things cloud, the possibilities and choices available to you around cost optimisations can be bewildering. Even restricting discussion to 'Quick Wins' that won't disrupt the existing technology or business still requires a significant amount of work to filter out the most effective avenues of improvement.
Container Solutions' Cost Optimisation Framework gives customers a quick and effective way to discover which tools, techniques and products are most appropriate for them, and how they can save money over the long term as their business evolves. Contact us to find out more.