AWS Savings Plans are a big deal, and they’re going to have a big impact on how companies optimize AWS costs. To help figure out how best to tackle this brave new world, the experts over at the FinOps Foundation put their heads together. In a call with over 100 experts from around the world, they talked about the possible advantages of Savings Plans and potential pitfalls — including taking a survey.
A few key data points stood put from the survey:
In the end, there was a general consensus to embrace Savings Plans, but only in a strategic way that would allow them to smoothly integrate into existing (and effective!) FinOps practices.
At Apptio, we couldn’t agree more. Keeping these lessons from the FinOps experts in mind, here are the four key parts essential for any Savings Plans strategy.
Currently, AWS has no plans to retire RIs any time in the near future. And if they do sunset RIs, there’s a multi-year transition period of at least three years to fulfill RI agreements. Even then, Savings Plans (as of now) can only cover compute resources. RIs are still necessary for RDS, DynamoDB, Elasticache, and Redshift. There are also plenty of other reasons why companies might choose to continue purchasing RIs in addition to Savings Plans, such as RI capitalization or lower agility in systems.
Regardless of the reason, using both means you need to have tooling and processes in place to track both RIs and Savings Plans are the same time. Otherwise, you risk not having an accurate picture of your actual cloud spend. Complicating this is the fact that Savings Plans and RIs show up differently in the CUR file. With Apptio Cloudability, we solved this by incorporating the Savings Plans line items into the powerful analytics engine behind our platform. Using tools like Cost Explorer, reports, and more, you can analyze both RIs and Savings Plans in a single pane of glass, enabling you to get the most savings out of your investments.
One of the biggest concerns mentioned in the FinOps Foundation article was ensuring that Savings Plans are correctly allocated to the right teams. And it’s a valid concern. Say your production team has put a lot of time and effort into rightsizing compute instances, then purchase Savings Plans to cover 90% of their resources. But then an R&D team spins up a couple hundred x1 instances that use up all of the Savings Plan compute prices for a day. Suddenly, the resources for the production team are running at On-Demand rates and spike dramatically. You need to be able to track this down, identify the cause of the spending spike, and correctly allocate savings where necessary.
Part of this can be done at the account level. By default, Savings Plans are utilized across all Consolidate Accounts, but you can set it up to only be used by the purchasing account. But beyond that, it really comes down to correctly allocating the savings on your terms to fit your organizational structure. Take the example above. In the end, you’re still getting compute resources at the Savings Plan discounted rate, and your overall compute spend won’t appreciably change depending on whether they’re applied to R&D or production. It really comes down to proper accounting and allocation. An analytics engine and mapping features like in Apptio Cloudability will allow you to tie those savings to the right teams.
Take the time to ensure you have the right processes and tools set up for allocation before you go too deep into Savings Plans. It’ll save you a lot of headaches in the long run.
Obviously, there are Savings Plan limits, like not applying to RDS or other services as mentioned above. But knowing how each is applied goes beyond that. Unless you’re one of a very small minority, you still have RIs you’ve purchased or committed to applying to your system. You need to be able to see the precise point where your RI coverage stops so you know where to purchase Savings Plans.
This it a little more complicated than just taking your non-RI spend and buying that many Savings Plans. Your infrastructure is fluid, which raises the pitfall of over-buying Savings Plans and having compute use drop below the threshold. This is where the tried-and-true waterline approach from RIs can be applied Savings Plans. Using Apptio Cloudability, you can view your total compute usage to define a savings watermark, such as at 90% coverage. Overlay that with the amount of compute usage currently covered by RIs. The delta between the RI coverage and waterline is the amount you want covered by Savings Plans. As your RIs expire, you can fill increase the Savings Plans coverage.
As always, the goal is to set a savings waterline, then hit it without your usage dropping below the waterline into waste.
This last point is crucial for any early adoption. AWS Savings Plans are still a very new product, which means there’s no tried-and-tested best practices specifically for their use. In fact, it’s probably going to be another six months to a year before those are fully developed. But you shouldn’t let that stop you from utilizing Savings Plans. Instead, adopt an experimental mindset. Try out Savings Plans in smaller purchases, then purchase more strategically as your ability and certainty with them grows. A good place to start is with the core concepts that applied to RIs. Just like RIs, you should look at the make-break point of Savings Plans, especially when looking at Partial Upfront or All Upfront options. You should also embrace the waterline model, albeit with compute or EC2 spend per hour instead of instance usage hours.
It’s an exciting new world in AWS cost optimization. Embrace it with a strategic, experimental mindset to get the most out of your AWS cloud.
Apptio Cloudability is here to help you craft and execute your Savings Plans strategy. Check out this video to find out more about our approach to Savings Plans — and our plans to fully support Savings Plans in our platform.
Ready to take the next step on your Savings Plans journey?
Contact us for a demo and find out how to get the most from your AWS Savings Plans.