How to operationalize cloud cost usage optimization
Sustainably eliminate cloud cost waste in your infrastructure. Create confidence in your cloud efficiency.
Sustainably eliminate cloud cost waste in your infrastructure. Create confidence in your cloud efficiency.
Usage optimization focuses on controlling cloud costs by minimizing unnecessary or inefficient resource consumption. Think removing idle resources, rightsizing, and taking advantage of cost saving mechanisms (e.g. S3/EFS intelligent tiering in AWS). For a deeper dive into how this is difference from rate optimization, see here.
At many organizations usage optimization doesn’t get active attention and opportunities to save money go unresolved causing ongoing waste.
A typical cycle: The monthly AWS bill comes in, the Cloud FinOps Team is frustrated at spend increases, they look at Trusted Advisor or Compute Optimizer for savings recommendations, they ping the relevant application owner about an underutilized instance in Slack, the application owner may or may not act on that ping, the Cloud FinOps Team won’t have the bandwidth to manually track progress on the savings remediation, and the opportunity to save money often will go without action, incurring wasteful spend until the next fire drill.
There is a better way!
Operationalizing cloud cost management means going from a reactive process like the one described above, which lets wasteful resources linger for months at a time, to a proactive process whereby any wasteful spend is surfaced to the rightful owner on a weekly (or even daily!) cadence so they can decide whether to act immediately (if it’s easy) or incorporate the fix into their sprint planning.
In this case, waste management is an ongoing process that happens week over week. Easily solvable waste (e.g. idle instances, unused load balacners, etc) shouldn’t persist in your environment. Waste that’s more tedious to solve (e.g. rightsizing databse - downtime and performance considerations) may have to exist until bandwidth is created to investigate. Regardless, a process to surface and track waste is essential to have confidence in cost efficiency as you grow.
We’ll cover four steps to operationalize cloud cost usage optimization.
This is a prerequisite to knowing who’s responsible for a specific instance of wasteful resources. Some organizations attribute with resource tags, other organizations attribute with accounts, regardless of how you attribute infrastructure you need to create a map of ownership and decide who the responsible person for each application / business unit is (we typically refer to this person as the Distributed Cost Owner).
In Cloudthread you can create Teams, attribute infrastructure based on any billing dimension, and then when users from that team login they’ll see their segment of attributed infrastructure.
Before you can operationalize usage optimization, you need to do an initial cleanup of your infrastructure and eliminate all of your high priority savings opportunities. Depending on the size and complexity of your infrastructure, this can take days or it can take quarters.
Map out what Savings Opportunities exist in your environment and create a phased plan to address the egregious waste offenders.
In Cloudthread, if there’s something that shows up as waste but is deliberate (e.g. an intentionally overprovisioned databse) you can “Archive” the savings opportunity so that it won’t appear as part of your waste analytics. You want your “Active Savings / Mo” to trend towards $0.
When you’ve finished your initial cleanup of infrastructure, waste in your environment should be at (or close to) zero, and it’s time to set up Savings Reports so that any waste that gets created week over week is surfaced to the appropriate owner.
When doing this it’s important to set the minimum threshold for being included in a report. For some companies this is $30 / month, for other companies it’s $1,000 / month.
From Cloudthread you can create a Savings Report, filter it by a Team, and then send it to users via email or channels via Slack.
Application cloud cost owners get a weekly report of any new waste that’s developed in the last week sorted by difficulty and can click into any opportunities to evaluate them easily.
If you want Engineering Teams to prioritize cloud cost savings, you need to create incentives. A previous article I posted goes into detail on how you can use leaderboards and gamification to create incentives - check it out here.
Examples of incentives you can provide:
Don’t let enforcement of savings opportunities fade away in slack threads or in google sheets. An essential part of the ongoing process is having a source of truth that’s referenced across the organization to showcase waste that exists and savings that have been achieved.
For application owners this provides a source of truth to validate cost saving measures taken.
For Cloud FinOps Teams this is referenced to ensure Teams are effectively achieving the organization’s cloud cost waste management goals.
The Cloudthread platform is designed to help organizations achieve distributed cloud cost ownership and conquer usage optimization. We’ve served growth stage organizations and Fortune 500 companies seeking to create a culture of distributed cloud cost ownership.
Cloudthread’s platform scans for a list of 50+ savings opportunities and helps tems operationalize turning those opportunities into actualized savings.
If you’re trying to achieve distributed cloud cost ownership at your organization, contact us.
