By Stefana Muller, 2nd Watch
The main reason companies delay and/or avoid confronting issues related to cloud cost optimization is because it’s incredibly complex. Challenges range from cloud sprawl to misaligned priorities. Understanding these challenges before you begin is essential to success. Here are the main issues:
- Cloud sprawl — This refers to the unrestricted, unregulated creation and use of cloud resources; cloud cost sprawl, therefore, refers to the costs related to the use of each and every cloud resource (i.e., storage, instances, data transfer, etc.). This typically presents as decentralized account or subscription management.
- Billing complexity — This is about the ever-changing, variable billing practices of cloud providers and the invoices they provide. AWS alone has 500,000 plus SKUs you could see on any single invoice. If you cannot make sense of your bill up front, your cost optimization efforts will languish.
- Lack of access to data — This is one of the biggest barriers to cloud cost optimization. Without billing data and application metrics over time, many incorrect assumptions will be made, resulting in higher costs.
- Misaligned policies and methods — This obstacle can make or break your optimization project. When every team, organization or department has its own method of managing cloud resources, the solution is more about organizational change and less about technology implementation. This problem can be tricky to solve, especially if the teams aren’t on the same page with needing to optimize.
- Lack of incentives — You might think everyone would be on the same page when it comes to saving money, but lack of clarity on incentives actually is the number one impediment to achieving cloud cost optimization. IT is laser-focused on cost management, while business units tend to be more focused on speed and innovation. Both are important, but without universally accepted incentives, your project is doomed to failure.
Our research indicates companies can achieve up to 70 percent savings on cloud through a combination of software and services. It often starts by implementing a solid cost optimization methodology. Here’s an approach that has worked for our clients:
Step 1 — Scope It Out!
As with any project, first identify the goals and scope and then uncover the current state environment. Here are a few questions to ask to scope out your work:
- Overall project goal — Are you focused on cost savings, workload optimization, uptime, performance or a combination of these factors?
- Budget — Do you want to sync to a fiscal budget? What is the cycle? What budget do you have for up-front payments? Do you budget at an account level or organization level?
- Current state — What number of instances and accounts do you have? What types of agreements do you have with your cloud provider(s)?
- Growth — Do you grow seasonally, or do you have planned growth based on projects? Do you anticipate existing workloads to grow or shrink overtime?
- Measurement — How do you currently view your cloud bill? Do you have detailed billing enabled? Do you have performance metrics over time for your applications?
- Support — Do you have owners for each application? Are people available to assess each app? Are you able to shut down apps during off hours? Do you have the resources to modernize applications?
Step 2 — Get Your Organization Excited
One of the big barriers to true optimization is gaining access to data. In order to gather the data (Step 3), you first need to get the team on board to grant you or the optimization project team access to the information. During this step, get your cross-functional team excited about the project, share the goals and current state info you gathered in the previous step and present your strategy to all your stakeholders. Stakeholders may include application owners, cloud account owners, IT Ops, IT security and/or developers who will have to make changes to applications.
Remember, data is key here, so find the people who own the data. Those who are monitoring applications or own the accounts are the typical stakeholders to involve. Then share with them the goals and bring them along this journey.
Step 3 — Gather Your Data
Generally speaking, data is grouped into a few buckets:
- Billing data — provides a clear view of your cloud bill over time
- Metrics data — CPU, I/O, bandwidth and memory for each application
- Application data — conduct interviews of application owners to understand the nuances. Graph out risk tolerance, growth potential, budget constraints and identify the current tagging strategy.
A month’s worth of data is good; three months is much better to understand the capacity variances for applications and how to project into the future.
Step 4 — Visualize And Assess Your Usage
This step takes a bit of skill. There are tools like CloudHealth that can help you understand your cloud cost and usage. Then there are tools that can help you understand your application performance over time. Using and correlating data from each of these sources is essential to understanding where to find savings opportunities.
I often recommend bringing in an optimization expert for this step. Someone with a data science, cloud and accounting background can help you visualize data and find the best options for optimization.
Step 5 — Plan Your Remediation Efforts And Get To Work!
Now that you know where you can save, take that information and build a remediation plan. For example, you might decide to shut down resources at night for an application and move it to another family of instances/VMs based on current pricing. Or you might implement a Reserved Instances purchase strategy. Remediation can take anywhere from one month to a year based on company size and the support of application teams.
Though complex, cloud cost optimization is an achievable goal. By addressing project risks like lack of awareness, decentralized account management, limited access to data and metrics, and lack of clear goals, your team can quickly achieve savings. With as much as 70 percent savings possible after implementing one of these projects, there’s a compelling reason to get started.
About The Author
Stefana Muller is Senior Product Manager at 2nd Watch.