Reduced Kubernetes Cloud Costs by Millions in 5 Months
Relativity makes software to help users organize data, discover the truth, and act on it. The platform is used by thousands of organizations around the world to manage large volumes of data and quickly identify key issues during litigation, internal investigations, and compliance operations with RelativityOne and its newest offering Relativity Trace. Relativity has over 180,000 users in 40+ countries from organizations including the U.S. Department of Justice, more than 70 Fortune 100 companies, and 198 of the Am Law 200.
Accelerating Deployment Velocity & Reducing Cloud Spend
Earlier this year Relativity kicked off an internal initiative to increase deployment velocity and reduce cloud spend as part of their cloud migration to Windows Azure.
Previously, finance initiated cost savings exercises directly with product managers and engineering teams. Cloud costs were ‘bubbled’ together into a high-level aggregate view across teams, which made it challenging for an individual product or engineering team to understand their costs.
This lack of visibility into cloud costs by application or microservice meant that engineers struggled to identify the real utilization and idle cost of the resources they were requesting in the Azure cloud, specifically relating to their Kubernetes clusters.
“We were seeing over-provisioning of 30-40% of our microservices and Kubernetes Clusters,” said Corey Wagehoft, Lead Systems Engineer at Relativity.
Cost Reductions in the First Month
Any engineering team that’s deploying Kubernetes and not proactively thinking about managing cost will fail to unlick its true benefits.Corey Wagehoft | Lead Systems Engineer | Relativity
Relativity participated in the early beta of Harness Cloud Cost Management — a new cloud cost management solution that empowers engineering and DevOps teams with unique cloud cost visibility so they can proactively manage cloud spend.
“During the first 30 days of implementation, we saw a noticeable change in our cloud spend across our engineering teams, with six-figure annualized savings,” said Shelby Lewin, Technical Product Manager at Relativity. “Not only did we save cost, but our engineering team also sped up two months of work on our roadmap for RelativityOne.
During the first 30 days of implementation, we saw a noticeable change in our cloud spend across our engineering teams, with six-figure annualized savings.Shelby Lewin | Technical Product Manager | Relativity
With Harness Cloud Cost Management, Relativity was able to reduce cloud spend by 30-40% for a new microservice in 30-days which represents a 6-figure annualized savings.
The biggest win for Corey and Shelby was unpacking the true cost of their SaaS tenants and customers running on the Relativity platform.
Here’s an example of one microservice at Relativity that was optimized after identifying high idle and unallocated costs within a Kubernetes cluster:
5 Months Later…
Fast forward five months, and Relativity has saved millions using Harness Cloud Cost Management.
Product and engineering teams at Relativity are now proactive with self-service access to their own cloud spend for application workloads so they can take ownership of cloud costs.
With Harness Cloud Cost Management, teams can view cloud cost by:
- Node and Pod
In addition, teams can set pro-active budgets and alerting associated with their application, microservice and clusters.
By analyzing the low utilization of Kubernetes pods running on cloud infrastructure (nodes), Relativity was able to double the density of Kubernetes pods per node from 40 to 100, and in some large nodes, 100 pods per node. The cost impact of this was a reduction in Kubernetes costs of 40% per day.
“Any engineering team that’s deploying Kubernetes and not proactively thinking about managing cost will fail to unlock its true benefits”.Corey Wagehoft, Lead Systems Engineer, Relativity
Finally, with Harness Cloud Cost Management, Relativity was able to defer 1 FTE costs associated with building an in-house cost management tool for product and engineering teams.