A great deal of Mux’s infrastructure decisions are driven by the cost effectiveness of various services, regions, and workloads within their multicloud (AWS+GCP) environment. Those decisions are largely controlled by their internal platform team, which provides a Kubernetes interface to the rest of their product engineering team, and manages how that Kubernetes layer is positioned on their different cloud providers.

Naturally, budgetary determinations play a large role in how those various teams are able to spin up and manage their workloads. Mux houses a business operations team that centralizes and analyzes their cloud cost data across the multicloud environment, which means that they need a single source of truth to ensure that they have the full view of their overall cloud spend.

However, when it came to optimizing that spend, Mux Co-Founder and Head of Technology Adam Brown needed additional assistance. The team was Reserved Instances on AWS, but struggling to manage their usage over terms of the commitments. There were occasions when the team decided to change machine types and had to go through the painful process of converting or selling the outdated reservations, or optimizing backend software and subsequently lowering their compute consumption, thus negating the need for the full capacity that they had purchased.

The time and expertise that would have been required for a fully optimized commitment strategy was too much for the growing company to manage on their own. This led to ad-hoc commitment purchases that failed to provide significant savings, while still draining the team of valuable time that could otherwise be spent improving their technology.