The Mythical Multi-Cluster Month
Posted on May 1, 2020 • 1 minutes • 207 words
Conference
Kubecon NA
Abstract
There’s a prevalent theory that adding one more cluster will solve your problems. It’ll unblock that team, or enable that new project to be successful. But the cluster sprawl and management overhead can slow you down. Local optimization can have big impacts to the global organization. In this talk we’ll look at operating cluster scale at scale. What patterns can we learn from the past to help us plan for future growth, and what tools exist to help us get there?
Benefits to the ecosystem
There are a lot of tools that help with parts of multi-cluster management (provisioning, configuration, deployments, networking) but rarely do talks or tools talk about the complete considerations of when to use multiple clusters or when something like multi-tenancy is a better option. Many talks focus on how they can optimize one portion but it’s hard to put those things into practice without understanding how tools fit together or what day 2 operations really look like. This talk attempts to address some of those areas for multiple use cases such as multi-cluster for environment splits, for team/application isolation, for on-prem/off-prem, and for multi-region deployments. Focusing mostly on cluster operators and how their choices impact application developers.