Has the Service Mesh Killed the Fat Network Client?
Jun 09, 2019
Applications now have so many services that it has become prohibitively expensive to put network features like load balancing, circuit breaking, etc, in a "fat" client on every service. So what are the alternatives?
Developer Advocate, HashiCorp
How many companies are running distributed systems that work 100% of the time? The answer, as any systems engineer knows, is none. The key to distributed systems reliability is managing failure.
Microservice design is deliberately loosely coupled and decentralized. While you gain the benefit of high availability, you need to implement particular network reliability patterns like load balancing, circuit breaking (health checks), retries, and timeouts, often as a “fat” network client in each service.
In this talk, HashiCorp developer advocate Nic Jackson will take a look in depth at these network reliability patterns and the benefits they give your systems. You'll also learn how to leverage the power of the service mesh with Envoy and Consul Connect to implement those patterns without the need for shared libraries or code changes.