Mastering Service Mesh

Introduction to policy controls

To tackle scalability issues, Istio uses a proxy that runs alongside any service, and this model fits well within a distributed environment. The distributed proxy (sidecar) caches the first level of information for the services, hence making distributed scaling easier. Each proxy calls a central control plane service (Mixer) to make precondition checks that contain the second layer of shared cache before and after every request.

Most of these operations can be performed from the local cache of proxy and hence considerably reduce the number of calls to Mixer. Each of the precondition check requests is synchronous and performed from the local cache. The sidecar buffers telemetry information and sends it asynchronously ...