For most scenarios, round-robin load-balance is good enough. But in our project, sometimes round-robin is not the best choice.
1. every instance of micro-service maintains a data zone by keys. especially using in-memory cache. Using round-robin, every instance may maintains all data. 2. every micro-service may have many instances. Using round-robin, may case m * n connection network. These scenarios can reference to Uber's Ringpop micro-service model. 1. Instances of a micro-service construct a consistence hash ring 2. Each instance maintains a piece of data by specified key. 3. Random select three instance deal with the request from client, so no m * n connections. [ Full content available at: https://github.com/apache/incubator-servicecomb-java-chassis/issues/942 ] This message was relayed via gitbox.apache.org for [email protected]
