|
|
|
|
Back to newsletter 183 contents
We're seeing an increased understanding of the pros and cons of microservices. In this newsletter, I'll list the ones I've gathered from a dozen other newsletters where I've previously extracted microservice performance tips.
Pros:
Decoupled systems have higher fault tolerance and are easier to understand and fix. Improved fault isolation. Ability to upgrade components independently. Easier to scale as you can scale only those services that need to be scaled. Encourages better handling of partial failures (as you can more clearly see interconnectedness). Fast starts. The ability to replace a service completely with a new implementation. Separate optimised data stores for each microservice. Reduced complexity. Less likely to use (and need) huge machines; can use cheaper hardware.
Cons:
Consistency issues across data stores. All the overheads of a service are multipled across the system, as each microservice needs to independently throttle, timeout, circuit break, manage connection and thread pools, pipeline requests asynchronously. Too much decomposition can cause too many network hops, excessive runtime overheads, and a complex interlinked difficult to understand system. Increased communication failures. Testing a distributed system is harder. Additional latency from additional hops. Inefficiencies from multiple processes duplicating their uses of similar resources (eg JVM GC threads). Distributed service issues: split brains, dropped packets, slow networks, other JVMs hitting GC pauses. You need to be much more rigorous in decoupling your components and there is additional deployment and distributed monitoring complexity.
Many of these cons are related to having distributed systems, but if you will be having a fault tolerant clustered system with failover capability you are likely to have the same issues anyway, so for large systems there are fewer cons than it seems here. And increased simplicity of understanding the system may override everything, as the real bottleneck is the people developing and running the system. Now on to our usual links to tools, articles, news, talks and as ever, all the extracted tips from all of this month's referenced articles.
Java performance tuning related news.
Java performance tuning related tools.
Back to newsletter 183 contents
