虚拟机中应用服务实例的负载预期

Question

We currently have multiple Websphere instances deployed in our production environment. They are all on a 64GB VM and each has their own slices of that memory. Currently one of the applications deployed in one of the instances will have an anticipated increase in its number of users (approximately additional 70k users).

Best the best way to assess if the current setup will be able to handle the increase in load? What metrics should be look into? What type of load test should we do? What should we need to tweak in the VM? what should we need to tweak in the Websphere Application server instances? etc.

We are using Websphere 9 by the way and they are in a load balanced setup in production.

Answer 1

What is the current resource consumption (usage of CPU, memory, network, Java heap, etc.) and workload behavior (number of users, request rate, response latency, etc) for the application and WAS instance in question?

Those are the questions that need to be answered to understand (a) how well the app is currently performing and (b) how the system/app is likely to respond to an increased number of users.

There is a huge amount of useful information about performance monitoring, tuning, and diagnosis in the WebSphere Performance Cookbook.

https://publib.boulder.ibm.com/httpserv/cookbook/