Domains
Using domains greatly enhance performance when you deploy multiple services on the same on-premises Mule runtime engine (Mule). By providing a centralized point for all the shared resources, domains make the class-loading process (and, therefore, metaspace memory usage) more efficient, especially because domain dependencies declared in the pom.xml file are also shared in the domain apps.
Example of Increased Performance Using Domains
Imagine that you have a set of HTTP Proxy applications that also apply one API Gateway policy, such as an IP whitelist. Each application interacts with the same backend. In a Mule instance configured to have 2 GB of heap memory and 256 MB of metaspace, deploying those applications independently without using a proper domain, results in a noticeable reduction in TPS (transactions per second) when deploying 60 or more applications due to saturated resources. At 80 applications, you see deployment issues due to out-of-memory exceptions.
You can greatly improve this outcome by using a domain that shares the backend server configuration, increasing the number of applications to beyond 100 while also experiencing a balanced use of machine resources and consistent good performance.
Note that despite the clear performance advantage in using domains, each deployed application adds its own unique complexity to the shared infrastructure resources. To avoid performance impact, before adding an application, identify the overhead by testing each application individually and then test it coexisting with other applications.