I have a gateway wherein I have deployed 10 services/policies. In which 5 services belong to a particular endpoint application and is consumed by a different set of people (Lets say set A) and the other 5 belong to another endpoint apllication consumed different set of people (Let's say set B).
Now, I need set A and set B to be independent of each other's performance. Say, If the load becomes high for set A or the endpoint faces some issue, I don't want the services in set B to be impacted due to this. What I basically need is to split the gateway's resources (Like CPU usage and Memory) between set A and set B services. While I totally understand that this can be acheived through deploying the other set in a seperate gateway, we don't have the budget and usage to do so currently. On the other hand I understand that the services run under a single gateway PID and hence this logical partition may not be possible (Please let me know if it's otherwise).
I was thinking whether allotting seperate listen ports to each set of services and defining seperate thread pool might address my issue. I am totally not sure on how this will work or if this approach is possible. Can someone please throw some light on this and possibly explain how this might work? Also, please let me know if there are any other ways of acheiving this.