Allow for Shared Instance Pooling Groups on ArcGIS Server

[RyanUthoff]

Currently, ArcGIS Server only allows for one shared instance pool of feature services that ALL feature services share.

My idea is that we have shared instance pool "groups." These groups would work the same as the current shared instance pool, except we could specify the number of shared instances per group.

This is beneficial to organizations who have multiple projects hosted on their server, or otherwise need to group feature services together in their own shared pooling without the hassle of managing dedicated instances. I envision that you can create shared instance "groups" in server manager where you can assign a number of shared instances, and then be able to assign individual feature services to that shared instance pool.

This is beneficial because you can still take advantage of the shared instance pooling, but at a project/group level where you don't have multiple projects/groups competing for resources in the single large shared instance pool, and without having to manage dedicated pooling at the feature service level.

[ReedHunter]

There is also a stability argument in favor of your request.  In my organization sometimes one of the shared pool services becomes unresponsive or otherwise becomes unstable.  It can happen because of database problems, or when we find out the service became suddenly too popular and should be switched to a dedicated service.  Our clue that this is happening is when various services in the shared pool start mysteriously failing and then recovering on their own.

This causes a scaling problem - the more services in the shared pool, the more risk to the shared pool.  If we could have sub-grouping of the shared pool, or in effect multiple shared pools, then this risk could be mitigated.

[RyanUthoff]

@ReedHunter 100% agree. We encountered an issue where a layer was causing the feature service to crash. And since it was a shared instance, it was causing all of the shared instances to crash. Changing it to a dedicated instance isolated the issue to only that feature service crashing. So, having multiple shared pools could help mitigate the risk of a single service causing all the shared instances to crash, and instead, just limit it to a smaller subset of shared instances in specific pool.