Prioritise selection of zOS Host Connections in Copy Services Manager

In our particular environment, we only have one Sysplex (with a mixture of LPARs all with different CPU weightings) and as we are very familiar with our business and for us, having the ability to inform CSM which order to use the zOS systems would be very helpful rather than just letting CSM arbitrarily pick a system
.
Also, all the documentation that we have seen to date does not advise the Customer that when you add zOS systems to the "zOS Host Connection" in the CSM GUI that the order entered is not taken into account and that CSM internally will randomly select one. Either the documentation should be updated and/or the "zOS Host Connection" GUI should have an explanation
.
Your question as to "if you have proven that the system with less CPU cannot handle the requests, then why would you use that as a redundant connection for CSM" - we did not do that on purpose. We are a new Customer to CSM (as we only implemented DS8K and CSM back in early 2018) and thus was all new for us. As stated in the preamble, we only have one Sysplex (which is a mixture of environments). We did not choose a PROD system at initial deployment of CSM and thus chose a DEVT/TEST system (which is going to have a low CPU weighting) as compared to a PROD system to ensure that everything was working (no Customer is going to deploy a new technology in PROD from day one). Using the DEVT/TEST system appeared to work fine as a zOS Host Connection until recently. Again, there was nothing in the CSM doc that cautioned the use of a low weighted LPAR as a zOS Host Connection
.
Giving the Customer a choice as to whether to just leave it to CSM to make the decision OR allow the Customer to make an informed decision based on their business requirements to select and prioritise which zOS Host Connections are to be used makes more sense that trying to assume "one size fits all" approach.

I understand the benefits of CSM dynamically choosing which z/OS to use. However, if CSM is going to take responsibility for that decision, does that not also mean that it needs to apply some intelligence to that decision? For example, it could work with WLM to determine which system currently has the most available capacity and then select that system. In the absence of that, or as an interim, easier-to-implement solution, giving the customer the ability to specify a preferred sequence would seem to be an improvement over the current situation.

We will consider this, however, the primary reason to have more than one system connected to CSM in a Sysplex, is to provide redundancy. This means that even if we did allow you to prioritize the order of the connections, if the system with more CPU available were to disconnect, CSM would still fall back on the system with less CPU. And if you have proven that the system with less CPU cannot handle the requests, then why would you use that as a redundant connection for CSM. Is there a reason that you need both connections? Having to manually choose a priority of connections is something that would not only be extraneous for customers but something they might forget or not even know they should do. This is partly why we automatically just choose a connection based off the connection state.