We're noticing a huge impact to the source system we are replicating data from when refreshing data with many threads (sliced and/or standard).
It is important to spread workload into several SAP source servers but the nature of the load balancer for SAP is not ideal - it's based rather on time of the session creation than utilization or classic round robin approach. When starting our refreshes via the HVR6 GUI, all of the threads mostly hit only one of the servers so that server is running at 99% CPU Usage while the other is severely underutilized.
This means the processes on SAP application servers created by HVR needs to be delayed between each other. We have a custom approach when refreshing to have the threads get refreshed with 15 seconds of delay between each other and this works very well. We were wondering if this could instead be added as well as a functionality in HVR (LDP) itself.