We have been running our virtualized Epicor for a while now and wanted to get some views on what to do form here.
We are running everything from 4 Virtual Machines:
Shipping 64 GB of Ram
Reports-64 GB of Ram
App Server and CRE-64 GB of Ram
Epicor DB-64 GB of Ram
I have been noticing slowness for some users but not all and its better in the morning but gets worse throughout the day. After searching, I found that VM 1 is only using between 13-20% of the Ram whereas VM 4 is running at 85-95% of Ram. So I am taking some Ram from VM 1 and adding it to VM 4.
However, what else could be causing slowness? How much of the local machine being older would cause slowness? Or could it be that the end users with the most issues are hitting the DB more often at the end of the day?
Lastly, any thoughts on my setup and how to make it better is always appreciated.
We have some users still talking about being slow. After timing what they were doing, some were seeing 5-15 min wait times for basic usage. Bellow is what we are running now. Wanted some input and thoughts on our current state.
Which programs are they running… are they on MES stations? If they are on MES, try CLOCKING them out, and in again to see if that resolves the issue.
They need to clock out on their MES Station. We have found that some users never “clock out” of MES, so they are clocked in forever… if they never clock out, then every time they log in, the MES module must look through all their work “today” (the day they clocked in) and it can take forever. Once they clock out, it clears the “today” work, and starts over.
To find out if your users fall into this issue, you can run the Shop Tracker and examine their Clock In date. if it is a really old date, then they are possible problems. The first time I experience this, I had a customer who complained that this was happening… they told me that every 2-3 months, they had to give the user a new clock user id and the problem went away for a while. (ding ding ding… alarms went off in my head.) Originally we though that maybe it was a client problem, but we found that if that user came up to a “fast” client, and they logged into MES, it was slow there also. So this was a user-specific problem.
I do not know the answer to that. But I do know that if you make it part of peoples jobs to clock out every day, and you tell them that there performance review is dependent on that, then they will take care of it. You can drive behavior through setting the ground rules. It doesn’t take a big stick to do this either… “A Pizza party at the end of the month.” for everyone clocking out every day goes a long way.
