Hi folks,

1,600 Users
3 Message stores (400,800,400)
Most recent ZCS 7.
we're running RHEL on Vmware - i've ruled out ESX issues

We've been toying with java heap and innodb_buffer_pool sizes and I'm wondering what we should be seeing for:

1. # of concurrent processes (700 users on a 12GB message store) - I'm only seeing 1 or 2. Should be around 8 I think.

2. CPU utilization characteristics (I'm seeing thrashing) - Should I see sustained 50-70% or should I see more erratic looking behaviour - I think it should be more sustained at higher levels.

We had consistent sustain of 70% on a system, then Zimbra instructed to the heap and increased the innodb_buffer_pool to 30 and 35% respectively. My system has 12GB of RAM, and these numbers more suit a 32GB message store.

Just want to see if we're moving in the right direction,