Full disclosure, I'm a VMware admin. We're trying to virtualize a SQL server that handles pricing information for multiple customers. Up until now, it has been a physical system with local SSDs. It's now running as a VM, Win 2016, 128GB memory, 6 vCPU (2sockets, 3 cores) against an all flash array. I/O has confirmed that this thing screams, storage admin says we're looking at very low latency, which confirms what I/O meter is seeing.
When it was physical, it performed a nightly process where it dumped table contents to a series of txt files per customer, then zipped those files up. Repeat again for the next customer.
On the physical system, files that are 250MB (text) would take around 6 seconds. On the VM they're taking around 60 seconds. I can tell the storage subsytem is in good shape. I've increased the queue depth for the vhba, and pvscsi controller as per vmware recommendations and no change.
My DBA's are saying "it worked on the physical server just fine, so it must be your system" If it is, there's no indication on my end that we're saturating the storage connections or running into resource contentions.
I'm attaching my sp_blitz and the subsequent results of the script that shows the wait for the various DBs... one is much higher than others.
