vCenter

Hi,

Has anyone else looked into changing the SCSI controller of the vCenter appliance? I find it strange that the appliance is shipped without using the paravirtual SCSI controller, we have seen good returns when moving to this and would like to across the board where possible (known limitations taken into consideration).

I appreciate the appliance is distributed 'as is' and modifying it in such a way would likely void any support but I'm still curious to know whether anyone has tried it.

As an update... it looks like you can put the second hard disk on a separate SCSI controller and set that to paravirtual, changing the boot disk however causes it to break.

I have raised a case with VMware to see what they say but am still interested to see if anyone else 'in the wild' has done this.

Looks like I'm the only one with anything to add..
.

Finished a call with a VMware engineer who confirmed that the PVSCSI controller is not supported on the existing vCenter appliance. This is due to the fact that it runs SUSE Linux 11 without service pack 1 which is required for support.

Obviously the engineer could not comment on future developments but perhaps a new release of the appliance will run on a later version of SUSE which does support the PVSCSI adapter.

I've just been through a best practices audit and VCSA 6 is still using LSI Logic Parallel (not even SAS) for the disk controller.  Furthermore, VMware is placing all the disks on one controller (against their own best practice).

This is also the case with several of VMware's other appliances :smileysad:

Thought I would chime in as well. Still LSI Logic Parallel with VCSA 6.0 Update 2.

Would be interesting if any VMware employees have any thoughts on this.

AK

It appears they also have it configured with a queue depth of only 128.  I suppose that might be fine for small environments that aren't very busy, but I don't know why they wouldn't set it higher to avoid that being a bottleneck in large environments.

cat /proc/scsi/mptspi/0

ioc0: LSI53C1030 B0, FwRev=01032920h, Ports=1, MaxQ=128

Kudos to Michael Webster for re-interating the queue depth issue with certain virtual disk controller configurations (which VMware continues to ignore for their own appliances):

Performance Issues Due To Virtual SCSI Device Queue Depths | Long White Virtual Clouds

Clearly VMware isn't really looking at this thread, but I thought I would revive it.. STILL with vcsa 6.7... using LSI Logic Parallel, 13 disks on 1 controller. (Oddly, the appliance OVA is configured with 2 LSI Logic Parallel Controllers, but nothing is attached to the 2nd one...)

OS is PhotonOS so they should have full control of this relative to driver support etc..

I was trying to set the drives to NVME controller but also failed (not sure if that is bad to do that though, im a newb). I am using NVME drives.