Wednesday, 11 February 2009

ESXi Health Status Inequality - HP vs Dell

Ok, so I'm pretty sure most of you know I've been haxing around with ESXi for quite some time, but I only recently noticed the difference in what you see in vCenter Health Status depending on which hardware vendor you use. I'm sure you also all know that I'm a massive fan of HP, which is why I'm a so unimpressed with their integration with vCenter. For example, lets have a look at recent generation HP BL685c blade running the HP OEM version of ESXi 3.5 (installable), downloaded from the VMware website.

Notice anything strange? Or rather, notice the distinct _lack_ of anything out of the ordinary. That's right, when comparing this blade to a non-OEM ESXi installable, it looks exactly the same in vCenter! At first I thought there must have been a corruption in the CIM providers or something, but after investigating with my VMware SE, we could find nothing out of order.

Now compare the image above to what you see in vCenter on an old Dell rackmount (note the bios date - nearly 2 years old!).


Now I can only hazard a guess that HP want you to use Insight Manager to perform health monitoring exclusively, but why force me to deploy more monitoring infrastructure than I need to? They could at least create a VI client plugin for Insight Manager so I can use a single pane of glass! But even with that, it doesn't seem like you get as much information as what you do with Dell - looking at Appendix B of this HP whitepaper, storage subsystem information is strangely absent.

If anyone else out there has seen this, seen different, or can compare with other hardware platforms like Sun or IBM, please share your experiences in the comments!


TechHead said...

Good article.

This has been something of an annoyance for myself for a while, as I too am an avid HP Proliant fan.

I thought that perhaps I was the only one bothered by the distinct lack of anything above the norm in the way of environmental monitoring in VCenter Environmentals for HP servers.

I can only hope that maybe something is in the pipeline from HP that'll rectify this.


Doug Hazelman said...

This is interesting indeed. Each hardware vendor has different implementations of CIM/SMASH. The VMware API can only see what has been exposed by the hardware vendor.

My only guess is that Dell has been a much better early adopter of the standards than HP. It's interesting that you're pointing this out on ESXi since this is essentially the only way to get this data. You'd like HP would jump on the standards bandwagon.

I also believe that Dell has been much better at working with VMware and also in helping to design the API. I do know there's a difference in Dells embedded vs. download able ESXi for this very reason.

Rob L. said...

You use HP's Virtual Machine Manager (VMM) plugin for Systems Insight Manager to give a sort of single plane of glass. It wil give some basic Virtual Center stats, start - stop VM's, vMotion, etc...

Jason Boche said...

The CIM aren't 100% cooked yet IMO and need improvement. My DL385's have a permanent yellow status due to "ProcHot for System Board 2 - Normal" according to the CIM providers, yet the HP SIM agents (latest version 8.11) show all green and no problems.