Comments on: DPM best practices. Look before you leap.

By: Louw Pretorius

Louw Pretorius — Mon, 06 Jul 2009 08:37:57 +0000

Just use the DPM slider people, it negates all your capacity worries.

By: alex

alex — Sun, 21 Jun 2009 06:00:58 +0000

I see DPM being useful for enterprises whose peak workload (eg mon-fri 9am-5pm) is significantly higher than off-peak workload, so they need more resource during the business hours than out of business hours.

However, I also see the following happening:

Out of hours DPM powers off one of the VMware ESX hosts, a VMware administrator then brings a further VMware ESX host into maintenance mode to do some patching, and boom, just in that instant a further VMware ESX host fails. Suddenly you don’t have enough capacity to power up all your VMs from the failed host. By the time DPM powers up the powered off host your entire VMware cluster has slown to a complete crawl.

I think I’m gonna wait and see feedback on others mileage with DPM before I look to implement.

By: Carl Skow

Carl Skow — Thu, 28 May 2009 20:41:44 +0000

Note that DPM in vSphere also supports the IPMI/iLO BMC connections offered on most enterprise class hardware as well as WoL.

See page 61:

http://www.vmware.com/pdf/vsphere4/r40/vsp_40_resource_mgmt.pdf

By: Ryan Benigno

Ryan Benigno — Thu, 23 Apr 2009 01:46:17 +0000

A use case for DPM not mentioned in the post is DR hot-sites. In a colo facility where you pay for actual power consumption, DPM represents a significant cost saving opportunity.

By: RaiulBaztepo

RaiulBaztepo — Sat, 28 Mar 2009 23:54:20 +0000

Hello!
Very Interesting post! Thank you for such interesting resource!
PS: Sorry for my bad english, I’v just started to learn this language
See you!
Your, Raiul Baztepo

By: Chris Dearden

Chris Dearden — Tue, 17 Mar 2009 15:29:29 +0000

I thought as much

tying maintenance mode in vcenter with maintnenace mode in MOM/SCOM ( via Nworks/Veeam ) would be a good start to this.

By: jason

jason — Tue, 17 Mar 2009 15:26:27 +0000

@Chris: Today, there is no distinguishable difference between a host that has gracefully been shut down, placed in standby mode, or as you put it has “fallen over”. This question was asked in a DPM session at VMworld 2008 in Vegas. Actually, I think I’m the one that asked it – I’ll see if I can dig up the video for that. At any rate, VMware didn’t have the technology at the time to address the question. I can’t say what will be embedded in vSphere. I will say that external/3rd party monitoring and change management systems will still apply though and you’ll need to figure out how those systems account for the (dynamically) planned outage/change. Many shops struggled to wrap a change management process around VMotion and even more so, fully automated DRS. A big difference being there though that VMotion and DRS results in no visible outage, whereas DPM does. If VMware has any brains at all, they’ll at least figure out how to allow vCenter Server to account for a DPM outage such that it DOES NOT fire alerts/alarms for a host outage that it planned and executed itself. If vCenter has this intelligence and your 3rd party monitoring solution ties into vCenter APIs or SNMP traps, then your mostly set but you’ll still need to deal with the change management piece.

By: Chris Dearden

Chris Dearden — Tue, 17 Mar 2009 08:50:44 +0000

Hi Jason,

How will virtual center & my monitoring solutions know the difference between a Host that has been powered down via DPM and a Host that has fallen over?