SCCM 2012 R2 - Maintenance Windows - a few questions

Question

Let's say I configure a maintenance windows (MW) for the 2nd saturday of every month from 6p-10p and I have 100 servers in this maintenance window,
are they all going to try and patch and reboot at the same time?
(Assuming they are all online)
This brings up concerns when trying to plan maintenance windows for our servers - especially when we are 80%+ virtualized. I can't find anything that states the sequence of events that happen when a machine goes to install patches based on a MW.
Do they wait on a heartbeat? Do they typically kick off right at the front of the MW? Does SCCM spread the installs out?
We currently use WSUS (working on migrating to SCCM) and in WSUS we have a ton of groups based on the hour so we spread out the machines being patched. I am trying to avoid having so many different options and times and instead just using MWs.
Another issue we are going to run into is certain machines need to be patched/rebooted in a certain order and with others we just cant have them all down at the same time (e.g. domain controllers). I supposed I could create 2-3 phases per patch weekend to
spread those machines out - is that the best option? (for instance; on the 2nd Saturday we have MW Phase 1 from 6p-8, MW Phase 2 from 8-10, and MW Phase 3 from 10-midnight)

Answer 1

Let's say I configure a maintenance windows (MW) for the 2nd saturday of every month from 6p-10p and I have 100 servers in this maintenance window,
are they all going to try and patch and reboot at the same time?
(Assuming they are all online)
This brings up concerns when trying to plan maintenance windows for our servers - especially when we are 80%+ virtualized. I can't find anything that states the sequence of events that happen when a machine goes to install patches based on a MW.
Do they wait on a heartbeat? Do they typically kick off right at the front of the MW? Does SCCM spread the installs out?
We currently use WSUS (working on migrating to SCCM) and in WSUS we have a ton of groups based on the hour so we spread out the machines being patched. I am trying to avoid having so many different options and times and instead just using MWs.
Another issue we are going to run into is certain machines need to be patched/rebooted in a certain order and with others we just cant have them all down at the same time (e.g. domain controllers). I supposed I could create 2-3 phases per patch weekend to
spread those machines out - is that the best option? (for instance; on the 2nd Saturday we have MW Phase 1 from 6p-8, MW Phase 2 from 8-10, and MW Phase 3 from 10-midnight)
Hi,
I often help customers with this scenario that you are facing. I'll try to answer your questions as clearly as possible.
A) Yes, if you have 100 servers in one MW during 6p-10p, they will start at 6p if they have recieved the policy before that. What you can do is to activate the Deadline randomization that was introduced in ConfigMgr 2012 SP1. You can read more on that here
http://technet.microsoft.com/en-us/library/gg682067.aspx
B) ConfigMgr does not wait for a heartbeat or anything like that. You need to manage to logic to move your workload from hosts before you take them down for a reboot. VMM could help you with this IIRC. Or at least SCSM SMA and PowerShell. The sequence is
basically that they install the patches, if a reboot is required, it will reboot at the end of the patches.
C) Yes you need to create different MWs and place those servers in the MW order that they need to be rebooted. Be careful tough. If you have ServerA, ServerB and ServerC that need to reboot in that order. You place them in MWs in that order. You might end
up getting patches to some or all of those servers from a Patch Tuesday but perhaps only ServerB will need a reboot so only ServerB will reboot and thus put you in a place you don't want to be. To solve this, create a re-running package/program deployment
that reboots ALL servers that are sensitive each and EVERY MW.
Tim Nilimaa | Blog: http://infoworks.tv | Twitter: @timnilimaa