I don't have to rebuild the whole thing, including drivers and applications.
If an update annihilates my original OS (and they often do), I can still fall back on my original without too much negative effect. If things go sour on production deployment, I just disable the updated one, revert to the original, and keep testing. If an application update comes out, I can make a copy of the original app, create the update TS in a test environment and make sure it works, then deploy it while deactivating the original version, just to make sure. This way, I only update what's needed without busting up the whole works. Create one TS to house all of your drivers, and update only that one whenever you get new machines in, and every other TS would draw from it? I'm long out of the deployment game so I don't know where that left off, but that would've been another godsend.
MS engineers were working on it for some future update of ConfigMgr (and none of the deployment experts I follow on Twitter have mentioned it yet, so I'm guessing it's not ready). I could just build a single reference task sequence, test the living tar out of it, then pick and choose the parts from it that I needed for specific groups of users or devices.Ī while ago, a request was submitted to User Voice for ConfigMgr to allow TSes to reference other TSes. Once the ConfigMgr team released the ability to copy TS steps from one TS to another, that cemented this method for me. Either way, though, my physical involvement was minimal, which was great.Ĭapture the base OS, package up common applications (all separately), package drivers, and then build task sequences (SCCM). For individuals, I could work with them over the phone to do it remotely. I was fortunate in that I could build out the images and make the junior techs go out to the buildings to deploy.Īt least in the case of mass deployments. This would include not just the source software, but number & distribution of clients plus the number of people you have available in the field to actually perform the deployment. Depends on your deployment environment, of course.