Forum Discussion
Deployment best practices
I would like to ask a question about the deployment process.
My organization is still struggling to come to grips with how best to deploy Windows 10.
We are an educational institution, so you may find 40 or more gigabytes worth of additional applications that we need to install in the image. What is your recommendation for a flowchart in how to handle this type of scenario?
18 Replies
- You may want to check out this blog post to consider the benefits of thick images.
https://svenaelterman.wordpress.com/2017/04/27/why-i-use-thick-images-for-lab-deployment/- John LouchCopper Contributor
I disagree with your assumptions. 95% of the software applications we use are packaged and deployed through SCCM, including updates. We do not routinely re-image machines, only when a machine has completely failed is it re-imaged. There is no difference in a machine I imaged six months ago to a freshly imaged one. If your workflow involves routine imaging then I could see the value in a thick image, but otherwise it is just extra work to maintain.
Also, what about your 5% of applications that are not deployed with SCCM?
- John LouchCopper Contributor
Also an education institution. My reference image is a modified stock image, only remove unwanted UWP apps, and add required components. All of the magic is done in the SCCM task sequence. Drivers per model are done by WMI query, common applications are installed as a group, site specific apps are also instead by checking computer name. Post image they will populate collections and install within an hour. We have very few manual installs.
- Luis YoungCopper Contributor
I'm in the same situation, educational institution that requires a lot of specialized applications installed (Geography and Math apps are some of the largest).
In my case, I've been able to leverage MS App-V with SCCM to push out a lot of the smaller applications to my computers after imaging. For some of the larger applications I use SCCM with application or package deployments. There are still some apps that simply cannot be deployed after imaging either because of size and complexity or licensing issues. Those I continue to add to my gold image and deploy them on the image.
Good Luck!
- Marc-Andrea KlimaschewskiBrass Contributor
This is a tough one.
Ideally, you would try to minimize the payload on every machine by only deploying what is needed. You could look into something like Software Center in Config Manager to enable a self service install of apps that are needed only for certain users.
If you have a large set of machines that need many apps, building a reference image and deploying that would most likely turn out to be faster than installing all the apps after the deployment of the actual OS. You will end up with a very large reference image though.
One other thing you may want to consider is in-place upgrade since that would carry forward all of your installed apps.
- Sean HaslingBrass Contributor
So in this case would you use MDT/SCCM to use task sequences to install everything on a single computer to setup a computer and then capture that image again with SCCM/MDT?
Also, in regard to in place upgrades, do you have a link that walks you through that process? This is a very new resource for imaging.
- Heather_Poulsen
Community Manager
Here is information on in-place upgrades:
We also have a virtual lab that will guide you step-by-step through the process: