How to Deploy Multiple OS
-
Monday, January 07, 2013 12:47 AM
HI,
I got Few Types of Laptops and Desktops So how do i deploy each using network boot?
AS
All Replies
-
Monday, January 07, 2013 2:22 AM
I am assuming you are looking to deploy Windows 7?
this a very open end question. here are some good links to get you started:
http://www.windowsnetworking.com/articles-tutorials/windows-7/Deploying-Windows-7-Part1.html
I would also get this book:
-
Monday, January 07, 2013 2:41 AM
Hi,
Thanks, I know MDT 2012 can deploy but if we create multiple Task Sequence and how do i push to different hardware without selecting?
i mean TS1--> for HP Laptop , TS2 ---> Dell Laptop and TS3---> HP Desktop
Can i specify for each Model?
AS
-
Monday, January 07, 2013 9:34 AM
you can work with a database or modify your cs.ini so that it automatically starts a task sequence based on what laptop the deployment is started.
with selection profiles you can limit what drivers can be accessed by which task sequence.
-
Sunday, February 03, 2013 10:53 PM
HI,
Stefan, this is what i'm looking for so is there any direction or sample config?
AS
-
Monday, February 04, 2013 3:54 AM
Hmm do you really REQUIRE a TS for each model? What are you trying to differentiate?
Just for an example, we deploy probably 12 model laptops and 8 model desktops and I just have a single TS. There might be better ways to accomplish what you are after!
-
Monday, February 04, 2013 7:42 AM
Hi,
Normally only one TS is required for handling several models.
I suggest you go to Johan´s great site and wtach his videos. They will give you a good insight how MDT works.
Here is the link to his site:
www.deploymentresearch.com
Rgds
Thomas
-
Monday, February 04, 2013 8:31 AM
take a look at this blog post by Andrew Barnes he has a lot of useful info.
-
Tuesday, February 19, 2013 10:20 PM
Hmm do you really REQUIRE a TS for each model? What are you trying to differentiate?
Just for an example, we deploy probably 12 model laptops and 8 model desktops and I just have a single TS. There might be better ways to accomplish what you are after!
HI,
I'm looking for the way to deploy same as you. So how do i set Model1 to go with Driver1 and Model 2 to go with driver 2 .... etc with one TS?
AS
-
Tuesday, February 19, 2013 10:26 PM
Hmm do you really REQUIRE a TS for each model? What are you trying to differentiate?
Just for an example, we deploy probably 12 model laptops and 8 model desktops and I just have a single TS. There might be better ways to accomplish what you are after!
HI,
I'm looking for the way to deploy same as you. So how do i set Model1 to go with Driver1 and Model 2 to go with driver 2 .... etc with one TS?
AS
http://www.deployvista.com/Home/tabid/36/EntryID/132/language/sv-SE/Default.aspx
Check out Scenario #3 - Total Control.
-Nick O.
-
Tuesday, February 19, 2013 10:58 PM
Correct, see FNO's answer for a link but basically, you create a folder structure based on the model name and make. (Make = Vendor, Model = Name)
It has to be exact so to make sure I usually just boot into the model and run:
wmic csproduct get name wmic csproduct get vendor
This will return the correct values you need. For instance Dell computers vary slightly but the correct Make (vendor) is Dell Inc. Once you have these values create your folder structure similar to mine in the screenshot:
Then from there all you need to do is make sure your DriverGroup001 variable is set, either through your CS.ini or your task sequence using:
DriverGroup001=DriverGroup001=%DEPLOYROOT%\Out-Of-Box Drivers\Win7\%Make%\%Model%
and you should be right, this will allow you to control which model gets what!
- Marked As Answer by AUSSUPPORT Wednesday, February 20, 2013 3:05 AM
-
Wednesday, February 20, 2013 3:09 AM
Hi Mh & FN
Thank you so much, i'll setup the TS and let you know the out come.
But not sure we can set up one TS for Laptop and desktop....? Laptops got wireless connection software and desktop not needed?
also laptop SOE is bit diferent to Desktop SOE...
AS
-
Wednesday, February 20, 2013 3:25 AM
There are many different ways you can accomplish what your after but I'll show you how I do it currently.
We have several different models (as you can see above) and most of them need some sort of 'helper' app for something or rather. Some of them share apps (like the Renasas USB 3 app), some of them are only for one model.
I created an application folder in MDT and added all of these apps in there then disabled the folder so it doesn't show in the wizard. Then I created an application bundle for each model, adding the relevant apps for the model as dependencies for the bundle.
I then copy the UID of the bundle (seen in the properties page) and create the following in the CS.ini (ignore the other priorities the one you want is Model)
[Settings] Priority=Make,Model,ByLaptopType, MMSettings, Default .... .... .... [3259T3M] Applications001={14986836-3035-4f73-b144-0aec230c3080} [Portege R830] Applications001={cae41d1a-8614-43b3-841a-9b9704afe057} [Portege R930] Applications001={f94f3fb4-7cba-4b3d-86b4-9e45e35db388} [Satellite PRO C660] Applications001={71486418-a468-4d49-967a-7ace91aa589a} [Tecra A10] Applications001={52f116b4-88f7-4905-83c1-99ce23b17c66} .... .... ....At the moment this just displays a list of models on the application install wizard screen with the correct bundle selected. In the near future I'll be changing these to MandatoryApplications001= and removing that screen altogether form the wizard (we are in a transition phase at the moment)
So the result will be:
- Check the model
- Assign the bundle as a mandatory application
- Install all the dependencies of that bundle in the TS
You can also just use the MDT DB functionality to assign settings based on model etc but I just find this way is a little more simplified.
It seems like a lot but once it's set up it never has to be changed, now when you need to add or update apps, you just update the bundles and nothing else.
-
Wednesday, February 20, 2013 3:26 AMBloody formatting in this forum, does anyone else have any problems with Chrome!?!?!?!
-
Wednesday, February 20, 2013 9:05 AM
We run a single MDT 2012, we have over 60 different models and a single Task sequence. We have however integrated our LDAP which contains information of each registerd computer we got, and pushes information with PowerShell synchronization scripts to the MDT database. We send over: macaddress, owner, and role to each computer account in the MDT database.
Doing that we have a ZTI scenario whenever we network boot into PXE, no input is given during the installation process. When the installation sequence is initiated it looks for information in the database to populate the variables needed for the installation.
The task sequence uses standard and simple inject drivers for 'good drivers' based on MODEL rules. (the model name you can get by simply starting cmd > wmic > csproduct get name), we then use selection profiles to bind the drivers together and push the right bundle of drivers to the detected hardware. We use the same method to push 'bad drivers' to each client. Pretty basic and Quick to setup. I opted for not adding tons of rules in the CS.INI and instead used the simplest way for anyone other then me to understand what is going on during a deployment (in the case I get ill or is not at work)
Each Role connects to what type of scenario we deploy: faculty domain joined client, student domain joined pclab client, standalone research client, application packaging client etc etc. Each role contains all the information needed to ZTI/streamline the installation + add the correct applications to the client.
1 MDT 2012 server, 60 clients, 1 TS, endless possibilities - we deploy to 3 different domains, currently 25 different role/scenarios, all automatically managed through our ldap (and ldap frontend). We also have setup 1 TS for Server 2008R2 and 1 TS for Server 2012, but the same methods apply to those deployment scenarios.
MDT is a great tool, and even standalone it is quite flexible and can be made immensely powerful.
//Lars
- Edited by Lars Forsman Wednesday, February 20, 2013 9:07 AM
-
Tuesday, February 26, 2013 6:18 AM
WOW that is nice but i need lots to learn ...lolWe run a single MDT 2012, we have over 60 different models and a single Task sequence. We have however integrated our LDAP which contains information of each registerd computer we got, and pushes information with PowerShell synchronization scripts to the MDT database. We send over: macaddress, owner, and role to each computer account in the MDT database.
Doing that we have a ZTI scenario whenever we network boot into PXE, no input is given during the installation process. When the installation sequence is initiated it looks for information in the database to populate the variables needed for the installation.
The task sequence uses standard and simple inject drivers for 'good drivers' based on MODEL rules. (the model name you can get by simply starting cmd > wmic > csproduct get name), we then use selection profiles to bind the drivers together and push the right bundle of drivers to the detected hardware. We use the same method to push 'bad drivers' to each client. Pretty basic and Quick to setup. I opted for not adding tons of rules in the CS.INI and instead used the simplest way for anyone other then me to understand what is going on during a deployment (in the case I get ill or is not at work)
Each Role connects to what type of scenario we deploy: faculty domain joined client, student domain joined pclab client, standalone research client, application packaging client etc etc. Each role contains all the information needed to ZTI/streamline the installation + add the correct applications to the client.
1 MDT 2012 server, 60 clients, 1 TS, endless possibilities - we deploy to 3 different domains, currently 25 different role/scenarios, all automatically managed through our ldap (and ldap frontend). We also have setup 1 TS for Server 2008R2 and 1 TS for Server 2012, but the same methods apply to those deployment scenarios.
MDT is a great tool, and even standalone it is quite flexible and can be made immensely powerful.
//Lars
-
Wednesday, February 27, 2013 3:03 PMAfter seeing how your Out-of-Box Drivers are setup I decided to redo my layout yesterday to match your so Mine is
Will the format of DriverGroup001=DriverGroup001=%DEPLOYROOT%\Out-Of-Box Drivers\Win7\%Make%\%Model% still work? I have a "DriverGroup1" in WDS but the only drivers there are for my OptiPlex 990 NIC, do I need to replicate all the drivers that are in MDT also? -
Wednesday, February 27, 2013 9:00 PM
It should be DriverGroup001=%DEPLOYROOT%\Out-Of-Box Drivers\%Make%\%Model% I must have mis-copied it.
Looks like it should work, that command is basically just saying 'To get the right drivers, look in the folder that matched the make then the model and include any subfolders in there'. One note, you will need to adjust the folder as above (remove the WIN7).
When your using MDT you won't need any drivers in WDS, MDT should take care of all the drivers necessary.
It is a lot to take in initially but it's a good learning curve and the knowledge is very transferable, so once you get cracking you can take it with you if you move to another package.
-
Thursday, February 28, 2013 10:01 AM
Why just not use selection profiles? Do a selection profile, name it accordingly for the client you will use it for. Add all the driver folders it should look in for that client. Then, in the task sequence, under the step for preinstall, add a new Inject Drivers Step, name it after the Client model, choose a slection profile - pick the one you created. Choose options - add 'if statement' - if any conditions are true. Highlight the 'if any conditions are true' choose add 'task sequence variable'.
Edit the task sequance variable.
In the Variable field type: MODEL
In the Condition field choose: equals
in the value: input the variable it should use, like for a X1 carbon: 34602SG
To get the variable you can run a cmd > wmic > csproduct get name on the client you want to add to get the correct variable name.
This is atleast for me the easiest way to manage 60+ diffrent hardware models. I used to use the whole drivergroup001=% thing back when I used MDT 2008, but has since moved away from that.
-
Friday, March 01, 2013 4:50 AM
Hi All,
I need to setup Branch Office deployment going through with this post but need bit help..
http://social.technet.microsoft.com/Forums/en-AU/mdt/thread/af779fea-33b0-40d0-a98a-03aa83507652
As
-
Friday, March 01, 2013 7:37 PMWhat I ended up doing was removing that line from my cs.ini and adding a new task sequence variable directly above Inject Drivers and the value = "Windows 7 x64\%Make%\%Model%" (minus the quotes).
I do appreciate you showing me how to set a good structure for my drivers.
Thanks mhouston100- Edited by localuserguest Friday, March 01, 2013 7:43 PM
-
Friday, March 01, 2013 8:34 PM
This is a guide how to use selection profiles for managing the correct drivers to the correct client model. I find it explaining why it has a few advantages over using the customsettings.ini method.
However, the best part of MDT is that you can always choose the one method that works for you, something that can be a bit more tricky with sccm osd.
-
Monday, March 04, 2013 12:29 AM
HI Lars,
Thank you so much for your reply and contribution. I'm new to MDT but looking at mhouston100 Method is much easier than the link you add.
anyway thanks for both of you.
AS
-
Monday, March 04, 2013 8:57 AM
Glad to able to help. One of the strongpoints are that MDT supports any number of ways to help a technican solve a problem. I no longer use the DriverGroup001=%DEPLOYROOT%\Out-Of-Box Drivers\%Make%\%Model% becuse I no longer manage singel models, but rather familys of models (ie, each new series of HP EliteBooks shares 95% the same hardware and drivers, why maintain a folder & driver for each model - 2570p, 8470p, 8750w and Folio, when I can use one folder to build a driver Collection for all EliteBooks, and use a selection proflie to just check off each driver needed for each specific model.)
Hopefully you will get started exploring all the methods you can use with MDT to find the best solution for your deployment scenarios.
-
Monday, March 04, 2013 8:47 PM
As Lars said, many ways to do the same thing, just pick whats best. As I did, you will probably pick one path and then let it evolve over time as you learn more about the system. The system we run today looks nothing like the one I created several years ago.
I like selection profiles but we stuck with the Drivergroup model as we tend to have very dynamic models and it's easier for the non-MDT trained techs to add models in. Other than that, both ways work equally as good, no doubt about it.
-
Monday, March 04, 2013 11:20 PM
Hi All,
True, I have managed to deploy to Laptops and desktops in my Head Office. Now i have to add this Technology to my branch office.
Do i need MDt 2012 Servers in each location?
AS
-
Tuesday, March 05, 2013 2:04 PM
I guess that a fairly low number of people are as spoiled as I am when it comes to infrastructure to our secondary campus sites, but there are a few ways to manage branch offices and off site deployment. Linked Deployment Shares are one, where you set up a network accessible share at or branch/off site location, and use the main MDT to replicate the data you need to that site. Another is to put up new MDT sites and replicate/copy the data needed with scheduled jobs transfering the data you need at a convenient time. I know there are a few guys on this board that has the same need you do to manage client deployment.
-
Tuesday, March 05, 2013 8:54 PM
We will be looking to roll out to our branches very shortly, they are all on incredibly slow WAN links.
Our initial assessment is to use a DFS namespace and just replicate the whole deployment share.
Any UNC path will then point to the DFS namespace, which will be the local branch copy. Then we would replicate once a week over the weekend. Pair that with the WDS role installed on the branch server in each subnet and should be good to go.
Using powershell it's pretty trivial to automate the update of the WDS boot images at each branch on a schedule after the sync.
That's just the initial assessment and I haven't spent to much time on it yet but in my mind that should work fine.
-
Wednesday, March 06, 2013 8:16 PM
HI
What do you mean WDS role installed on the branch server in each subnet ? One server for each branch or multiple WDS?
As
-
Wednesday, March 06, 2013 8:58 PM
Each branch has it's own subnet + W2k8 server. It would be too much traffic for clients to boot over the network from our main branch so we would need to install the WDS role on each of the branch servers so PXE / TFTP is server locally.
Not really sure what your asking but WDS is just a role on Windows server which provides PXE / TFTP functionality, paired with the DHCP on each subnet we can point each branch to it's own WDS.
The links we are talking about are often only ADSL(1) connections so less than 1Mb download.
Some people get worried about having to update boot media / deployment shares when talking about multi-site MDT / WDS but with Powershell it really is trivial to accomplish. (ha ha saying that I haven't actually done it yet so I cant be TOO confident).
.png)
