Monday, 18 February 2019

Application access using groups in G Suite

Managing application access through the organisational tree in G Suite is both a blessing and a curse.

On the plus side it’s a simple, easy to use framework which has proven extremely effective in deploying dozens of applications across thousands of user accounts with minimal management overhead.

On the minus side it has one major weakness that has been detailed in a number of posts. The organisational tree cannot support a situation in which a user is a member of many different sets. For this reason it’s not possible to deploy an app to just the students in the History Class in Year 10. To deploy an app to a subject set you need a one-to-many relationship and that can only be done through groups.

Other platforms such as InTune for Education use this approach as do many other MDM platforms including Google's  Mobile Manager. That’s not to say groups don't create their own unique management issues but the point remains that to deploy subject based apps you need groups.

Fortunately allowing groups to control app deployment is a feature that has recently made an appearance in the G Suite Management console. So are the problems solved ?

The first thing to know is that the type of group used to control app deployment cannot be defined using Google Groups for Business (GG4B) but must be created using either the Groups icon within the admin console or through the Directory API. For this reason groups maintained using Google Cloud Directory Sync and most other third party products will work fine. Once created they can be managed through GG4B. Nested groups are supported.

The second thing is that groups can only be used to turn apps ON. This means that groups only have an effect if the sub-OU holding the user account sets the app to OFF. In this case the group setting (ON) overrides the sub-OU setting (OFF).

This might sound a bit limiting but that's not really the case.

Let's take a few examples.

Task 1: You need to turn on Blogger/Sites/Hangouts for a 6th form Media Class that contains user accounts held in a number of different sub-OU’s. The default action is set lower down the tree and turns these items OFF for all students. Prior to groups this was a big problem. You could either turn it ON for students or tie yourself in knots creating sub-OU’s. Now you can just create a new access group called app_Media_Group drop and turn on whatever apps you need for that group.

Task 2: You’ve been asked to trial Classroom in your school. You need to grant access to two teachers who’ll be taking the English class and a member of the SLT team. The problem here is the teachers reside in an sub-OU with all the other teachers, the SLT team are in the parent OU and the students in a complete different part of the tree with 60 other accounts. You consider the OU approach before realising that one of the students is already residing a sub-OU Penalty Box and can’t be be moved. Groups to the rescue again.

While we are on the subject of a Penalty Box it might seem like a good idea to allow groups to  turn apps OFF as well as ON to cover exactly this scenario but when you consider ease of use that may not be the case.  Flexibility often brings more complexity and in this case I believe Google got it exactly right because even with the added group filter it’s still pretty to work out what policy applies to which app.

In this example we have turned Jamboard OFF for all users at the root but then enabled it for the group called groupwithbusinessoff.

The console immediately shows it as ON for some with the View Details link to break out the information you need.

To examine the apps from the user perspective you can use the Apps card on the user account dialog.

Dropping the user account into the groupwithbusinessoff group shows the app being applied for that single user.

Like most of what Google does, application control through groups is a simple idea that’s been well implemented, but could it be better?

At the moment groups can only be applied for G Suite and Additional Google Services but not Marketplace apps which are more likely to be subject based. Also there's no sign yet of group based deployment from the Chrome Store which would be a real breakthrough.

Lastly since membership of a group can now enable an app, why not a Google Classroom.

Now there’s an idea, over to you Google.

Friday, 18 January 2019

InTune for Chromebook Admins (p3).

Application Deployment.

In the last of three posts we take a look at application deployment using InTune for Education in a SaaS based school

Intune for Education (IT4E) can be used to deploy three types of application.

Web Apps. These are no more than shortcuts to web sites. If you have an externally hosted learning platform or any application that has a web front end you’ll use the web apps option to deploy the link.

Microsoft Store Apps. This feature allows you to install apps from the Microsoft Store for Education. All store apps use the Universal Windows Platform (UWP) which is a unified format for Windows 10, Windows 10 Mobile, Xbox One and HoloLens but not Windows 7. Therefore the school needs to be running Windows 10 to pull apps from the store. However since you also need Windows 10 for IT4E the two requirements go together.

Desktop Apps. Also known as Line of Business apps in the InTune console and probably Legacy Apps in the not too distant future.  These are the standard .msi deployment packages that have been around since Windows 95. Note that IT4E does not support .exe install packages, it’s the .msi format or nothing.

Deploying Apps.
Applications are allocated in the same the way as policy, through a hierarchy of groups. This strategy is a little more flexible than the Google approach which has only just started to adopt group based application deployment.  Another area where the Microsoft approach differs from Google is the ability to assign applications to devices. This has no counterpart in the Google world where user and device policies are strictly divided.

The application set that’s delivered to the user session is a combination of all the groups that reference both the user and device. If an installed application falls out out scope when a user or device is removed from a group the application will uninstall on the next policy refresh.

The end result is much the same as general policy management, you have great flexibility but this power comes with the ability to tie yourself into knots if you don't have a plan to start with.

First impressions working with app deployment in IT4E is that some tasks are really easy and other tasks are impossibly hard.

On the easy side, deploying a Desktop App (.msi) is a very slick process.  Fill in a simple dialog in IT4E and select the .msi file. The file itself is uploaded to a central data store within Azure which is provided as part of the platform and then pushed down to the user/device groups on a policy refresh. This works surprisingly well allowing for the time to reset the policy and to transfer the .msi data package. Web Apps behave in the same way except that everything works a little quicker as there’s no install package to move about.

Although the process allows you to select an icon for the application this only shows up on console itself and doesn’t get pushed down to the device as you might expect. This irritation will probably get fixed at some stage.

Deploying apps from the Microsoft Web Store requires an initial effort to set up the Store account for the school. Currently this can only be achieved using the Intune Azure interface rather than through IT4E. Until the store account is linked back to Azure the option remains greyed out in IT4E.

Once the account is activated the process involves selecting the app from the store interface and then synchronizing it with the InTune database so the application displays on the IT4E console. Once that’s done you are presented with the same group allocation model you have with Web Apps and Desktop Apps.

The synchronization between the Store and InTune runs as a background process so it’s unlikely that the app will immediately display in the IT4E console, just give it a little time. The full Intune Azure interface has an option to force a sync which is a benefit for impatient admins and it would be nice to see this feature in IT4E as well (above).

If these are the easy bits - what are the difficult bits?
  • You can’t install from a .exe file, a common format for legacy software and drivers.
  • You can’t place a link onto the desktop to create a customised user experience or do anything in the way of file management.
  • You can’t map Windows file shares.
  • Print support is underdeveloped at the moment.

Currently the solution is to create a Powershell package deployed through InTune or create an .msi wrapper. In fact just about anything is possible if you are willing to spend some time scripting but why make it so difficult ?

The answer lies with Microsoft's cloud focused vision. In the future you’ll be unlikely to map Windows file shares - you’ll use OneDrive. You won’t sideload local apps, you’ll download them from the Microsoft store or consume SaaS.  You won’t disseminate information by printing, you’ll share, publish and collaborate using Teams and you won’t be able to use the desktop as a filestore because the vision is cross platform.

These things are hard because by adopting the easier options you’ll be nudged towards Microsoft's cloud wonderland.

In conclusion let’s return to our Google administrator living in an alternative reality who has the job of integrating Microsoft's emerging technology into a serverless SaaS based school.

What are the principal features of Microsoft modern management platform that a Chromebook admin will have to come to terms with ?

  • A cloud based platform managed through a web interface without any dependency on local infrastructure.
  • Locally installed applications delivered using a store model.
  • Simple “lifetime” device level licencing as an option.
  • Close integration with a cloud based productivity platform to deliver LOB applications.
  • Strict application control limited to web links and store apps with restricted support for side loading.
  • A updated security model based on web open standards with backend integration into common SaaS platforms.
  • Reliance on a cloud based user directory that all also supports groups and devices.
  • Security patches and updates delivered centrally on a predictable release schedule.
  • Management controlled through a hierarchical policy set.

Does any of this sound familiar?  I suspect it won’t be too hard.

Monday, 7 January 2019

InTune for Chromebook Admins (p2).

Understanding user policy management.

The key to mastering Intune for Education is understanding that policy is filtered through a hierarchy of groups. This is a fundamental difference between MS InTune and Google G Suite.

In G Suite, policy is applied using a simple parent-child folder tree that’s constructed by the admin user. A user or a device (chromebook) can only exist at a single point in the the tree. Policies flow up from the root to the branch of the tree where the object resides. If the user or chromebook is moved from one sub-OU (branch) to another the policies are updated in a consistent and predictable manner.

With InTune you still have a hierarchy but it’s a hierarchy of groups and since an object can exist in many groups it’s possible to create a structure that places an object at multiple points in the tree. While this model provides a level of flexibility that can’t be matched by the Google system it can create extra complexity if you have many groups and many branches in your tree.

For both platforms it’s best to keep it simple but that's particularly true for InTune so let’s see how this might work with out new InTune user Joe Schmoe.

Each InTune group has a built-in tab that lists it’s configuration settings in the same way that the User Chrome Management in the G Suite console presents a list of options.

The major difference between the two approaches is that with Google you navigate to the object you wish to manage (G Mail, Chromebook Devices, Calendar), open the settings tab and then select the point in the OU policy tree you wish the setting to apply..

With InTune the settings are embedded into the group itself and not the service and this creates a number of challenges for the administrator.

The most obvious one is that settings tab must contain a link to every possible value you can set in a policy.  At the moment this is fairly manageable but if the list ever expands to anything close to the number of options available to the G Suite admin it could get quite ugly. Fortunately Microsoft has provided a search facility that works in the same way as some of the larger G Suite config lists. I can see this becoming essential.

The situation is compounded by the fact that a group can contain both users and devices and therefore the policy list must contain the settings for both. The settings tab has a pulldown section labelled Windows Device Settings (above) but this also contains user profile settings as well.

For the full confusion effect if you open the settings tab for the built in All Users groups you are still only presented with the Windows Device Settings pull down. Renaming it to Windows Settings would help as would breaking the policies into two discrete sets in the same way that G Suite works.
Which brings me to the next point. Because the policies are mixed it’s difficult to know which are controlled by the user object and which by the device object. If Cortana is turned ON by group user policy but then turned OFF by a Device group policy in a higher level group, which takes precedence? Where this gets really hard to comprehend is when both the user and device are members of two entirely different policy trees and so you’re unable to use the position in the tree to work it out. Let’s see how this could happen and what you can do to avoid it..

First we need to understand the role of the two built-in groups All Devices and All Users which sound like they would sit at the root of your policy tree but of course being Microsoft, they don’t.

Both groups are system defined which means the membership cannot be edited and user defined groups can’t sit above them to inherit policy. So both groups exist in quiet isolation. The main use is to act as the default policy set for any user or device not immediately assigned to a group, a sort of catch-all template policy. If you have a simple setup you can use these groups to push configurations out to your entire estate of users and devices in one easy step which is a great alternative if you wish to avoid the complexities of a group hierarchy.

Nested group policy only applies to user defined groups, so let's create a group called All Students and turn Cortana ON for that user group. You can now create another group called Year 10 as a child of All Students and you’ll see this automatically inherits all the settings of the All Students group with Cortana turned ON. However you can edit the settings for Year 10 to turn Cortana OFF overriding the inherited settings from All Students .

So Joe Schmoe could be in three groups.

  • All Users  - Cortana turned OFF.
  • All Students  - Cortana turned ON.
  • Year 10  - Cortana turned OFF.

To find out whether Joe has access to Cortana you need to know which groups Joe is a member of and how those groups interact. Simply checking that Joe is a member of All Students where Cortana turned is ON tells you nothing because he could be a member of another group higher up the tree where Cortana is turned OFF.

And then it gets really complicated because you also have to consider that the device that Joe is using also has it’s own policy tree which could be completely independent of Joes groups and even fixed to a completely separate group at the root.

So when you turn on the computer is Cortana going to turned ON or OFF?

There’s no way of viewing  the resultant set of policy (RSOP) from the console so it’s quite easy for incompatible settings to be applied to the same group. These inconsistencies result in errors when a user or device is being set up with different settings in multiple places.

For example let’s assume Joe is a member of the Year11 group and is also a member of the Geography group. If you configure a homepage setting and assign to Year11, and you configure a different homepage setting and assign it to Geography, Joe has two conflicting homepage settings which leads to an error. Microsoft have thoughtfully provided a report that lists those errors - the group settings errors report.

I suspect any admin trying to implement a fairly complex group hierarchy is going to spend quite a while going through those reports.

Alternatively it might be a better idea to adopt a system that avoids the problem in the first place, so how do you do that?

First understand that although the All Users and All Devices groups are useful tools if you need to differentiate policy between sets of users and devices they are of no real use.

So as a first step create a new root group that represent your school and call it ITE-MySchool for example. Two important things about this group. Preface it with something like ITE so it can be easily identified as a policy group and then invest some time transferring the settings from the All Users group into it as nothing is set by default. Don’t put any users or devices in this group.

Create other groups such as ITE-All Students and ITE-All Staff in the same way.

Organise these groups into a structure that makes applying policy easy to manage using inheritance to do most of the work. Try and set the majority of settings close to the root group.

When you have finished you have created a policy tree but without any objects.

Now drop your production user and device groups (Year 11) at the point of the tree where you want policy to be applied. Unlike the ITE groups you never set policy on these groups that hold users or devices, they are just containers into which the policy flows.

As users are added or removed from the production group the policy will change without any intervention on your part. At the end of year simply move production groups to a new point in the policy tree to allow the inherited policies take effect.

Google admins will recognise this pattern from the G Suite org tree. Working with InTune we are simply replacing sub-organisations for groups and then making sure that a device or user only ever exists within one group in the tree. That way you can ensure consistent results and you have a system that’s far easier to troubleshoot... and it works just like Google.

InTune for Chromebook Admins (p3).