Exploring Azure AD Privileged Identity Management (PIM) – Part 4 – Access Review and Azure RBAC

Exploring Azure AD Privileged Identity Management (PIM) – Part 4 – Access Review and Azure RBAC

Access Reviews

Welcome to my final post on Azure Active Directory Privileged Identity Management (AAD PIM).  Over this series of posts I’ve provided an overview of the service, guidance on how to set the service up, and a deep dive and look at the user and approver experience.  We’ll wrap up the series by looking at the Access Review feature, take an intermission to look at a new feature, and wrap up with reviewing the Azure RBAC integration.

We have a lot to cover, so let’s jump into it.

As a quick refresher, I’ll be using my Journey Of the Geek tenant.  Within the tenant I have some Office 365 E5 and EMS E5 licenses provisioned.  Our admin user will be initiating the access review and Homer Simpson will acting as a reviewer.

I first log into the Azure Portal as the admin user and open up the AAD PIM shortcut from my dashboard.  Once the application opens, I’m going to navigate to the Azure AD directory roles option.

4aadpim1

After selecting the option my main menu is refreshed to show the management options for the various AAD PIM features.  As a quick refresher, let’s look at the settings I’ve configured for Access Reviews in my tenant.  We navigate to those Settings by clicking the Settings option as seen below and selecting Access Reviews.

4aadpim2

As you can see from the settings in the screenshot below, my tenant is set to send mail notifications to reviewers when a review is started and to admins when it finishes.  It’s also configured for reminders to be sent out to reviewers who haven’t yet completed their review.  I’ve configured reviewers to provider a reason as to why continued access to a privileged role needs to be maintained.  This is a great little option to capture the business requirements behind the access.  Finally, my access reviews are configured to run a total of 30 days.

4aadpim3

Let’s navigate back to the Access Review blade under the management menu.

4aadpim4

On the Access Reviews blade we see a listing of the access reviews in progress.  You can see I setup an access review for users that are members of the Global Admins role.  On the top we have the menu options to start a new access review, filter what access reviews are displayed, change the way they are grouped, and go back to the Access Review settings I showed earlier.

4aadpim5

et’s spin up a new access review for users who are permanent or eligible members of the User Administrators Azure AD role.  We click the Add link and a new blade opens where we can configure a number of options.  We have the basic options of naming the access review, providing a description, and a start and end date.

I’ve selected the User Administrator role as the role being reviewed during this access review.  Notice the Scope option with the Everyone radio button.  Perhaps that’s a placeholder for functionality that will be introduced in the future to limit the users within a role that the access review will cover.  I’ve selected Homer Simpson to be the reviewer for the role.  The advanced settings have inherited the global settings for my tenant for access reviews I covered previously.  Once the information is filled in, I hit the start button to kick off the access review.

4aadpim6

It takes a few minutes for the access review to be created and then it’s displayed in the listing of access reviews with a status of active.

4aadpim7

If we navigate over to Homer Simpson’s Outlook inbox, we see he has received an email informing him an access review has been kicked off and he has been designated as a reviewer and must approve or reject other members’ continued eligibility for the role.

4aadpim8

If we delay acting on the access review for a day we receive another reminder email per our settings.  The email can be seen below.

4aadpim9

If the approvers do not respond to the access review, the review completes but records that none of the users have been reviewed.

4aadpim10.png

Let’s spin up another review and complete this one.

4aadpim11

Homer Simpson again receives the notice that an Access Review has been kicked off.  Clicking the Start Review button in the email opens up the Azure Portal and the AAD PIM blade.  Here Homer gets an overview of the access review including the user who created the review, the length of the review, the description of the review, and the users who are members (permanent or eligible) for the role.

The filter option allows us to filter on the listing of users based upon whether they still need to be reviewed or have been approved or denied.

4aadpim12

We first check off Bart Simpson and see that we are required to input a reason for Bart’s approval or denial.  I input a reason and choose the deny button.  Bart disappears from the menu.  If I use the filter option to show all three categories of users, Bart now reappears under the denied category.

4aadpim13

I check off both Homer and Marge and provide a reason for both users and hit the approve button. All users have been reviewed by Homer Simpson. After refreshing the page the review now shows 0 users remaining to be reviewed.

4aadpim14

Switching over to the browser for the admin user we see that the access review is still open.

4aadpim15

If we open the access review we can see that all users have been reviewed even though the review is still active.  We have the option to Reset the access review to force the approvers to perform the access review activities again or we can stop it.  We’re going to choose end the access review early now that all the reviews have been completed.

4aadpim16

Re-opening the access review, we now have the option to apply the results of it.  After clicking he Apply button the changes are applied and we’re notified via the Portal notification system.

4aadpim17

Navigating to the roles blade under the Manage section now shows only Homer and Marge as being eligible for the User Administrator role verifying that the changes made during the access review have taken effect.

The access review feature is a wonderful addition by Microsoft.  Back in the olden days of Windows Active Directory, managing the entire lifecycle of an identity and its entitlements often involved complex third-party identity management solutions in combination with request management system.  By including this feature out of the gates, Microsoft is showing a real maturity in its identity offerings.

A Brief Intermission

Before I get into what AAD PIM can do for Azure RBAC, I want to touch on a new feature that went into public preview while I was working on this post.  Notice in the Manage section the Roles blade now has a (Preview) notation after it.

4aadpim18

Navigating into the blade shows an entirely new interface with far more useful information.  We now have a complete list of the roles AAD PIM can manage including descriptions.  If we select a role we go a level deeper and can add users to the role as we would expect.

4aadpim19

We also have two new menu options for Description and Definition.  The Description blade opens up and gives us a link to the Microsoft documentation on the role as well as every permissions the role has (AWESOME!).  The Definition blade gives us a JSON view of the role information.  Perhaps we’ll be able to create custom AAD / O365 roles in the future and we’ll be able to use these JSON views as ARM templates?  Time will tell.

4aadpim20

The introduction of this new feature is a great demonstration of how quickly things change in the cloud.

AAD PIM and Azure RBAC

Most organizations consuming Microsoft cloud services don’t just consume Office 365.  These organizations want to yield the benefits of the infrastructure-as-a-Service (IaaS) and platform-as-a-Service (PaaS) services provide by Microsoft’s Azure offering.  Managing authorization in Azure is handled through Azure Role-Based Access Control (RBAC).  In short, Azure RBAC provides a method of authorizing a security principal (user, group, or service principal) to perform an action on a resource (VM, storage account, Azure SQL, etc) based upon membership in a role.  Out of the box Microsoft provides a few roles such as owner, reader, and contributor.  You can also create custom roles to fit your business needs.

 

Similar to Office 365 prior to AAD PIM, preventing standing access of security principals in Azure RBAC roles was left to custom scripts and third-party solutions.  Last year it was announcedthat AAD PIM capabilities were being extended to Azure RBAC.  The integration of AAD PIM and Azure RBAC become generally available in the commercial offering of Azure AD in May of 2018.

For this demonstration I’m going to switch over to my Geek In The Weeds tenant.  Recall that the tenant is a synchronized and federated tenant using Azure AD Connect and Active Directory Federation Services.  I’ve already activated AAD PIM for the tenant so I’ll be jumping right into its integration with Azure RBAC.

After logging into the portal as a user who has permanent membership in the Privileged Role Administrator role I’m faced with the standard admin view of AAD PIM.  In the Manage menu I’m going to select the Azure resources option.

4aadpim21

If this is your first time using AAD PIM with Azure RBAC you’ll need to go through the discovery stage.  This will discover Azure Resources that you have write permissions to and thus have the ability to manage privileged access to.  After discovery is complete you’ll see a screen similar to the below.  You can see that my user is a member of the owners role for the Visual Studio Enterprise Azure subscription and that there are 77 roles defined for the subscription with three security principals holding one or more roles.

4aadpim22

Selecting the subscription resource gives us a dashboard displaying key metrics about PIM activity within the subscription.

4aadpim23.png

One of the metrics that caught my eye was the single user in the User Access Administrator role.  Selecting that area of the dashboard opens a new blade which lists out the members of the role.  We can see the service principal for PIM has been added to the User Access Administrator role to grant the service permissions to administer the roles within the resource (in this case a subscription).

4aadpim24

Notice also that the PIM menu for managing Azure AD/Office 365 differs for the menu for managing Azure RBAC.  We see that the new Role options I outlined above haven’t been migrated to the Azure RBAC integration yet.  Additionally we see that the request approval workflow is still in public preview in Azure RBAC.  In the Azure RBAC menu we also get a Resource Audit log which details PIM activity within the resource.

4aadpim25

Notice also that the general Settings option isn’t present in the Azure RBAC menu.  Instead we have a Role Settings option.  Selecting this option opens a new blade that lists out the Roles associated with the Resource.  Selecting any of the resources opens a new blade where we have the options of configuring a large selection of options for the role for both assignment (making the user eligible or a permanent member) as well as activation.  If you recall the configurable options for the Azure AD / Office 365 roles, these are far more granular.  The additional flexibility makes sense because these roles are going to managing IaaS and PaaS resources which are much more catered to programmatic access by non-humans.  The non-human access tends to be much more predictable than human access, so enforcing controls such as temporary eligibility for a role makes a lot of sense.

4aadpim26

Let’s take a look at what the experience is adding a user to one of the RBAC roles.  The process is very similar to AAD PIM with Azure AD / Office 365 in that we select the Roles option from the Manage section.  For this demonstration I’m going to add a user to the Virtual Machine Contributor role.

Clicking the Add Member option allows me to assign Ash Williams as an eligible member of the role.  Notice the additional option called Set membership settings.  Here I can set a timespan that Ash is eligible for the role.  This option isn’t available in AAD PIM for Azure AD / Office 365 that I could see.

4aadpim27

After hitting the add button Ash is successfully added as a Direct member to the role.  Notice that I can also add groups as members of the Role.  This is another capability unit to the Azure RBAC integration.

4aadpim28

Let’s go through the user experience for activating a role.  For the sake of simplicity I’m going to cover differences in the user experience.  You can reference my third post if you’re curious of the full user experience.

At this point I’ve logged into a virtual machine as Ash Williams and have authenticated to the Azure Portal.  I’ve entered the Azure resources blade.  Here we see the user being informed that no Azure resources are protected by PIM.  In this instance hitting the Discover resources permission will not update this menu because Ash Williams isn’t a member of any role that would grant him write permissions on an Azure Resource.  Instead I’m going to click the Activate Role button.

4aadpim29

After clicking the Activate role button I’m shown the roles Ash Williams is eligible to activate.  Notice Ash has the ability to activate the role due to both his direct membership and his membership in the GIW AIP Users group.  I’d recommend leveraging groups for this access where possible so you don’t get in the situation where you grant a security principal longer access to the role than you wanted due to a direct role assignment situation.

4aadpim30

he activation experience and approval experience is the same from this point forward so I’m going stop here.

Summing It Up

I really enjoyed this blog series.  I hadn’t done a deep dive into AAD PIM since it was in public preview and much has changed since then.  I really like how Microsoft is finally exposing capabilities which have historically been more Azure AD / Office 365 centric to Microsoft Azure.  It’s an excellent marketing tool for companies who may already be using Office 365 but are using another cloud provider for IaaS and PaaS. The product team has also done great job integrating much needed features such as approval workflows, access reviews, and metrics.

I’m not going to have the time to do a post about the AAD PIM PowerShell module but I recommend you check it out if you have some bandwidth.  There are some great opportunities there to integrate PIM functionality with third party workflow management tools to automate the entire user experience behind a GUI you users are already familiar with.

That wraps up my series on Azure AD Privileged Identity Management.  I hope you enjoyed it as much as I did.

See you next post!

Exploring Azure AD Privileged Identity Management (PIM) – Part 3 – Deep Dive

Exploring Azure AD Privileged Identity Management (PIM) – Part 3 – Deep Dive

Welcome back fellow geeks to my third post on my series covering Azure AD Privileged Identity Management (AAD PIM).  In my first post I provided an overview of the service and in my second post I covered the initial setup and configuration of PIM.  In this post we’re going to take a look at role activation and approval as well as looking behind the scenes to see if we can figure out makes the magic of AAD PIM work.

The lab I’ll be using consists of a non-domain joined Microsoft Windows 10 Professional version 1803 virtual machine (VM) running on Hyper V on my home lab.  The VM has a local user configured that is a member of the Administrators group.  I’ll be using Microsoft Edge and Google Chrome as my browsers and running Telerik’s Fiddler to capture the web conversation.  The users in this scenario will be sourced from the Journey Of The Geek tenant and one will be licensed with Office 365 E5 and EMS E5 and the other will be licensed with just EMS E5.  The tenant is not synchronized from an on-premises Windows Active Directory.  The user Homer Simpsons has been made eligible for the Security Administrators role.

With the intro squared away, let’s get to it.

First thing I will do is navigate to the Azure Portal and authenticate as Homer Simpson.  As expected, since the user is not Azure MFA enforced, he is allowed to authenticate to the Azure Portal with just a password.  Once I’m into the Azure Portal I need to go into AAD PIM which I do from the shortcut I added to the user’s dashboard.

3pim1.png

Navigating to the My roles section of the menu I can see that the user is eligible to for the Security Administrator Azure Active Directory (AAD) role.

3pim2

Selecting the Activate link opens up a new section where the user will complete the necessary steps to activate the role.  As you can see from my screenshot below, the Security Administrator role is one of the roles Microsoft considers high risk and enforces step-up authentication via Azure MFA.  Selecting the Verify your identity before proceeding link opens up another section that informs the user he or she needs to verify the identity with an MFA challenge.  If the user isn’t already configured for MFA, they will be setup for it at this stage.

3pim3.png

Homer Simpson is already configured for MFA so after the successful response to the MFA challenge the screen refreshes and the Activation button can now be clicked.

3pim4.png

After clicking the Activation button I enter a new section where I can configure a custom start time, configuration an activation duration (up to the maximum configured for the Role), provide ticketing information, and provide an activation reason..  As you can see I’ve adjusted the max duration for an activation from the default of one hour to three hours and have configured a requirement to provide a ticket number.  This could be mapped back to your internal incident or change management system.

3pim5.png

After filling in the required information I click the Activate button, the screen refreshes back to the main request screen, and I’m informed that activation for this role requires approval.  In addition to modifying the activation and requiring a ticket number, I also configured the role to require approval.

3pim6.png

At this point I opened an instance of Google Chrome and authenticated to Azure AD as a user who is in the privileged role administrator role.  Opening up AAD PIM with this user and navigating to the My roles section and looking at the Active roles shows the user is a permanent member of the Security Administrators, Global Administrators, and Privileged Role Administrators roles.

3pim7.png

I then navigate over to the Approve requests section.  Here I can see the pending request from Homer Simpson requesting activation of the Security Administrator role.  I’m also provided with the user’s reason and start and end time.  I’d like to see Microsoft add a column for the user’s ticket number.  My approving user may want to reference the ticket for more detail on why the user is requesting the role

3pim8.png

At this point I select the pending request and click the Approve button.  A new section opens where I need to provide the approval reason after which I hit the Approve button.

3pim9.png

After approving the blue synchronization-like image is refreshed to a green check box indicating the approval has been process and the user’s role is now active.

3pim10

If I navigate to My audit history section I can see the approval of Homer’s request has been logged as well as the reasoning I provided for my approval.

3pim11.png

If I bounce back to the Microsoft Edge browser instance that Homer Simpsons is logged into and navigate to the My requests and I can see that my activation has been approved and it’s now active.

3pim12.png

At this point I have requested the role and the role has been approved by a member of the Privileged Role Administrators role.  Let’s try modifying an AIP Policy.  Navigating back to Homer Simpsons dashboard I select the Azure Information Protection icon and receive the notification below.

3pim13.png

What happened?  Navigating to Homer Simpsons mailbox shows the email confirming the role has been activated.

3pim14.png

What gives?  To figure out the answer to that question, I’m going to check on the Fiddler capture I started before logging in as Homer Simpson.

In this capture I can see my browser sending my bearer token to various AIP endpoints and receiving a 401 return code with an error indicating the user isn’t a member of the Global Administrators or Security Administrators roles.

3pim15.png

I’ll export the bearer token, base64 decode it and stick it into Notepad. Let’s refresh the web page and try accessing AIP again. As we can see AIP opens without issues this time.

3pim16.png

At this point I dumped the bearer token from the failure and the bearer token from a success and compared the two as seen below.  The IAT, NBF, and EXP are simply speak to times specific to the claim.  I can’t find any documentation on the aio or uti claims.  If anyone has information on those two, I’d love to see it.

3pim17.png

I thought it would be interesting at this point to deactivate my access and see if I could still access AIP.  To deactivate a role the user simply accesses AAD PIM, goes to My Roles and looks the Active Roles section as seen below.

3pim18.png

After deactivation I went back to the dashboard and was still able to access AIP.  After refreshing the browser I was unable to access AIP.  Since I didn’t see any obvious cookies or access tokens being created or deleted.  My guess at this point is applications that use Azure AD or Office 365 Roles have some type of method of receiving data from AAD PIM.  A plausible scenario would be an application receives a bearer token, queries Azure AD to see if the user is in one a member of the relevant roles for the application.  Perhaps for eligible roles there is an additional piece of information indicating the timespan the user has the role activated and that time is checked against the time the bearer token was issued.  That would explain my experience above because the bearer token my browser sent to AIP was obtained prior to activating my role.  I verified this by comparing the bearer token issued from the delegation point at first login to the one sent to AIP after I tried accessing it after activation.  Only after a refresh did I obtain a new bearer token from the delegation endpoint.

Well folks that’s it for this blog entry.  If you happen to know the secret sauce behind how AAD PIM works and why it requires a refresh I’d love to hear it!  See you next post.

Exploring Azure AD Privileged Identity Management (PIM) – Part 2 – Setup

Exploring Azure AD Privileged Identity Management (PIM) – Part 2 – Setup

Welcome back to part 2 of my series on Azure Active Directory Privileged Identity Management (AAD PIM).  In the first post I covered the basics of the service.  If you haven’t read it yet, take a few minutes to read through it because I’ll be jumping right into using the service going forward.  In this post I’m going to cover the setup process for AAD PIM.

Before you can begin using AAD PIM, you’ll need to purchase a license that includes the capability.  As we saw in my last post, at this time that means a standalone Azure AD Premium P2 or Enterprise Mobility + Security E5 license.  Once the license is registered as being purchased by your tenant, you’ll be able to setup AAD PIM.

Your first step is to log into the Azure Portal.  After you’ve logged in you’ll want to click the Create a Resource button and search for Azure AD Privileged Identity Management.

1pim1.png

Select the search result and AAD PIM application will be displayed with the Create button.  Click the create button to spin the service up for your tenant.

1pim2.png

It will only take a few seconds and you’ll be informed the service has successfully been spun up and you’ll be given the option to add a link to your dashboard.

1pim3.png

The global admin who added AAD PIM to the tenant will become the first member of the Privileged Role Administrator role.  This is a new role that was introduced with the service.  Members of this role are your administrators of AAD PIM and has full read and write access to it.  Beware that other global admins, security administrators, and security readers only have read access to it.  As soon as you successfully spin up the service, you’ll want to add another Privileged Role Administrator as a backup.

Upon opening AAD PIM for the first time, you’ll receive a consent page as seen below.  The consent process requires confirmation of the user’s identity using Azure MFA.  If the user isn’t enabled for it, it will be configured at this point.

1pim4.png

After successfully authenticating with Azure MFA. The screen will update to show the status check was completed as seen below. This is a great example of Microsoft exercising the concept of step-up authentication. The user may have authenticated to the Azure Portal with a password or perhaps a still-valid session cookie. By prompting for an Azure MFA challenge the assurance of the user being the real user is that much higher thus reducing the risk of the user accessing such sensitive configuration options.

1pim5

After clicking the Consent button the service becomes fully usable.  The primary menu options are displayed as seen in the picture below.  For the purpose of this post we’re going to click on the Azure AD directory roles option under the Manage section.

1pim6.png

The Manage section of the menu is refreshed and a number of new options are displayed.  Before I jump into the Wizard, I’ll navigate through each option in the section to explain its purpose.

1pim7.png

The Roles option gives us a view of all of the users who are members of privileged roles within Azure AD  and Office 365.  In the activation column it’s shown as to whether or not the user is a permanent or eligible admin.  The expiration column shows any user that is eligible and has actively requested and been approved for temporary access to the privileged role.  As you can see from my screenshot from my test tenant I have a number of users in the global admin roles which is a big no no.  We’ll remediate that in a bit using the Wizard.

1pim8.png

Selecting the Add user button brings up a new screen where new users can be configured for privileged roles.  Microsoft has done a good job of giving AAD PIM the capability of managing a multitude of Azure AD and Office 365 roles.  Adding users to roles through this tool will make automatically make the user an eligible for the role rather than a permanent member like through other means would.

1pim9.png

The Filter button allows for robust filtering options including the permission state (all, eligible, permanent), activation state (all, active, or inactive), and by role.  The Refresh button’s function is obvious and the group option allows you to group the data either by user or by role.  The Review button allows you to kick off an access review which we’ll cover in a later post.  Lastly we have the Export button which exports the data to a CSV.

The Users option under the Manage section presents the same options as the Roles option except it takes a user-centric view.

The Alerts option under the Manage section displays the alerts referenced here.  You can see it is alerting me to the fact I have too many permanent global admins configured for my tenant.  I also have the option to run a manual scan rather than waiting on the next automatic scan.

1pim10.png

The Access Reviews option under the Manage section is used to create new access review.  I’ll cover the capability in a future post.

Skipping over the Wizard option for a moment, we have the Settings option.  Here we can configure a variety of settings for roles, alerts, and access reviews.

Let’s dig into the settings for roles first.

1pim11.png

Here we can configure the default settings for all roles as well as settings specific to one role.  When a user successfully activates a privileged role, the membership in that role is time bound with a default of one hour.  If after doing some baselining we find one hour is insufficient, we could bump it up to something higher.  We can also configure notifications to notify administrators of activation of a role.  There is also the option to require an incident or request reference that may map back to an internal incident management or request management system.  Azure MFA can be required when a user activates a role.  You’ll want to be aware that the MFA setting is automatically enforced for roles Microsoft views as critical such as global administrator.

Finally we have the option to require an approval.  If you’ve played around with AAD PIM since preview, you may remember the approval workflow.  For some reason the product team removed it when AAD PIM original went general available.  This effectively meant users could elevate their access whenever they wanted.  Sure they weren’t permanent members but there were no checks and balances.  For organizations with a high security posture it made AAD PIM of little value and forced the on-demand management of privileged roles to be done using complicated PowerShell scripts or third-party tools that were integrated with the Graph API.  It’s wonderful to see the product team responded to customer feedback and has added the feature back.

Toggling to Enable for the require approval option adds a section where you can select approvers for requests for the role.

1pim12.png

Moving on to the Alerts settings we have the ability to configure parameters for some of the alerting as can be seen from the examples below.

1pim13.png

The default values for the configurable thresholds around the “There are too many global administrator” should be a good wake up call to organizations as to the risk Microsoft associates with global admin access.  The thresholds for the “Roles are being activated too frequently” should be left as the default until the behavior of your user base is better understood.  This will help you to identify deviations from standard behavior indicating a possible threat as well as to identify opportunities to improve the user experience by bumping up the activation time span for users holding privileged roles that the hour long default activation time is insufficient.

Lastly we have Access Review settings.  Here we can enable or disable mail notifications to reviewers are the beginning and end of an access review.  Reminders can also be sent to reviewers if they have no completed a review they are a part of.  A very welcome feature of requiring reviewers to provide reasons for approvals of a review.  This can be helpful to capture business requirements as to why a user needs continued access to a role.  Finally, the default access review duration can be adjusted.

1pim14

Now on to the Wizard.  The Wizard is a great tool to use when you first configure AAD PIM in order to get it up and running and begin capturing behavioral patterns.  The steps within the Wizard are outlined below.

1pim15.png

The Discover privileged roles step displays a simple summary of the privileged roles in use and the amount of permanent and eligible users.  We can see from the below my tenant has exceeded either the 3 global admins or greater than 10% of users default thresholds for the “There are too many global admins” alert.  Selecting any of the roles displays a listing of the users holding permanent or eligible membership in the role.

1pim16.png

Clicking the next button bring us to the “Convert users to eligible step” where we can begin converting permanent members to eligible members. From a best practices perspective, you should ensure you keep at least two permanent members in the Privileged Role Administrator role to avoid being locked out if one account becomes unavailable. You can see that I’m making Ash Williams and Jason Voorhies eligible members of the global admins group.

1pim17.png

After clicking the Next button I’m moved to the “Review the changes to your users in the privileged roles” step.  I commit the changes by hitting the OK button and my two users are now setup as eligible members of the roles.

1pim18.png

As you’ve seen throughout the post AAD PIM is incredibly easy to configure.  I firmly believe that the only successful security solutions moving forward will be solutions that are simple to use and transparent to the users.  These two traits will allow security professionals to focus less of their time on convoluted solutions and more time working directly with the business to drive real value to the organization.

I’m going to start something new with a quick bulleted list of key learning points that I came across while performing the lab and doing the research for the post.

  • AAD PIM can be configured after the first Azure AD Premium P2 or EMS E5 license is associated with the tenant
  • Be aware that at this time Microsoft does not enforce a technical control to prevent all users from benefiting from PIM but the licensing requirements require an individual license for each user benefitting from the feature.  Make sure you’re compliant with the licensing requirements and don’t build processes around what technical controls exist today. They will change.
  • Once AAD PIM is activated by the first global admin, immediately assign a second user permanent membership in the Privileged Role Administrators role.

That’s it folks.  In the next post in my series I’ll take a look at what the user experience is like for a requestor and approver.  I’ll also look at some Fiddler captures to see I capture any detail as to how/if the modified privileges are reflected in the logical security token.

Thanks!

 

Exploring Azure AD Privileged Identity Management (PIM) – Part 1

Exploring Azure AD Privileged Identity Management (PIM) – Part 1

We’re going to take a break from Azure Information Protection and shift our focus to Azure Active Directory Privileged Identity Management (AAD PIM).

If you’ve ever had to manage an application, you’re familiar with the challenge of trying to keep a balance between security and usability when it comes to privileged access.  In many cases you’re stuck with users that have permanent membership in privileged roles because the impact to usability of the application is far too great to manage that access on an “as needed basis” or as we refer to it in the industry “just in time” (JIT).   If you do manage to remove that permanent membership requirement (often referred to as standing privileged access) you’re typically stuck with a complicated automation solution or a convoluted engineering solution that gives you security but at the cost of usability and increasing operational complexity.

Not long ago the privileged roles within Azure Active Directory (AAD), Office 365 (O365), and Azure Role-Based Access Control had this same problem.  Either a user was a permanent member of the privileged role or you had to string together some type of request workflow that interacted with the Graph API or triggered a PowerShell script.  In my first entry into Azure AD, I had a convoluted manual process which involved requests, approvals, and a centralized password management system.  It worked, but it definitely impacted productivity.

Thankfully Microsoft (MS) has addressed this challenge with the introduction of Azure AD Privileged Identity Management (AAD PIM).  In simple terms AAD PIM introduces the concept of an “eligible” administrator which allows you to achieve that oh so wonderful JIT.  AAD PIM is capable of managing a wide variety of roles which is another area Microsoft has made major improvements on.  Just a few years ago close to everything required being in the Global Admin role which was a security nightmare.

In addition to JIT, AAD PIM also provides a solid level of logging and analytics, a centralized view into what users are members of privileged roles, alerting around the usage of privileged roles, approval workflow capabilities (love this feature), and even provides an access review capability to help with access certification campaigns.  You can interact with AAD PIM through the Azure Portal, Graph API, or PowerShell.

To get JIT you’ll need an Azure Active Directory Premium P2 or Enteprise Mobility and Security E5 license.  Microsoft states that every use that benefits from the feature requires a license.  While this is a licensing requirement, it’s not technically enforced as we see in my upcoming posts.

You’re probably saying, “Well this is all well and good Matt, but there is nothing here I couldn’t glean from Microsoft documentation.”  No worries my friends, we’ll be using this series to walk to demonstrate the capabilities so you can see them in action.  I’ll also be breaking out my favorite tool Fiddler to take a look behind the scenes of how Microsoft manages to elevate access for the user after a privileged role has been activated.

 

The Evolution of AD RMS to Azure Information Protection – Part 7 – Deep Dive into cross Azure AD tenant consumption

The Evolution of AD RMS to Azure Information Protection – Part 7 – Deep Dive into cross Azure AD tenant consumption

Each time I think I’ve covered what I want to for Azure Information Protection (AIP), I think of another fun topic to explore.  In this post I’m going to look at how AIP can be used to share information with users that exist outside your tenant.  We’ll be looking at the scenario where an organization has a requirement to share protected content with another organization that has an Office 365 tenant.

Due to my requirements to test access from a second tenant, I’m going to supplement the lab I’ve been using.  I’m adding to the mix my second Azure AD tenant at journeyofthegeek.com.  Specific configuration items to note are as follows:

  • The tenant’s custom domain of journeyofthegeek.com is an Azure AD (AAD)-managed domain.
  • I’ve created two users for testing.  The first is named Homer Simpson (homer.simpson@journeyofthegeek.com) and the second is Bart Simpson (bart.simpson@journeyofthegeek.com).
  • Each user have been licensed with Office 365 E3 and Enterprise Mobility + Security E5 licenses.
  • Three mail-enabled security groups have been created.  The groups are named The Simpsons (thesimpsons@journeyofthegeek.com), JOG Accounting (jogaccounting@journeyofthegeek.com), and JOG IT (jogit@journeyofthegeek.com).
  • Homer Simpson is a member of The Simpsons and JOG Accounting while Bart Simpson is a member of The Simpsons and JOG IT.
  • Two additional AIP policies have been created in addition to the Global policy.  One policy is named JOG IT and one is named JOG Accounting.
  • The Global AIP policy has an additional label created named PII that enforces protection.  The label is configured to detect at least one occurrence of a US social security number.  The document is protection policy allows only members of the The Simpsons group to the role of viewer.
  • The JOG Accounting and JOG IT AIP policies have both been configured with an additional label of either JOG Accounting or JOG IT.  A sublabel for each label has also been created which enforces protection and restricts members of the relevant departmental group to the role of viewer.
  • I’ve repurposed the GIWCLIENT2 machine and have created two local users named Bart Simpson and Homer Simpson.

Once I had my tenant configuration up and running, I initialized Homer Simpson on GIWCLIENT2.  I already had the AIP Client installed on the machine, so upon first opening Microsoft Word, the same bootstrapping process I described in my previous post occurred for the MSIPC client and the AIP client.  Notice that the document has had the Confidential \ All Employees label applied to the document automatically as was configured in the Global AIP policy.  Notice also the Custom Permissions option which is presented to the user because I’ve enabled the appropriate setting in the relevant AIP policies.

7aip1.png

I’ll be restricting access to the document by allowing users in the geekintheweeds.com organization to hold the Viewer role.  The geekintheweeds.com domain is associated with my other Azure AD tenant that I have been using for the lab for this series of posts.  First thing I do is change the classification label from Confidential \ All Employees to General.  That label is a default label provided by Microsoft which has an RMS Template applied that restricts viewers to users within the tenant.

One interesting finding I discovered through my testing is that the user can go through the process of protecting with custom permissions using a label that has a pre-configured template and the AIP client won’t throw any errors, but the custom permissions won’t be applied.  This makes perfect sense from a security perspective, but it would be nice to inform the user with an error or warning.  I can see this creating unnecessary help desk calls with how it’s configured now.

When I attempt to change my classification label to General, I receive a prompt requiring me to justify the drop in classification.  This is yet another setting I’ve configured in my Global AIP policy.  This seems to be a standard feature in most data classification solutions from what I’ve observed in another major vendor.

7aip2.png

After successfully classifying the document with the General label protection is removed from the document. At this point I can apply my custom permissions as seen below.

7aip3.png

I repeated the process for another protected doc named jog_protected_for_Ash_Williams.docx with permissions restricted to ash.williams@geekintheweeds.com.  I packaged both files into an email and sent them to Ash Williams who is a user in the Geek In The Weeds tenant.  Keep in mind the users in the Geek In The Weeds tenant are synchronized from a Windows Active Directory domain and use federated authentication.

After opening Outlook the message email from Homer Simpson arrives in Ash William’s inbox.   At this point I copied the files to my desktop, closed Outlook, opened Microsoft Word and used the “Reset Settings” options of the AIP client, and signed out of my Office profile.

7aip4

At this point I started Fiddler and opened one of the Microsoft Word document. Microsoft Word pops-up a login prompt where I type in my username of ash.williams@geekintheweeds.com and I’m authenticated to Office 365 through the standard federated authentication flow. The document then pops open.

7aip5.png

Examining the Fiddler capture we see a lot of chatter. Let’s take a look at this in chunks, first addressing the initial calls to the AIP endpoint.

7aip6

If you have previous experience with the MSIPC client in the AD RMS world you’ll recall that it makes its calls in the following order:

  1. Searches HKLM registry hive
  2. Searches HKCU registry hive
  3. Web request to the RMS licensing pipeline for the RMS endpoint listed in the metadata attached to the protected document

In my previous deep dives into AD RMS we observed this behavior in action.  In the AIP world, it looks like the MSIPC client performs similarly.  The endpoint we see it first contacting is the Journey of the Geek which starts with 196d8e.

The client first sends an unauthenticated HTTP GET to the Server endpoint in the licensing pipeline. The response the server gives is a list of available SOAP functions which include GetLicensorCertificate and GetServerInfo as seen below.

7aip8.png

The client follows up the actions below:

  1. Now that the client knows the endpoint supports the GetServerInfo SOAP function, it sends an unauthenticated HTTP POST which includes the SOAP action of GetServerInfo.  The AIP endpoint returns a response which includes the capabilities of the AIP service and the relevant endpoints for certification and the like.
  2. It uses that information received from the previous request to send an unauthenticated HTTP POST which includes the SOAP action of ServiceDiscoveryForUser.  The service returns a 401.

At this point the client needs to obtain a bearer access token to proceed.  This process is actually pretty interesting and warrants a closer look.

7aip9

Let’s step through the conversation:

  1. We first see a connection opened to odc.officeapps.live.com and an unauthenticated HTTP GET to the /odc/emailhrd/getfederationprovider URI with query strings of geekintheweeds.com.  This is a home realm discovery process trying to the provider for the user’s email domain.

    My guess is this is MSAL In action and is allowing support for multiple IdPs like Azure AD, Microsoft Live, Google, and the like.  I’ll be testing this theory in a later post where I test consumption by a Google user.

    The server responds with a number of headers containing information about the token endpoints for Azure AD (since this is domain associated with an Azure AD tenant.)

    7aip10.png

  2. A connection is then opened to odc.officeapps.live.com and an unauthenticated HTTP GET to the /odc/emailhrd/getidp with the email address for my user ash.williams@geekintheweeds.com. The response is interesting in that I would have thought it would return the user’s tenant ID. Instead it returns a JSON response of OrgId.

    7aip11.png

    Since I’m a nosey geek, I decided to unlock the session for editing.  First I put in the email address associated with a Microsoft Live.  Instead of OrgId it returned MSA which indicates it detects it as being a Microsoft Live account.  I then plugged in a @gmail.com account to see if I would get back Google but instead I received back neither.  OrgId seems to indicate that it’s an account associated with an Azure AD tenant.  Maybe it would perform alternative steps depending on whether it’s MSA or Azure AD in future steps?  No clue.

  3. Next, a connection is made to oauth2 endpoint for the journeyofthegeek.com tenant. The machine makes an unathenticated requests an access token for the https://api.aadrm.com/ in order to impersonate Ash Williams. Now if you know your OAuth, you know the user needs to authenticate and approve the access before the access token can be issued. The response from the oauth2 endpoint is a redirect over to the AD FS server so the user can authenticate.

    7aip12.png

  4. After the user successfully authenticates, he is returned a security token and redirected back to login.microsoftonline.com where the assertion is posted and the user is successfully authenticated and is returned an authorization code.

    7aip13.png

  5. The machine then takes that authorization code and posts it to the oauth2 endpoint for my journeyofthegeek.com tenant. It receives back an Open ID Connect id token for ash.williams, a bearer access token, and a refresh token for the Azure RMS API.

    7aip14.png

    Decoding the bearer access token we come across some interesting information.  We can see the audience for the token is the Azure RMS API, the issuer of the token is the tenant id associated with journeyofthegeek.com (interesting right?), and the identity provider for the user is the tenant id for geekintheweeds.com.

    7aip15.png

  6. After the access token is obtained the machine closes out the session with login.microsoftonline.com and of course dumps a bunch of telemetry (can you see the trend here?).

    7aip16.png

  7. A connection is again made to odc.officeapps.live.com and the /odc/emailhrd/getfederationprovider URI with an unauthenticated request which includes a query string of geekintheweeds.com. The same process as before takes place.

Exhausted yet?  Well it’s about to get even more interesting if you’re an RMS nerd like myself.

7aip17.png

Let’s talk through the sessions above.

  1. A connection is opened to the geekintheweeds.com /wmcs/certification/server.asmx AIP endpoint with an unauthenticated HTTP POST and a SOAP action of GetServerInfo.  The endpoint responds as we’ve observed previously with information about the AIP instance including features and endpoints for the various pipelines.
  2. A connection is opened to the geekintheweeds.com /wmcs/oauth2/servicediscovery/servicediscovery.asmx AIP endpoint with an unauthenticated HTTP POST and a SOAP action of ServiceDiscoveryForUser.  We know from the bootstrapping process I covered in my last post, that this action requires authentication, so we see the service return a 401.
  3. A connection is opened to the geekintheweeds.com /wmcs/oauth2/certification/server.asmx AIP endpoint with an unauthenticated HTTP POST and SOAP action of GetLicensorCertificate.  The SLC and its chain is returned to the machine in the response.
  4. A connection is opened to the geekintheweeds.com /wmcs/oauth2/certification/certification.asmx AIP endpoint with an unauthenticated HTTP POST and SOAP action of Certify.  Again, we remember from my last post that this requires authentication, so the service again responds with a 401.

What we learned from the above is the bearer access token the client obtained earlier isn’t attended for the geekintheweeds.com AIP endpoint because we never see it used.  So how will the machine complete its bootstrap process?  Well let’s see.

  1. A connection is opened to the journeyofthegeek.com /wmcs/oauth2/servicediscovery/servicediscovery.asmx AIP endpoint with an unauthenticated HTTP POST and SOAP action of ServiceDiscoveryForUser.  The service returns a 401 after which the client makes the same connection and HTTP POST again, but this time including its bearer access token it retrieved earlier.  The service provides a response with the relevant pipelines for the journeyofthegeek.com AIP instance.
  2. A connection is opened to the journeyofthegeek.com /wmcs/oauth2/certification/server.asmx AIP endpoint with an authenticated (bearer access token) HTTP POST and SOAP action of GetLicensorCertificate.  The service returns the SLC and its chain.
  3. A connection is opened to the journeyofthegeek.com /wmcs/oauth2/certification/certification.asmx AIP endpoint with an authenticated (bearer access token) HTTP POST and SOAP action of Certify.  The service returns a RAC for the ash.williams@geekintheweeds.com along with relevant SLC and chain.  Wait what?  A RAC from the journeyofthegeek.com AIP instance for a user in geekintheweeds.com?   Well folks this is supported through RMS’s support for federation.  Since all Azure AD’s in a given offering (commercial, gov, etc) come pre-federated, this use case is supported.
  4. A connection is opened to the journeyofthegeek.com /wmcs/licensing/server.asmx AIP endpoint with an uauthenticated HTTP POST and SOAP action of GetServerInfo.  We’ve covered this enough to know what’s returned.
  5. A connection is opened to the journeyofthegeek.com /wmcs/licensing/publish.asmx AIP endpoint with an authenticated (bearer access token) HTTP POST and SOAP action of GetClientLicensorandUserCertificates.  The server returns the CLC and EUL to the user.

After this our protected document opens in Microsoft Word.

7aip18.png

Pretty neat right? Smart move by Microsoft to take advantage and build upon of the federated capabilities built into AD RMS. This is another example showing just how far ahead of their game the product team for AD RMS was. Heck, there are SaaS vendors that still don’t support SAML, let alone on-premises products from 10 years ago.

In the next few posts (can you tell I find RMS fascinating yet?) of this series I’ll explore how Microsoft has integrated AIP into OneDrive, SharePoint Online, and Exchange Online.

Have a great week!

The Evolution of AD RMS to Azure Information Protection – Part 4 – Preparation and Server-Side Migration

The Evolution of AD RMS to Azure Information Protection – Part 4 – Preparation and Server-Side Migration

The time has finally come to get our hands dirty.  Welcome to my fourth post on the evolution of Active Directory Rights Management Service (AD RMS) to Azure Information Protection (AIP).  So far in this series I’ve done an overview of the service, a comparison of the architectures, and covered the key planning decisions that need to place when migrating from AD RMS to AIP.  In this post I’ll be performing the preparation and server-side migration steps for the migration from AD RMS to AIP.

Microsoft has done a wonderful job documenting the steps for the migration from AD RMS to AIP within the migration guide.  I’ll be referencing back to the guide as needed throughout the post.  Take a look at my first post for a refresher of my lab setup.  Take note I’ll be migrating LAB2 and will be leaving LAB1 on AD RMS.  Here are some key things to remember about the lab:

  • There is a forest trust between the JOG.LOCAL and GEEKINTHEWEEDS.COM Active Directory Forests
  • AD RMS Trusted User Domains (TUDs) have been configured on both JOG.LOCAL and GEEKINTHEWEEDS.COM

I’ve created the following users, groups, and AD RMS templates (I’ve been on an 80s/90s movies fix, so enjoy the names).

  • GEEKINTHEWEEDS.COM CONFIGURATION
    • (User) Jason Voorhies
      • User Principal Name Attribute: jason.voorhies@geekintheweeds.com
      • Mail Attribute: jason.voorhies@geekintheweeds.com
      • Group Memberships: Domain Users, GIW Employees, Information Technology
    • (User) Theodore Logan
      • User Principal Name Attribute: theodore.logan@geekintheweeds.com
      • Mail Attribute: theodore.logan@geekintheweeds.com
      • Group Memberships: Domain Users, GIW Employees, Information Technology
    • (User) Ash Williams
      • User Principal Name Attribute: jason.voorhies@geekintheweeds.com
      • Mail Attribute: ash.williams@geekintheweeds.com
      • Group Memberships: Domain Users, GIW Employees, Accounting
    • (User) Michael Myers
      • User Principal Name Attribute: michael.myers@geekintheweeds.com
      • Mail Attribute: michael.myers@geekintheweeds.com
      • Group Memberships: Domain Users, GIW Employees, Accounting
    • (Group) Accounting
      • Mail Attribute: giwaccounting@geekintheweeds.com
      • Group Type: Universal Distribution
    • (Group) GIW Employees
      • Mail Attribute: giwemployees@geekintheweeds.com
      • Group Type: Universal Distribution
    • (Group) Information Technology
      • Mail Attribute: giwit@geekintheweeds.com
      • Group Type: Universal Distribution
    • (Group) GIW AIP Users
      • Group Type: Global Security
    • (AD RMS Template) GIW Accounting
      • Users: giwaccounting@geekintheweeds.com
      • Rights: Full Control
    • (AD RMS Template) GIW Employees
      • Users: giwemployees@geekintheweeds.com
      • Rights: View, View Rights
    • (AD RMS Template) GIW IT
      • Users: giwit@geekintheweeds.com
      • Rights: Full Control
  • JOG.LOCAL CONFIGURATION

    • (User) Luke Skywalker
      • User Principal Name Attribute: luke.skywalker@jog.local
      • Mail Attribute: luke.skywalker@jog.local
      • Group Memberships: Domain Users, jogemployees
    • (User) Han Solo
      • User Principal Name Attribute: han.solo@jog.local
      • Mail Attribute: han.solo@jog.local
      • Group Memberships: Domain Users, jogemployees
    • (Group) jogemployees
      • Mail Attribute: jogemployees@jog.local
      • Group Type: Universal Distribution

After my lab was built I performed the following tests:

  • Protected Microsoft Word document named GIW_LS_ADRMS with GEEKINTHEWEEDS.COM AD RMS Cluster and successfully opened with Luke Skywalker user from JOG.LOCAL client machine.
  • Protected Microsoft Word document named GIW_GIWALL_ADRMS with GEEKINTHEWEEDS.COM AD RMS Cluster and GIW Employees template and unsuccessfully opened with Luke Skywalker user from JOG.LOCAL client machine.
  • Protected Microsoft Word document named GIW_JV_ADRMS with GEEKINTHEWEEDS.COM AD RMS Cluster using Theodore Logan user and opened successfully with Jason Voorhies user from GEEKINTHEWEEDS.COM client machine.
  • Protected Microsoft Word document named JOG_MM_ADRMS with JOG.LOCAL AD RMS Cluster using Luke Skywalker user and opened successfully with Michael Myers user from GEEKINTHEWEEDS.COM client machine.
    Protected Microsoft Word document named GIW_ACCT_ADRMS with GEEKINTHEWEEDS.COM AD RMS Cluster and GIW Accounting template and was unsuccessful in opening with Jason Voorhies user from GEEKINTHEWEEDS.COM client machine.

These tests verified both AD RMS clusters were working successfully and the TUD was functioning as expected.  The lab is up and running, so now it’s time to migrate to AIP!

Our first step is to download the AADRM PowerShell module from Microsoft.  I went the easy route and used the install-module cmdlet.

AIP4PIC1.png

Back in March Microsoft announced that AIP would be enabled by default on any eligible tenants with O365 E3 or above that were added after February.  Microsoft’s migration guide specifically instructs you to ensure protection capabilities are disabled if you’re planning a migration from AD RMS to AIP.  This means we need to verify that AIP is disabled.  To do that, we’re going to use the newly downloaded AADRM module to verify the status.

AIP4PIC2.png

As expected, the service is enabled.  We’ll want to disable the service before beginning the migration process by running the Disable-Aadrm cmdlet.  After running the command, we see that the functional state is now reporting as disabled.

AIP4PIC3.png

While we have the configuration data up we’re going to grab the value (minus _wmcs/licensing) in the  LicensingIntranetDistributionPointUrl property.  We’ll be using this later on in the migration process.

In most enterprise scenarios you’d want to perform a staged migration process of your users from AD RMS to AIP.  Microsoft provides for this with the concept of onboarding controls.  Onboarding controls allow you to manage who has the ability to protect content using AIP even when the service has been enabled at the tenant level.  Your common use case would be creating an Azure AD-managed or Windows AD-synced group which is used as your control group.  Users who are members of the group and are licensed appropriately would be able to protect content using AIP.  Other users within the tenant could consume the content but not protect it.

In my lab I’ll be using the GIW AIP Users group that is being synchronized to Azure AD from my Windows AD as the control group.  To use the group I’ll need to get its ObjectID which is the object’s unique identifier in Azure AD.  For that I used the Get-AzureADGroup cmdlet within Azure AD PowerShell module.

AIP4PIC6

Microsoft’s migration guide next suggests come configuration modifications to Windows computers within the forest.  I’m going to hold off on this for now and instead begin the server-side migration steps.

First up we’re going to export the trusted publisher domains (TPDs) from the AD RMS cluster.  We do this to ensure that users that have migrated over to AIP are still able to open content that was previously protected by the AD RMS cluster.  The TPD includes the Server Licensor Certificate (SLC) keys so when exporting them we protect them with a password and create an XML file that includes the SLC keys and any custom AD RMS rights policy templates.

AIP4PIC7.png

Next we import the exported TPD to AIP using the relevant process based upon how we chose to protect the cluster keys.  For this lab I used a software key (stored in the AD RMS database) so I’ll be using the software key to software key migration path.  Thankfully this path is quite simple and consists of running another cmdlet from the AADRM PowerShell module.  In the first command we store the password used to protect the TPD file as a secure string and use the Import-AadrmTpd cmdlet to pull it into AIP.  Notice the resulting data provides the cluster friendly name, indicates the Cryptographic Mode was set to 1, the key was a Microsoft-managed (aka software key) and there were three rights policy templates attached to the TPD.

Keep in mind that if you have multiple TPDs for some reason (let’s say you migrated from Cryptographic Mode 1 to Cryptographic Mode 2) you’ll need to upload each TPD separately and set the active TPD using the Set-AadrmKeyProperties cmdlet.

AIP4PIC8.png

Running a Get-AadrmTemplate shows the default templates Microsoft provides you with as well as the three templates I had configured in AD RMS.

AIP4PIC9

The last step of the server side of the process is to activate AIP.  For that we use the Enable-Aadrm cmdlet from the AADRM PowerShell module.

AIP4PIC10

At this point the server-side configuration steps have been completed and AIP is ready to go.  However, we still have some very important client-side configuration steps to perform.  We’ll cover those steps in my post.

Have a great week!

AWS and Microsoft’s Cloud App Security

AWS and Microsoft’s Cloud App Security

It seems like it’s become a weekly occurrence to have sensitive data exposed due to poorly managed cloud services.  Due to Amazon’s large market share with Amazon Web Services (AWS) many of these instances involve publicly-accessible Simple Storage Service (S3) buckets.  In the last six months alone there were highly publicized incidents with FedEx and Verizon.  While the cloud can be empowering, it can also be very dangerous when there is a lack of governance, visibility, and acceptance of the different security mindset cloud requires.

Organizations that have been in operation for many years have grown to be very reliant on the network boundary acting as the primary security boundary.  As these organizations begin to move to a software defined data center model this traditional boundary quickly becomes less and less effective.  Unfortunately for these organizations this, in combination with a lack of sufficient understanding of cloud, gives rise to mistakes like sensitive data being exposed.

One way in which an organization can protect itself is to leverage technologies such as cloud access security brokers (cloud security gateways if you’re Forrester reader) to help monitor and control data as it travels between on-premises and the cloud.  If you’re unfamiliar with the concept of a CASB, I covered it in a previous entry and included a link to an article which provides a great overview.

Microsoft has its own CASB offering called Microsoft Cloud App Security (CAS).  It’s offered as part of Microsoft’s Enterprise Mobility and Security (EMS) E5/A5 subscription.  Over the past several months multiple connectors to third party software-as-a-service (SaaS) providers have been introduced, including one for AWS.  The capabilities with AWS are limited at this point to pulling administrative logs and user information but it shows promise.

As per usual, Microsoft provides an integration guide which is detailed in button pushing, but not so much in concepts and technical details as to what is happening behind the scenes.  Since the Azure AD and AWS blog series has attracted so many views, I thought it would be fun and informative to do an entry for how Cloud App Security can be used with AWS.

I’m not in the habit of re-creating documentation so I’ll be referencing the Microsoft integration guide throughout the post.

The first thing that needs doing is the creation of a security principal in AWS Identity and Access Management (AWS IAM) that will be used by your tenant’s instance of CAS to connect to resources in your AWS account.   The first four steps are straightforward but step 5 could a bit of an explanation.

awscas1.pngHere we’re creating a custom IAM policy for the security principal granting it a number of permissions within the AWS account.  IAM policies are sets of permissions which are attached to a human or non-human identity or AWS resource and are evaluated when a call to the resource is made.  In the traditional on-premises world, you can think of it as something somewhat similar to a set of NTFS file permissions.  When the policy pictured above is created the security principal is granted a set of permissions across all instances of CloudTrail, CloudWatch, and IAM within the account.

If you’re unfamiliar with AWS services, CloudTrail is a service which audits the API calls made to AWS resources.  Some of the information included in the events include the action taken, the resource the action was taken upon, the security principal that made the action, the date time, and source IP address of the security principal who performed the action.  The CloudWatch service allows for monitoring of metrics and optionally triggering events based upon metrics reaching specific thresholds.  The IAM service is AWS’s identity store for the cloud management layer.

Now that we have a basic understanding of the services, let’s look at the permissions Microsoft is requiring for CAS to do its thing.  The CloudTrail permissions of DescribeTrails, LookupEvents, and GetTrailStatus allow CAS to query for all trails enabled on an AWS account (CloudTrail is enabled by default on all AWS resources), lookup events in a trail, and get information about the trail such as start and stop logging times.  The CloudWatch permissions of Describe* and Get* are fancy ways of asking for  READ permissions on CloudWatch resources.  These permissions include describe-alarms-history, describe alarms, describe-alarms-for-metric, get-dashboard, and get-metric-statistics.  The IAM permissions are similar to what’s being asked for in CloudWatch, basically asking for full read.

Step number 11 instructs us to create a new CloudTrail trail.  AWS by default audits all events across all resources and stores them for 90 days.  Trails enable you to direct events captured by CloudTrail to an S3 bucket for archiving, analysis, and responding to events.

awscas2.png

The trail created is consumed by CAS to read the information captured via CloudTrail.  The permissions requested above become a bit more clear now that we see CAS is requesting read access for all trails across an account for monitoring goodness.  I’m unclear as to why CAS is asking for read for CloudWatch alarms unless it has some integration in that it monitors and reports on alarms configured for an AWS account.  The IAM read permissions are required so it can pull user  information it can use for the User Groups capability.

After the security principal is created and a sample trail is setup, it’s time to configure the connector for CAS.  Steps 12 – 15 walk through the process.  When it is complete AWS now shows as a connected app.

awscas3.png

After a few hours data will start to trickle in from AWS.  Navigating to the Users and Accounts section shows all of the accounts found in the IAM instance for my AWS account.  Envision this as your one stop shop for identifying all of the user accounts across your many cloud services.  A single pane of glass to identity across SaaS.

awscas4.png

On the Activity Log I see all of the API activities captured by CloudTrail.  If I wanted to capture more audit information, I can enable CloudTrail for the relevant resource and point it to the trail I configured for CAS.  I haven’t tested what CAS does with multiple trails, but based upon the permissions we configured when we setup the security principal, it should technically be able to pull from any trail we create.

awscas7.png

Since the CAS and AWS integration is limited to pulling logging information, lets walk through an example of how we could use the data.  Take an example where an organization has a policy that the AWS root user should not be used for administrative activities due to the level of access the account gets by default.  The organization creates AWS IAM users accounts for each of its administrators who administer the cloud management layer.  In this scenario we can create a new policy in CAS to detect and alert on instances where the AWS root user is used.

First we navigate to the Policies page under the Control section of CAS.

awscas6.png

On the Policies page we’re going to choose to create a new policy settings in the image below.  We’ll designate this as a high severity privileged account alert.  We’re interested in knowing anytime the account is used so we choose the Single Activity option.

awscas7

We’ll pretend we were smart users of CAS and let it collect data for a few weeks to get a sampling of the types of events which are captured and to give us some data to analyze.  We also went the extra mile and leveraged the ability of CAS to pull in user information from AWS IAM such that we can choose the appropriate users from the drop-down menus.

Since this is a demo and my AWS lab has a lot of activity by the root account we’re going to limit our alerts to the creation of new AWS IAM users.  To do that we set our filter to look for an Activity type equal to Create user.  Our main goal is to capture usage of the root account so we add another filter rule that searches for a User with the name equal to aws root user where it is the actor in an event.

awscas8.png

Finally we configure the alert to send an email to the administrator when the event occurs.  The governance capabilities don’t come into play in this use case.

awscas9

Next we jump back to AWS and create a new AWS IAM user named testuser1.  A few minutes after the user is created we see the event appearing in CloudTrail.

awscas10

After a few minutes, CAS generates and alert and I receive an email seen in the image below.   I’m given information as to the activity, the app, the date and time it was performed, and the client’s IP address.

awscas11.png

If I bounce back to CAS I see one new Alert.  Navigating to the alert I’m able to dismiss it, adjust the policy that generated it, or resolve it and add some notes to the resolution.

awscas12.png

I also have the option to dig deeper to see some of the patterns of the user’s behavior or the pattern of the behaviors from a specific IP address as seen below.

awscas13.png

awscas14.png

All this information is great, but what can we do with it?  In this example, it delivers visibility into the administrative activities occurring at the AWS cloud management layer by centralizing the data into a single repository which I can then send other data such as O365 activity, Box, SalesForces, etc.  By centralizing the information I can begin doing some behavioral analytics to develop standard patterns of behavior of my user base.  Understanding standard behavior patterns is key to being ahead of the bad guys whether they be insiders or outsiders.  I can search for deviations from standard patterns to detect a threat before it becomes too widespread.  I can also be proactive about putting alerts and enforcement (available for other app connectors in CAS but not AWS at this time) to stop the behavior before the threat is realized.  If I supplemented this data with log information from my on-premises proxy via Cloud App Discovery, I get an even larger sampling improving the quality of the data as well as giving me insight into shadow IT.  Pulling those “shadow” cloud solutions into the light allow me to ensure the usage of the services complies with organizational policies and opens up the opportunity of reducing costs by eliminating redundant services.

Microsoft categorizes the capabilities that help realize these benefits as the Discover and Investigate capabilities of CAS. The solution also offers a growing number of enforcement mechanisms (Microsoft categorized these enforcement mechanisms as Control) which add a whole other layer of power behind the solution.  Due to the limited integration with AWS I can’t demo those capabilities with this post.  I’ll cover those in a future post.

I hope this post helped you better understand the value that CASB/CSGs like Microsoft’s Cloud App Security can bring to the table.  While the product is still very new and a bit sparse on support with 3rd party applications, the list is growing every day. I predict the capabilities provided by technology such as Microsoft’s Cloud App Security will be as standard to IT as a firewall in the years to come.  If you’re already in Office 365 you should be ensuring you integrate these capabilities into your arsenal to understand the value they can bring to your business.

Thanks and have a wonderful week!