Tag: governance

Default Environment – How to handle?

As we’re all aware, the default (Power Platform) environment in any Azure tenant is a very ‘interesting’ thing to have. It’s there by default when an Azure tenant is created, all users within the Azure tenant automatically have access to it, we’re not able to restrict users from being in it, etc etc.

Though it’s able to be backed up, it’s not able to be restored over itself, there’s no SLA/support available on it….the list goes on & on…!

Many of us have come up against issues caused by people using the default environment whilst not knowing about challenges involving it, which usually results in pulling out our hair, banging our head against the wall, and other like-minded productive approaches.

However, it is the first place that users, being new to Power Platform, land up, and instinctively they’ll start building applications, automations etc within it (though usually without using solutions as a container for the development of items). So to date, there’s not really anything that’s been able to be done around this, apart from monitoring users & chasing them after the fact.

Now, we’re all about enabling our users in the right way, helping educate & support them. Telling them a big NO doesn’t help, and can even be an initial blocker to having people start playing around & building technological solutions.

So how can we go about enabling our users, but also having the appropriate level of governance over the top? Well, there are several steps that I think we can take, which will help us with these. Now, not all of these are yet in place, though they have been talked about publicly. So let’s go take a look at them

  1. The first step, in my mind, is to start off with enabling the default environment as a managed environment (yes, this can ACTUALLY be done!). Managed environments have many different properties associated with them, but the one of most interest (for this at least) is the requirement to have a premium license in place.

All users within an organisation should by default have an M365 license SKU against them (usually this would be an E3 or E5). Users with these can immediately use the seeded Power Platform capabilities within them to create Power Platform collateral (using standard connector capabilities). However, with the default environment being managed, they will NOT be able to access it!

Note: For the moment, I’m leaving out users who have premium Power Platform licenses – this is deliberate

  1. Environment routing. Announced recently is the environment routing capabilities. This will enable users to be automatically routed to an appropriate environment, based on various conditions that can be set. With this, we could create appropriate business unit ‘sandboxes’, and we could route users to these. The user experience would be that when logging in, they would automatically then go to the right environment, rather than trying to work out which environment they should actually go to. This will save on confusion, and be a good user experience (in my opinion).
  1. Just-In-Time (JIT) Environment Creation. One of the items mentioned by Charles Lamanna at the European Power Platform Conference 2023 in Dublin is a new capability that’s coming in soon (I hope!). From the sound of it, this will give the ability to automatically create a new environment for users who do not already have one.

This sounds really cool. With the recent advent of Development Environments (& the ability for all users to have multiples of these), this could work REALLY well with the environment routing capability mentioned above. When a user would log in for the first time, it could look to see if they have a developer environment – if yes, then route them to it. But if the user didn’t, then to automatically spin up & create a new developer environment, and route them to it.

Now there are some caveats with this approach, leaving aside that some of the functionality isn’t GA yet.

It would mean that organisations would need to be alright with changing the default environment to become a managed environment. Obviously, risk assessments would need to be carried out with this, and non-premium solutions migrated elsewhere.

It’s also important to call out that organisations which have a CDS 1.0 implementation (ie before Power Platform became GA etc) will only have the ability to upgrade default to managed. They are not able to downgrade back to an unmanaged default environment, given limitations of the original CDS implementation (I’ve heard some truly HORRIFIC stories around this, so be careful!)

The above, however, is just the start of things. There are many other concepts to keep in mind, such as Landing Zones, Policies, etc. I’m going to be looking to cover these in upcoming posts, so keep an eye out for them!

Developer Environment Deletion!

Strong title for a blog post, right? Well, I did want to catch your attention! So what exactly are we talking about here?

For the last few years, it’s been possible for users to sign up for a ‘Developer’ plan, which gives them a full capability Power Platform environment for free (though with some limitations to them). This used to be be called the ‘Community’ plan, and is an amazing resource for everyone, whether they’re a professional or citizen developer, to have their own personal ‘sandpit’ to play in, and try things out.

Let’s wind back a few months in time now – earlier this year, Microsoft announced that users would be able to create THREE of these Developer environments, rather than having just a single one! This was mind blowing news, and something that has been extremely welcomed. If you’re wanting to see more on the announcement, Phil Topness has a great video on it at Dataverse Environments For Everyone – New Developer Plan – Power CAT Live – YouTube.

Incidentally, I’m curious as to how much storage space Microsoft has in the background to handle these. After all, each environment takes up a minimum of 1GB of space (& can grow to 2 GB). That means that each user could have 6GB of storage being used….which when multiplied, gives a VERY large number!

Microsoft has now announced that these developer environments, however, need to be utilisied. Ie if they’ve been created, but aren’t being used, Microsoft is going to delete them! Now, from a certain perspective, this is actually quite good – after all, there are all of the storage considerations for environments that have been created, but not being used. However from a different perspective, this could be a problem. What about if you’re doing something occasionally in an environment, but not too often? What about if you decide to go on a ‘Round the World’ cruise for several months?

So let’s look at the definition for this. Microsoft states that an environment is considered to be inactive when it hasn’t been used for 90 days. At that point in time, it is disabled, and the administrator or environment owner is notified. If there is no action taken within the next 30 days, then the developer environment is automatically deleted.

Now, how does Microsoft define ‘Activity’? It goes something like this:

  • User activity: Launch an app, execute a flow (whether automatic or not), chat with a Power Virtual Agents bot
  • Maker activity: Create, read, update, or delete an app, flow (desktop and cloud flows), Power Virtual Agents bot, custom connector
  • Admin activity: Environment operations such as copy, delete, back up, recover, reset

The above is all user driven – ie a user needs to interact with something within the environment. However, it’s also important to note how automation is viewed:

  • Activity includes automated behaviors such as scheduled flow runs. For example, if there’s no user, maker, or admin activity in an environment, but it contains a cloud flow that runs daily, then the environment is considered active.

It’s also important to note that at this point in time, the above only applies to Developer environments. Other types of environment (Production, Sandbox etc) don’t have any auto-deletion policies called out for them – well, at least not yet (if something does pop up around these, I’ll definitely look to talk about them too!).

So to answer our question above about what happens with a (developer) environment that is only being used infrequently – the way to stop it being auto-deleted is to put some automation in place. This doesn’t need to be lightweight – it can be something simple & easy, just to ensure that the environment registers activity happening within it.

In my view, it would be nice to have some granularity & control over this as well – allowing organisations to set their own deletion policies. We have this in place for things like audit log retention – it would be nice to have have it in here too.

The story of MFA & the Centre of Excellence

I’ve been rolling out the Microsoft Centre of Excellence solution for several years now at customers. It’s a great place to start getting a handle on what exactly is going on within a Power Platform tenant, though there’s obviously so much more that takes place within a Centre of Excellence team.

The solution gathers telemetry around environments, Power Apps, Power Automates etc through the usage of the Power Automate Admin connectors for Power Platform (see Power Platform for Admins – Connectors | Microsoft Learn for further information on these).

Now obviously we need a user account to run these, and this usually has been through the use of a ‘pseudo service account’, as using a service principal has been tricky, to say the least. So we would get customers to set up an appropriate account with licensing & permissions in place, and use this to own & run the Power Automate flows that bring in the information to the CoE solution.

It is important to note that usage of these connectors do require a pretty high level of permissions – in fact, we usually suggest applying the Power Platform Admin security role (within the Microsoft 365 Admin Centre) to the user account. All good so far.

The tricky part has, to date, been around security. Organisations usually require (for good reasons) multi-factor authentication to be in place (aka MFA). Now this is fine for users logging in & accessing systems. However, it proves to be somewhat tricker for automations.

See, when a user logs in & authenticates through MFA, a token is stored to allow them to access systems. Automations can also use this. However the token will expire at some point (based on how each organisations has implemented MFA access/controls). When the token expires, the automations will stop running, and fail silently. There’s no prompt that the token has expired, and the only way of knowing is to take a look at the Power Automate flow history. This can be interesting though, as signing in (with the pseudo service account) will prompt for MFA authentication, and then everything will start running again!

So this has usually resulted in conversations with the client to politely point out that implementing MFA on the service account will mean that, at some point, the Power Automate flows are going to start failing. Discussions with security teams take place, mitigation using tools such as Azure Sentinel are implemented, and things move ahead (cautiously). It’s been, to date, the most annoying pain for the technical implementation (that I can think of at least, in my experience).

Now you’d think that a change in this would be shouted from the rooftops, people talking about it, social media blowing up, etc. Well, I was starting an implementation recently for a customer, and was talking to them around this, as I’d usually do. Imagine my surprise when Todd, one of the Microsoft technical people attached to the client, asked why we weren’t recommending MFA.

Taking a look at the online documentation, I noticed that something had slipped in. Finally there was the ability to use MFA!

Trawling back through the GitHub history (after all, I wanted to find out EXACTLY when this had slipped in), I discovered that it was a few months old. I was still very surprised that there hadn’t been more publicity around this (though definately a good incentive to write about it, and a great blog post to start off 2023 with!).

So moving forward, we’re now able to use MFA for the CoE user account. This is definately going to put a lot of mind at rest (especially those who are in security and/or governance). The specifics around the MFA implementation can be found at Conditional access and multi-factor authentication in Flow – Power Automate | Microsoft Learn – but it’s important to note that specific MFA policies will need to be set up & implemented for this account.

So, now the job will be to retro-fit this to all organisations that already have the CoE toolkit in place. Thankfully this shouldn’t be too difficult to do, and will most definitely enhance the security controls around it!

Have you implemented any mitigation in the past to handle non-MFA? I’m curious if you have – please drop a comment below!

Power Platform Capacity Monitoring

If I look back at customer engagements over the last few years around Power Platform, whether it was a new capability or an existing capability, there was ONE thing that stood out above all. This was the ability to be able to track capacity usage over time, and to be honest, most organisations weren’t really doing very well at it.

For those who are unaware, there are actually three different types of capacity present within Power Platform environments. These are:

  • Data
  • File
  • Log

Each one is used for a specific purpose – broadly speaking, File holds all attachements that are uploaded directly into Dataverse, Log is used for auditing purposes, and Data holds everything else (hence the name)!

Now this data is shown within the Power Platform Admin Centre, under the ‘Resources/Capacity’ section’. An example of this is:

There’s also a nice little breakdown of capacity allocation through licenses etc, which essentially shows where the available capacity has come from:

If we drill down a bit further, we can open up a specific environment, and see not only the overall usage per capacity type, but also which tables are consuming the most amount of data:

All of this is well & good so far, for someone wanting to take a look at what is currently happening. But this is a manual action – it is possible to manually export the data, but again, this isn’t automated.

It’s also not possible (at least not at this point in time) to query the underlying records that hold these values. So we’re a little stuck. If an organisation wanted to see historical data usage, and/or predict data trends (such as ‘how much capacity would we need to have in 6 months if we continued our scaling’), there’s no way to do this. At least not automatically – someone would need to store the values down manually, then report on it. A hassle, to say the least.

Now when it comes to looking overall at Power Platform, the Centre of Excellence Starter Toolkit is really quite amazing. The Microsoft PowerCAT team continue to iterate existing functionality within it, as well as bring new functionality as well.

At this point in time, however, it doesn’t have any capacity monitoring in it. Well, it sort of does – we can implement notifications to alert us when capacity reaches a certain value. But this doesn’t solve the challenge as laid out above.

So with this in mind, I set out to create a solution to handle it. I’ve always wanted to create some sort of tool for giving back to the community & helping others, and I saw this as my chance to do so (I’m in awe of the various XrmToolBox tool creators, for the record).

So, I’m releasing a capacity monitoring tool. I’m using GitHub as the host, and the repo can be accessed at https://github.com/thecrmninja/Power-Platform-Capacity-Monitoring (it was a learning experience as well as how to use GitHub as a source repository, as I’ve not done that before!).

Model-Driven App:

Reporting Dashboard:

This is just the first version – I have various ideas about how to iterate on it, and tweak functionality. Each release will include release notes & important information to be aware of (such as security needing to run it). Also importantly, thanks to the amazing Matt Collins-Jones for reviewing some of my work around this.

The audience for this tool is aimed at IT/Power Platform admins who are already familiar with the Microsoft CoE toolkit solution, and have appropriate access to it.

If you find any issues, please raise an appropriate GitHub Issue item, and I’ll look into it. Also, if you have any ideas that you think could be worthwhile, please feel free to suggest them!

Finally, I’d be interested in hearing how you think this could support you or your organisation – feel free to drop a comment below!

New Platform DLP Capabilities

DLP (or Data Loss Prevention) is a very important capability in the Power Platform. With being able to bring together multiple data sources, both within the Microsoft technology stack as well as from other providers gives users amazing capabilities.

However with such great capabilities comes great responsibility. Of course, we trust users to be able to make proper judgements as to how different data sources can be used together. But certain industries require proper auditing around this, and so being able to specify DLP policies are extremely important to any governance team.

Being able to set how data connectors can be used together (or, in the reverse, not used together) across both Power Apps as well as Power Automate flows is imperative in any modern organisation.

To date, Power Platform DLP capabilities have existed that allow us to be able to categorise connectors (whether Microsoft provided or custom) into three categories. These categories specify how the connectors are able to function – they’re able to work with other connections that are in the same category group, but cannot work with connectors that are in a different category group.

So for example, it’s been possible to allow a user to create a Power App or a Power Automate flow that interacts with data from Dataverse, but cannot interact with Twitter (in the same app or flow).

With this approach, it’s possible to create multiple DLP policies, and ‘layer’ them as needed (much like baking a 7 layer cake!) to give the functionality required per environment (or also at the tenant level).

Now this has been great, but what has been missing has been the ability to be more granular in the approach to this. What about if we need to read data from Twitter, but just push data out to Twitter?

Well, Microsoft has now iterated on the DLP functionality available! It’s important to note that this is per connector, and will depend on the capabilities of the connector. What we’re now able to do is to control the specific actions that are contained within a connector, and either allow or not allow them to be able to be utilised.

Let’s take the Twitter connector as an example:

We’re able to see all of the actions that the connector is capable of (the scroll bar on the side is a nice touch for connectors that have too many actions to fit on a single screen!). We’re then able to toggle each one to either allow or disallow it.

What’s also really nice are the options for new connector capabilities.

This follows in the footsteps of handling connectors overall – we’re able to specify which grouping they should come under (ie Business, Non-Business, or Blocked). As new connectors are released by Microsoft, we don’t need to worry that users will automatically get access to them.

So too with new actions being released for existing connectors (that we’ve already classified). We’re able to set whether we want them to be automatically allow, or automatically blocked. This means that we don’t need to be worried that suddenly a new connector action will be available for users to use, that they perhaps should not be using.

From my perspective, I think that any organisation that’s blocking one or more action capabilities for a connector will want this to be blocked by default, just to ensure that everything remains secure until they confirm whether the action should be allowed or not.

So I’m really pleased about this. The question did cross my mind as to whether it would be nice to be able to specify this on a per environment basis when creating a tenant-level policy, but I guess that this would be handled by creating multiple policies. The only issue I could see around this would be the number of policies that could need to be handled, and ensuring that they’re named properly!

Have you ever wanted these capabilities? How have you managed until now, and how do you think you’ll roll this out going forward? Drop a comment below – I’d love to hear!