Why Power Platform licensing is complex, part 4: the role of CDS

Three months ago I set myself a goal to publish a 4-part series on the many mysteries around Power Platform licensing models. I’ve been known to write blog posts that are a bit on the long side and sometimes it surprises myself when after announcing the new article on LinkedIn I see a “20 minute read” info tag added alongside what I though was just a quick exploration into a topic. So, learning to better pace myself is a skill worth practicing, and dividing my writings into more digestable portions is one such exercise.

To recap, the original outline of this series was to cover these sources of licensing complexity:

• Protection of intellectual property as well as development budgets across Microsoft product portfolio (part 1, done✔)
• The old world of apps as data silos & the concept of multiplexing (part 2, done✔)
• Light use scenarios for information workers vs. “Real Business Applications” (part 3, done✔)
• The role of CDS as the platform, not just a data source/target (part 4, you are here 📍)

The Common Data Service is more than a data service

For the majority of users and developers out there who work with Microsoft technologies on a daily basis, CDS is not yet a “common” part of the toolkit. It’s not a visually sexy thing like Power Apps pixel-perfect UI’s, nor an superglue that connects different cloud services like Power Automate. There’s nowhere near as much flare in the data part as with Power BI. It’s not (in itself) an intelligent thinking AI machine with fancy algorithms either. For many it simply looks like a boring database that sits somewhere in the MS Cloud.

These people got it all wrong, of course, but it’s hardly their fault. After all, “CDS” sounds a bit like “SQL” and the term often isn’t used in any other context than when you’re planning the data storage part of your solution architecture. What CDS really should have been called is “Power Platform”, because essentially it is the single place that represents all the platform elements that makers, developers and admins can take advantage of when building on Microsoft’s low-code application platform.

The above image is the best illustration of the many dimensions of Common Data Service that I have yet come across in Microsoft’s slide decks. “Yeah, that’s pretty much all the important stuff you get with CDS”, I was thinking for a long time. Until someone pointed out to me that this picture is actually missing one obvious thing: solutions! That’s perhaps the single most important thing that CDS from its XRM origins is bringing to the table. Without solutions, what Microsoft would otherwise have is just a collection of independent cloud services that each have a completely separate way to manage their configuration, with no way to package these components into one logical container covering the functionality of an app, nor any valid ALM story for enterprise scenarios. In short, Power Platform would not be a platform without the solution model that’s powered by CDS.

The fact that we can so easily miss a critical part of the Common Data Sevice while talking about it just goes to show how much of a foundational part CDS is in the business applications offering. It’s not just one icon in the cloud architecture pictures. In fact, we also should keep in mind how many different Azure services CDS runs on, to deliver its feature set to users, makers and developers:

Yes, there is of course a whole bunch of deep tech stack found behind each and every icon in that picture too – say CosmosDB, for example. It is “big”, in the same way that CDS is “big”. What I think makes the Common Data Service special in this discussion is how it can abstract all that techy stuff from Azure and allow citizen developers (like myself) to control it indirectly, either via our own configuration or through the predefined patterns built by MS.

Essentially CDS is a value-added layer that brings much of the cool new cloud technology into my business apps while I’m sleeping. What I mean by this is that the architecture evolves to cover new requirements and open up new possibilities, all the while the original business logic and data of the app is preserved. If I create a custom entity, like I did for the first time in 2007, it will keep gaining new superpowers every year as application platform beneath it matures and evolves. All I have to do is keep paying for the service.

How do you price such a thing then?

The answer: in all the wrong ways. At least that’s what it often looks like at that moment when the cost factor is brought into the discussion in Power Apps business scenario planning.

The more central CDS becomes in the app story of MS Cloud, the more dissatisfaction there seems to be on the impact that its premium licensing model has on business decisions. There have been many debates on Twitter during the past few months, with valid arguments and counterarguments presented between the participants that range from customers to Microsoft 365 pros to Dynamics 365 pros to MSFT employees. Threads like this highlight an important aspect:

If you were to invent a brand new service out of the blue, had zero customers and then started to plan the pricing model for it, life would be easy. If, on the other hand, you take an online service that has been offered to customers for over a decade now, then try and mold that into whole new use cases, you’re inevitably going to cause some anger among the existing customer base.

In the above example, Neil’s point about the CDS storage cost increase in the new pricing model that moved from a flat GB rate to database/log/file storage type separation is entirely true: some part of the bill will go up 8x and that quite frankly sucks. Imagine if that would happen to the price of gasoline and your fuel costs would suddenly be 8 times higher. But hold on: the price of diesel went down considerably at the same time, and filling the tank with natural gas like LNG is now practically FREE!!! “Well whoop-de-doo, I’ll just switch my old gasoline Volkswagen to use LNG the next time I need to stop at the station.” Yeah, real life comes with a bit more friction to change the system that powers your vehicle, or your business processes.

Going back to my CDS custom entity example, though, if we think about the types of cars that Volkswagen sold back in 2007 vs. in 2020, of course there are slight differences in what the product is. The transition to electric cars is a very apt metaphor when thinking about the difference between XRM and CDS. Even though the Volkswagen engineers were able to create an electic version of the popular VW Golf model and sell the e-Golf alongside its combustion engine powered cousins since 2014 already, that particular path wasn’t going to get them far enough. In order to remain a competitive vehicle manufacturer in the global game, Volkswagen has had to design a pure electric compact car in the form of ID.3, and struggle with the many issues that come with brand new technology (also on the software side).

Similarly, it wasn’t ever really enough for Microsoft to just take XRM to the cloud and host it there (which they did a decade ago already). A lot of R&D budget has surely been spent in building the new foundation of Power Platform. At the same time, some existing features that the loyal fans of the classic VW Golf would have enjoyed could not be brought over to the ID.3, because there just isn’t a logical place for the combustion history to co-exist with the electric future all within the same frame. Witness the same pain with CDS, Power Apps and Power Automate never quite reaching full parity with what was available in the world that came before them.

Aside from just going electric, there’s one even bigger existential question the car makers are facing: do people even want to purchase cars in 2020 or would they prefer to just pay for the transportation services that they need at any given time? A bit like the dilemma of the light use scenarios, whereby it would make more business sense to lower the barrier of entry by not requiring the purchase of a full user license for the complete platform but rather selling tickets to enter the app when the need arises. This growth in revenue wouldn’t necessarily come from the existing customer base that pays more – it could be a whole new revenue stream made possible by the way the technology as well as the world out there has evolved.

Think about a simple, timely example: customer self-service terminals. In the aftermath of COVID-19, our day-to-day interactions with humans directly are being greatly reduced due to regulations on keeping a safe distance when roaming out there in the physical world. Instead of talking to clerks and cashiers, we are now acquiring the same information and completing the same transactions by using touch screens that allow the customer to do the data entry and retrieval themselves. Now, Power Apps Canvas apps would of course be the perfect fit for quickly creating touch friendly tablet UIs for various self-service scenarios. There’s only one blocker: Canvas apps don’t have a licensing model for kiosk scenarios where the person interacting with the app isn’t equipped with a user license for Power Apps. Doh!

It’s hard to imagine Microsoft not wanting to expand the footprint of Power Platform to more and more scenarios that aren’t tied to named users with dedicated licenses. Yes, Portals can already do that, Power Virtual Agents can of course serve anonymous customers, Power BI can even be embedded inside third party applications. It’s that direct interaction with the CDS entities through a user identity mapped into security roles that isn’t quite as flexible yet in the current world… because like I’ve been telling you all along, Power Platform licensing is complex for the customers, the partners and surely Microsoft themselves as well.

Life finds a way, licensing may follow

The good news is that CDS as a technology is doing better than ever, thanks to the critical role it is playing in so many different Microsoft products these days. It’s further expanding inside the Dynamics 365 portfolio via Dual-write. Data platform is embracing it via the native sync to Azure Data Lake. It’s the ALM backbone for all Power products (aside from Power BI – and of course PowerPoint!). Project got rebuilt on top of CDS to replace the earlier SharePoint based architecture. It’s perfectly feasible to imagine more and more MS apps and products gravitating towards this growing platform and starting to orbit around the central body that is CDS.

That still leaves open the question of the customer specifc business apps that are getting built every day as Power Apps Canvas apps. In order to ensure the citizen developers don’t just default to using SharePoint Lists (soon to be called Microsoft Lists) as their data storage solution, MS needs to find ways to lower the barriers to go with CDS instead. There’s the technical barrier from the long legacy of CRM requirements, then we have this licensing complexity. Even though the recent launch of Per App passes as a concept has in some high value scenarios made the business case calcuations easier, the vast majority of true citizen developer apps won’t ever even reach that stage where official calculations for a specific app’s ROI would be analyzed in detail. For this long tail of app development, there should just be ample CDS capacity available to be used for the purposes that the business experts themselves have identified as worthy to spend their time on (which often may cost a lot more than the actual software licenses).

Capacity based licensing of course wouldn’t make the tools free either, it just changes the way the final bill is calculated. The example of CDS storage pricing change is a reflection of exactly what happens when trying to find non-user based billing metrics that would reflect the rate at which the platform is used. Looking at the Power Platform price points, Microsoft may well feel that it’s perfectly OK to sell the CDS database capacity at $40/GB/month when their traditional CRM rival Salesforce has set their list price per GB to be a whopping $250 (which I don’t think they publicly advertise anywhere, so your real mileage may vary). Both figures have nothing to do with how much it actually costs to store data in the cloud, they are purely intended for building a pricing curve where the growth in platform usage (and hopefully business value) results in growth of revenue to the service provider.

The more we have capacity based pricing of Power Platform features, the bigger the issue of capacity limits and their enforcement will become in solution architecture design. API calls are much more difficult to estimate than user counts, so there is bound to be some increasing complexity ahead for scenarios where CDS and other Power Platform tools are used in a “pay per request” model. MS will likely continue to bundle quotas of API and storage capacity into their own software products that use CDS and where the revenue is attained via other means, in an effort to simplify the experience for that one product. The challenge that remains is the more holistic usage of an application platform, where not only do you buy the Apps but use features powered by CDS in your own unique ways.

Other parts in this article series can be found here:

• Why Power Platform licensing is complex, part 1: products
• Why Power Platform licensing is complex, part 2: multiplexing
• Why Power Platform licensing is complex, part 3: light use scenarios