I’m Just Gonna Leave This Here
Sometimes my value as a blogger is to make complex topics simple, or to deliver industry news to you, or to entertain you. Sometimes, it’s just to keep you posted on things that are happening in the industry.
So with that in mind, at Microsoft FabCon Barcelona this week, there was a partner day for Microsoft partners. From there, Eelco Drost (shout out to the smart & fun folks at MasterMinds) shared a slide:
The text is garbled at the bottom of the slide because Eelco edited out the presenter standing there, but I’m going to zoom in on one particular pair of columns:
To be explicitly clear: this Microsoft slide is telling their partners that SQL Server is a legacy database. That’s what “1st party legacy” means – Microsoft is the first party.
I’m just going to leave that there, and give y’all a place to discuss it here in the comments.
Free, 3× a week
Get my new posts by email
Three posts a week, plus a Monday roundup of the best database news from around the web.
Microsoft really wants folks to move to Postgres.
If my choices are Fabric or Postgres that’s a pretty easy choice.
That was the part that really surprised me. If you’re going to call your own SQL Server product “legacy” in the year 2026, it’s pretty courageous to think that the natural next step would be either Azure SQL DB or Fabric, especially when Microsoft doesn’t dominate the cloud hosting space. What exactly do they expect customers in AWS to do?
Not remotely surprised. This is bonus (or not-getting-fired) motivated.
There are groups in MS being pushed to grow compute in Azure and get more people there. PaaS is where they’d like to lock you in. IaaS isn’t as much of a lock-in, so this is likely partially marketing, partially sales, and partially self-motivated MS staff.
Yes, Postgres for new projects. But a migration from TSQL is daunting. (We’re piloting this with assistance from Claude right now).
Microsoft Solution Partners are still stuck with SQL Server 2019 licenses. That sounds pretty legacy—so the move makes sense.
And when you’re making 140 billion in profit, you really couldn’t care less about any individual customer.
True enough, but thinking that everybody can be “heavy-handed” onto the cloud might be a miscalculation. There are a whole lot of individual customers who can’t for institutional policy reasons. When you add them up that could be a chunk of those 140 billion
This is much more prevalent than I think they realize. Take government for example, there are a large number of cases where it is required that data be held on-prem and not in the cloud. If they were to actually kill SQL server the choice wouldn’t be “go to Azure”, it would be “which non-MS on-prem SQL server is the best replacement”
Yep, did a course with someone and he explained that while learning the differences between the cloud platforms was interesting, his company (for security reasons) would always be on-prem with servers under their control.
This is just bizarre that they would put it out there. At a conference!
I never expected to be on the side of Amazon in a contest, but here we are…
They have been trying to get rid of WinForms since the 2000s. I am not going to start sweating just yet. lol
Since the $ is king, if you are leaving legacy systems behind, any cloud provider will do…As long is it’s the lowest bidder.
lets see how quick they roll those words back.
We are doing this data lake stuff already in Snowflake. SQL Server needs to hurry up and add support for iceberg tables
This feels like they want to push the “No one gets fired for buying Microsoft” folks to move into Azure, and figure they can risk the rest of the business.
Got involved with MS and SQL Server through Ashton-Tate (we had a development partnership with Ashton-Tate). At some of those meetings various presenters (especially MS) had the habit of talking about their envisioned “tomorrow space” as if it was historical fact. Most of those proved to be rubbish and BS. move forward 41 years and here we are “Going back, Jack to do it again.” These MS evangelists will end up on the same ash-pile as DR DOS
You can have my on-premise SQL Server core license when you pry it from my cold… before it moves to Azure. 1 DBA, 2 Windows guys – almost a 2 billion $$$ company. Why do I need Azure (someone else who doesn’t care about my company) to run my companies $$$ workload? Azure makes sense for Office365, Outlook.. not for critical low-latency robot/Automated OT manufacturing processes.
Jeff beat me to it, “from my cold…”
Linux is looking better and better!
This was the vision of Satya Nadella when he took over as CEO of Microsoft 14 years ago. All us SQL gurus should have seen this coming maybe?
I remember the ‘Lift and Shift’ days to Azure for everything on prem that MS was touting. Eventually they started using ‘hybrid environment’ type phrases know that a complete move was unlikely. If it ever does happen with SQL Server, it will be a while I think. We are still rocking a ton of SSRS reports too. 🙂
Same thing is happening to Power BI, being replaced by Fabric Apps.
https://community.fabric.microsoft.com/blog/-/Power-BI-s-next-chapter-The-evolution-of-business-intelligence/5369131
I’m going to put this in the same bucket as when SQL 7 was going to get rid of DBA because we now had Auto statistics
Microsoft pushing everyone to the cloud is understandable, but the reality is much more complicated than just ‘migrate everything’.
A huge roadblock that often gets ignored is data sovereignty.
For many European companies, using US-based cloud providers is a legal minefield.
Because of regulations like the CLOUD Act, US agencies can potentially access data even if it’s physically sitting in a European datacenter.
For organizations handling highly sensitive data, that lack of control is just a hard ‘no’.
We’re already seeing pushback. France, for example, created the ‘SecNumCloud’ standard for critical infrastructure like their national e-invoicing.
To get it, the cloud provider must be European-owned to completely wall off the data from foreign extraterritorial laws.
So while Microsoft’s marketing slides might call on-prem SQL Server ‘legacy’, the legal and privacy realities mean that on-premise databases aren’t going anywhere anytime soon for a lot of businesses.
I’m still firmly of the opinion that Fabric is not a serious platform which is ready to run production workloads. Not because it has any particular missing features in it’s data or processing engines, but for the simple fact that it has no observability and monitoring story that lives outside of the platform itself.
As a former (reformed) DBA, Database Team Manager, and Operational IT professional, the thought of running database workloads in a platform that I cannot effectively monitor and answer questions from users and my employer about why things are not working as planned means this is something that I can not get behind. I want to see a proper instrumentation framework that allows for ecosystem partners and community OSS solutions to make running it easier.
Once they fix that, then we can talk about the impracticality of moving third party workloads to Azure SQL Database or, if you’re feeling really brave, Fabric Database.
The thing is, I really love the concept of OneLake and what they are trying to achieve, but as normal the execution is flawed.
Why does it have to be cloud or on-prem? Doing both seems to be serving Microsoft (and their users) well so far. Other companies out there have reached that same conclusion and are perfectly happy offering products for both worlds. Take for example ESRI with their ArcGIS Enterprise (on prem) and ArcGIS Online (cloud) architectures.
Because if they can get you up in the cloud, they can sell you more services more easily, without you having to spin things up, and you can’t cheat on licensing costs, and you won’t be able to move to another hosting provider easily.
Depending on how you read my words, it can sound like I’m a paranoid guy in a tin foil hat, but honestly, if I was in their shoes, I can understand how some folks in finance & marketing would come to that conclusion, thinking that the new revenue they’d gain would offset any holdouts they’d lose. After all, if you’re using the boxed product today, it’s not like you’re gonna rewrite your entire app to move it quickly. You’ll just keep paying support for forever.
Perhaps the whole slide snafu is just a test to see who they can scare into jumping onto Fabric. If they’re measuring I think they’ll find out that there are a lot of players out there who don’t scare that easily.
I prefer the term “vintage” myself ?
Hahaha, nice!
Fair reaction. The wording on this partner slide was poor, and I understand why it was read that way. The two lines described separate modernization motions; “1st-party” [SQL Server] and “3rd-party legacy data estate” was not intended as a designation of SQL Server.
To be direct: Microsoft does not consider SQL Server a legacy product. It remains actively developed and supported, with millions of deployments powering businesses around the world. Modernization looks different for every customer.
Anna – the slide literally says “Modernize every SQL Server”.
There’s no room for wishy-washy interpretations of that, is there?
What wording do you think the slide should have said instead?
Maybe this is just a case of a terribly worded slide by somebody at MS who had no business wording anything. Time will tell, I guess
I’ve heard enough Microsoft stories to know that it got approved by somebody.
This is damage control, but the cat was already let out of the bag.
Glad I am learning other database products. Nadella’s relentless push for enshitification and push for ever decreasing reliability in everything MS cloud marches on.
It’s gotta just be them trying to get people to move to Fabric so that can upsell the AI features to grow their revenue there. Plus if folks are using on-prem SQL outside of Azure VMs, they make money on the storage and compute that they would not be getting from those customers today.
Microsoft is not discontinuing SQL Server on VMs or on-premises. However, they are clearly pushing customers toward SaaS/PaaS rather than IaaS or on-premises deployments.
The reality is that, for some workloads, you may not get anywhere close to the same level of performance or cost efficiency that you can achieve with a well-designed IaaS or on-premises SQL Server environment. I have some Azure SQL Database workloads running on the Business Critical tier, and the performance has not been very good. One of the biggest issues for me is that you lose a lot of control over the underlying infrastructure and database environment.
Microsoft also changed the name of MPP on premise to Azure SQL Data Warehouse to Synapse, and now Microsoft Fabric. From what I have seen, Fabric doesn’t provide the same level of performance for transactional workloads either. It also feels like you have less control over traditional database optimization techniques, such as indexing and physical database design.
I don’t know about you, but my plan is to continue using SQL Server, either on virtual machines or on-premises/IaaS, rather than moving everything to PaaS or SaaS. For new transactional databases, I’m also considering PostgreSQL.
For me, the decision ultimately comes down to having control over the environment, predictable performance, and being able to optimize the database based on the workload rather than relying entirely on what the managed service allows.