SQL ConstantCare® Population Report: Summer 2023

The short story for this quarter: SQL Server 2022 adoption rates have stalled, even backtracked, and it doesn’t appear to be due to the cloud, either.

For the long story: ever wonder how fast people are adopting new versions of SQL Server, or what’s “normal” out there for SQL Server adoption rates? Let’s find out in the summer 2023 version of our SQL ConstantCare® population report.

Out of 3,124 monitored servers (up from 3,002), here’s the version adoption rate:

The big 3 versions are all within 1% of the last quarter’s numbers:

  • SQL Server 2019: 38% (up 5% since last quarter)
  • SQL Server 2017: 19% (steady)
  • SQL Server 2016: 27% (down 1%)

On the other extreme:

  • SQL Server 2022: 1% – actually down, from 52 servers last time to 46 this time
  • Azure SQL DB: 2%
  • Azure SQL DB Managed Instances: <1%

Just 13% of the population are running unsupported major versions (2014 & prior), and that’s steady from the last report.

Here’s how adoption is trending over time, with most recent data at the right:

It looks like companies are standardizing on SQL Server 2019 for new installs rather than SQL Server 2022. In the past, we’ve seen development environments going live on new versions first, ahead of production pushes, but we’re not even seeing that anymore.

I’ve written about how SQL Server 2022 still isn’t ready yet, and how even the updates are breaking. This feels like the moment where we call it: companies are skipping SQL Server 2022. It’s a shame, and I know Microsoft employees have to be frustrated about this, and I know I’m not going to make any friends at Microsoft by saying the quiet part out loud, but here we are.

Previous Post
Free Azure Networking Training This Week
Next Post
[Video] Office Hours: While Stack Overflow is Down

15 Comments. Leave new

  • That up-tick in Azure SQL DB in 2022 Q3 is interesting.
    Can you tell us if it’s an indication of adoption of Azure DB (followed by move back to on-prem)?

    • Hmm, how would I be able to tell that?

      • I’m guessing the granularity of the data is not sufficient to determine that. That is, customer A had 5 instances and one moved to Azure DB for a month or two making 1 in Azure DB and 4 on-prem. Later the 1 in Azure DB moved back to on-prem making 5 on-prem again. Interesting bit of data to know but that granularity doesn’t seem particularly relevant in the big picture of monitoring the servers with your toolset.

    • Sad to here. Looking forward to 2025 ?

  • Could all this be an artifact of which customers had SQL ConstantCare installed at a point in time, or which instances they chose to install it on? No way to correct for the latter possibility, but what would the graphs look like if you considered only customers that used SQL ConstantCare for the duration of the study’s time period? (In other words, exclude newcomers and folks not currently using it.)

  • I don’t think the companies are skipping SQL 2022, they may not be that eager to use it yet. I have plans to move our ~ 250 servers to SQL 2022 in 2023 with the next version of the main application because the current version does not support it yet; by that time, SQL 2022 will probably be ready for production.

    In the past we skipped 2014 and 2017 because they were too close vs our regular upgrade cycle, but we are in 2023 and SQL 2019 will be too old for new installations soon, for us. I am not betting on a SQL 2025 to be ready for prime time in 2026.

  • I think it would be good for MS to have 2022 just crash and burn become an embarrassment. It may trigger them to be introspective about what they are doing. They are succeeding in making open source alternatives compelling with narrowing functioning feature set and narrowing support and documentation quality, while also continually increasing prices of the software. The days of extensive QA sadly are over I think, but if they are going to have us do their QA and documentation for them, or train their support techs, at least give us a break on the licensing for the labor costs they are saving not doing it themselves.

  • “I know I’m not going to make any friends at Microsoft by saying the quiet part out loud, but here we are.”
    I that is what gains you respect in the SQL community, that you are not simply a parrot for promoting everything that Microsoft puts out.

    Run out and get SQL 2022 because it’s GREAT!
    Oh, and while you’re at it, grab a copy of MS Works!

  • So, almost half of the prod SQL server (40%) is 2016 or less… heavy!

  • Geoff Langdon
    July 19, 2023 7:08 am

    Vendors are not supporting 2022 SQL installations for us.

  • John Trollope
    July 19, 2023 9:09 am

    I’m not surprised by the low update of SQL Managed Instance: the impact of losing control of the OS has more impact than you might expect. Also even basic things like the backup and restore syntax have lots of things that will catch you out!

    In terms of my database code, the biggest impact is the complete drop of support for creating your own system procedures.

    I do run SQL 2022 as well so I can restore from Azure to on prem but that is really the only compelling use case.

  • Glenn Berry
    July 24, 2023 3:17 am

    I don’t think SQL Server 2022 is going to be a “skipped version” (like SQL Server 2017 was).

    The fact that Microsoft did not release SQL Server 2022 CU1 for three months after GA has definitely slowed evaluation and early adoption of SQL Server 2022. Not having the RTM version of SSMS 19.0 available when SQL Server 2022 went GA was another delaying factor.

    SQL Server 2019 falls out of Mainstream support on Feb 28, 2025. As we get closer to that date, it will be a harder sell to deploy SQL Server 2019 vs. SQL Server 2022. I also doubt that SQL Server “vNext” will GA before SQL Server 2019 falls out of Mainstream support.

    One good sign for SQL Server 2022 is that Microsoft has been hitting their stated one-month release cadence for CU2 through CU6, rather than regularly slipping by weeks for every release.

    • I hear you, but even *before* 2019 came out, folks had put in development environments with it. I’m just not seeing that now, even though 2022 has been out for several months.

      Interesting thought about 2022’s replacement not being out in 2025. Do you think Microsoft’s going back towards the 2000/2005 release cadence?

      • Glenn Berry
        July 24, 2023 6:40 pm

        I don’t think SQL Server “vNext” will take five years. I think Microsoft will be under some pressure to get vNext done before February 28, 2025. They usually like to do major SQL Server releases in November, so I can see two scenarios:

        A. They cut scope to get it done by November 2024
        B. They do more and release it in November 2025

        In either scenario, I think SQL Server 2022 adoption will go up over time, as less people choose to use SQL Server 2019 for new deployments. It will be interesting to see what your data shows over the next year.

  • Ivan Argentinski
    July 24, 2023 5:33 pm

    SQL 2022 is *still* not included in our Partner Program!

    Don’t know why. In the past, new versions were there close to release.

    So, without it in the partner channel we are with our hands tied.


Leave a Reply

Your email address will not be published. Required fields are marked *

Fill out this field
Fill out this field
Please enter a valid email address.