Supplier proposes upgrade to 2016, we're running 2008r2, should we do it?

Solution 1:

It's hard to answer your question. You won't find public insights that tells you you should upgrade to the "not latest" Windows Server. I guess because Microsoft's influence in the IT press is very high and they always push their latest system, and so it's hard to find someone telling you the other way around.

As insight, i can share my 20 years experience on Microsoft OSs. The latest is never been a good option. The only exception was upgrading Vista to 7, for who went to Vista (I had all my clients still to XP ofc). Someone says Microsoft likes to use their users as beta testers, but many people may argue on this. Someone say you should wait between 1 to 2 years before installing latest OSs.

For instance, with latest updates from last patch tuesday we experienced on a few 2016 Std this error: "The stub received bad data". This just happens the first time you open up an application in the desktop environment. If you click the program icon again, it just normally opens. Now, such a weird thing never happened on a "at least two years old" microsoft os. This is just an example.

Waiting a few years is always been the best option with MS Servers as for my experience. The only one thing that should lead you to the latest software is if there are unbackported features you desperately need.

Also, you talk about a small env, and as thus i imagine you don't have a test env. So implementing latest softwares will be even more risky.

Also, consider what ppl wrote in the comments. Some of them won't aggree with me. Licensing: purchase 2016 Std. Run 2012 R2. When it comes to the next server upgrade, that will probably be before 2012R2 EOL (2023), you will find a good chance to upgrade to 2016. Again, you'll find yourself asking if you should upgrade to 2020R5, or to 2016. You'd better choose 2016, or, at least, that's what i would do.

Solution 2:

Going 2008R2 -> 2012R2 in mid-2017 is pretty much pointless, I'd wait for 2016 successor (there are rumors MSFT will release it soon to tap numerous holes in their WS2016 GA) and put IT into production.

Solution 3:

I pretty much second Marco's experience. In February for instance, I tried to setup a Windows Server Failover Cluster with Storage Spaces Direct for an SQL Failover Cluster instance. The way to achieve this was very rocky and last week, the whole thing deadlocked and crashed in on me. Finding proper, detailed and relevant documentation on Windows Server 2016 features is slowly getting better now, but still a major pain. And I ran into a lot of not so funny stuff feeling a bit like a beta tester. In this case, I had no choice due to the required feature set, but my experience holds true that the latest version is not necessarily the most mature one.

So if you just need a plain server for "normal" duty, I don't see a reason not to go to 2016. We deployed several, it will give you the platform and increase the interval before the next upgrade. But if you have something more taxing in mind, I'd check whether 2012 R2 delivers all features you need.