Weekly reading #27
September is over. This was not a good month for us. We had to cancel the conferences planned for September and were not able to write as much as we wanted to. We have a lot of great topics to share next weeks! Today – let’s start with some fresh information from the data platform world
Hot Patching SQL Server Engine in Azure SQL Database
How Azure SQL Database does the patching? Interesting.. because it has to be done in a very tiny period of time to make sure your business is not interrupted!
Non-Root SQL Server 2019 Containers
It is all about privileges. How to run SQL Server on Docker as a non-root user. This is important as if the root user breaks out of the container, they would have root permissions on the host.
[Video] Data Lake Design Patterns
I have been working with data lakes quite a lot recently and I can recommend the video if you need to learn the basic design patterns and architectural principles to make sure you are using the data lake and underlying technologies effectively.
Nested Loops Joins and Performance Spools
Paul White has written an excellent article – as usual. This time he has desrcibed the performance spools operators. He has also described some (undocumented) trace flags so you can test the behaviour of the queries
Azure SQL DB Design Note: Choosing the Right Database Reservation Size
Once upon a time Azure SQL Database was very simple to choose (1GB or 5GB) and that was it! Now things are more complicated because the database must be aligned to your business. Conor Cunningham explains what options you have and which to choose and when.
Troubleshooting NUMA Node Imbalance Problems
Klaus Aschenbrenner found a really huge thing. He has found how to reproduce HUMA Node imbalace problem and how to fix it! Definitely a ***MUST READ*** for all DBA’s!!!
By the way – do not forget to visit the Interviews page – there are 6 interviews as of now but much more to come!
Interview no. 1 – Uwe Ricken
Interview no. 2 – Andre Melancia
Interview no. 3 – Rob Farley
Interview no. 4 – Grant Fritchey
Interview no. 5 – Aaron Bertrand
Interview no. 6 – Greg Low
Cheers,
Damian
coding family 
Well, regarding that NUMA imbalance from Klaus Aschenbrenner – it’s worth mentioning, that he fixed only the testing tool, not the root cause of the issue :|.
I had this problem once or twice and definitely it is how SQL Server connection stack is implemented.
It drives you nuts if your applications are making so many connections and some of them are dropped quickly, this definitely leads to this nasty NUMA overload.
Round robin is the cheapest solution and of course there are many reasons for keeping it like that. I think compromise is the best of them ;]
So, discovering a tool with a bug, that can reproduce this problem is a cool situation, but unfortunately there is no general solution or fix to SQL server. You need to fix your architecture and applications first :]
LikeLike