One of the big value I get from your videos, especially this kind of video is the vocabulary around the topic (aka the domain model), and the fact it fleshes out or enriches one intuition; which helps us find a solutions quicker, when you face a narrow issues.
Wow, there are so many fundamental and deep concepts in your video that I feel dumb not to have known most of them, despite working with Postgres for years. Thank you so much.
Thanks, I wanted to know more about Postgres and in the end I got much more knowledge about Architecture from it which is much more valuable. Your first video that I watch and I immediately subscribed
Thank you Hussein! Love your content, from all the tech RU-vidrs, i can follow you the best, hope you doing good in life! As soon as I found my place in tech, I will donate to you. Keep up the great work!
Might I suggest that you use some kind of pointer ( mouse cursor ) to point to the items in the presentation as you speak. It would really further improve your presentation. Because as viewers, it is sometimes difficult to predict where exactly focus / intention is. As always, your videos are extremely helpful, thank you so much.
Maybe, but it's fine to me and helps me be more engaged, Also additionally Naseer can flow his thoughts as usual without having to bother about pointers.
Great summary. An improvement I can suggest would be to expand the acronym WAL (Write-Ahead Logging). I had to look that up elsewhere. I guess you could do a whole webcast on it & how it works.
On your point about there being lot of TLB cache misses due to Postgres having processes instead of Threads, this is not really the case - There is one TLB per core for most modern processors (Intel - core/AMD - ryzen) and all processes on that core can share that TLB, so there will be no TLB miss due to postgres having multiple processes. Different core processes will face TLB miss but that is true for Threads as well. There is one TLB per process in different processors such as ARM processors used in mobiles where the use case support this as every app's TLB entries are flushed after that app is killed.
Great info... I always guess what was the 'real' difference and strengths between PostGRE and MySQL/MariaDB... now please do the same for MariaDB... and for MS-SQL... and in the end a comparative of use cases and why....
@hnasr Could you do a comparaison with GCP AlloyDB ? To understand the difference and the architectural changes that GCP decided to apply to PG, especially the decoupling of compute from storage. I just don't know if there is enough information out there.
It's crazy how fork is still the standard mechanism to spawn new processes. It's unnecessarily expensive even with the optimizations and has thread safety issues, plus it's horrible from API perspective. We should use alternatives like posix_spawn for better efficiency, less bugs and easier to use code.
The number of adverts on this video is crazy. Constant interruptions make it harder for me to absorb the information in the video. Only got half way through then gave up.