3 ms·
The what to optimize then also becomes an optimization problem. Devs struggle to plan for read/write workloads. Making more indexes can then make the same inbo
by cmgbhm 4y ago
The what to optimize then also becomes an optimization problem.
Devs struggle to plan for read/write workloads. Making more indexes can then make the same inbound volume from the previous day require n more writes per row.
You can still get surprised by query optimizations on relatively simple joins just changing query strategy when hitting tuning parameters.
Making analyzers more visual and giving lots of suggestions (that could then be ignored) would b a good intermediate step.
- marcosdumay 4y ago> You can still get surprised by query optimizations on relatively simple joins just changing query strategy when hitting tuning parameters. I imagine there is a large bias about caring or not about this depending on what DBMS you are used to. It just happens that on some DBMS you get "surprised" as "oh, shit, this problem again", and on others it usually comes as a surprise as "wait, I never thought about optimizer limits". The second set is much more prone to automating.