3 ms·
Yes, this is a very bad idea. Due to the way that query planners work, it's not even a good idea to vary the shape of the dataset between testing and productio
by calpaterson 11y ago
Yes, this is a very bad idea. Due to the way that query planners work, it's not even a good idea to vary the shape of the dataset between testing and production. If you have a different (bigger, higher stddev, etc) distribution of data in column x, the planner will often complete a query differently
- ams6110 11y agoAgree, unless your need for a DB is quite trivial, then perhaps something like SQLLite for the devs and whatever else for production. However this breaks down quickly once you grow beyond needing very simple INSERT and SELECT type operations.
- sitkack 11y agoAt which point one should just run SQLite everywhere, It is often excellent for datasets up to about a gig.