Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

imo, json dt should be an intermidiary step in db struc in rdb, never the final. Once you know & have stable columes, unravel the json into proper cols with indexing, it should improve the situation

if youre having issues with 5gb, you will face exponential problems when it grows due to lack of indexing



Cheers for the response (and affirmation). After some latency profiling I am convinced proper cols with indexing will vastly improve our situation since the queries themselves are very simple.


Depending on how much of the data in your json payload is required, extract data into their own table/cols. And store the full payload in a file system/cloud storage.


Also there's a way to profile which queries take longest via DB itself and then just run EXPLAIN ANALYZE to figure what's wrong


You can take incremental approach and p.o.c with the data that you have so you can justify your move too!




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: