I am trying to estimate what is a 'reasonable' number of rows the DB can handle for heroku Postgres plans:
Assumptions:
- 1 row: 100 bytes
- Reasonable indexes are created on different columns
- Most of the queries are done on a single table. So not many joins and at a given point only one table's data is in memory.
Here is a rough calculation: 1GB Ram = 10M rows each of 100 Bytes. But what's the space required for
- Caching
- Indexes etc.
So what is a usable 'rule of thumb' rows I can fit in 1GB RAM, 15GB RAM for simple query optimizations.
It's fine to assume std Heroku architecture and any optimizations they provide.
Please share any experiments you have done or heuristics you have from your usage on this tier.
PS:
1] I understand that this is hard to predict accurately but I need some ballpark estimate to tell the client what plan we could use and what would be the cost.
