What size does my database need?
Estimate the size of a PostgreSQL, a MySQL or a Redis® on Tenkai from your data, your rows and your connections, and see the memory and the settings it gets.
Choose your service, enter how much data you keep, how many rows and how many connections your applications open, and the calculator suggests a size. It is an estimate: it shows what it assumes, so you can judge it.
Estimate
Medium PostgreSQL
512 MB reserved · 1 GB limit
For PostgreSQL, the planner cache estimate is 384 MB, compared with your estimated need of 318 MB. The calculator estimates room for 80 connections. You entered 20.
- Shared buffers
- 128 MB
- Planner cache estimate
- 384 MB
1 server, 512 MB reserved on it
One node
- PostgreSQL512 MB
What the estimate assumes
Your queries, your indexes and your load change the real need. For a single-instance service, start with this estimate and resize when needed.
- Data
- The share your queries read often, plus 32 bytes of index for each row
- Cache assumption
- The calculator compares your estimated working data with the planner cache estimate for this size.
- Each connection
- About 8 MB, from the rest of the limit
- Disk
- Not part of the size: the data lives on the disk of its node
How it estimates
A database answers fast when the data its queries read is in memory. The selected size determines these memory settings:
- PostgreSQL plans its queries as if three quarters of its reserved memory held data: a quarter in its shared buffers, the rest in the system cache.
- MySQL keeps data in its InnoDB buffer pool: half of its reserved memory, and no more than a quarter of its limit.
- Redis® and Valkey keep all of their data in memory, in three quarters of their reserved memory, and refuse writes when it is full.
The calculator compares that memory with the part of your data that your queries read often, plus an index entry for each row. Then it counts the connections. Each one takes memory from what is left of the limit, and a server accepts only so many: 100 for PostgreSQL, 100 for a MySQL whose limit is 512 MB or less, 151 above.
It suggests the smallest size where both fit. If Large is too small, the calculator suggests a custom size. Its memory limit is twice its reserved memory, as with the standard sizes.
What it does not know
- How your queries read the data. A full-table scan can read much more data than the share you selected, so this estimate may understate the memory needed for your target performance.
- The indexes you add beyond one per row.
- The CPU your queries need. The size sets it too.
- Disk. The size sets CPU and memory only: the data lives on its node's disk.
Start from the estimate and choose Change size when the database needs more.