Open joshua-spacetime opened 10 months ago
100 is not so far from a suitable N (for context)
We would like to see the number of bytes per table so that we can track which tables have very large rows.
We would like to see the number of bytes per table so that we can track which tables have very large rows.
It would be very easy to compute a conservative overestimate of this by simply tracking the number of pages per table in the committed state. A precise count is more challenging.
I split this ticket into https://github.com/orgs/clockworklabs/projects/22/views/38?pane=issue&itemId=47389802 as well
We should track the persistent memory associated with tables and indexes for energy purposes. This means each table and index needs access to an
EnergyMonitor
for recording energy use.If
1 eV
is the amount of energy used to store 1 byte on disk for 1 second, thenN eV
is the amount of energy required to store 1 byte in memory for 1 second for someN
. We should pick a suitableN
and record, for every insert/delete, the amount of energy used per table, since the previous insert/delete.As part of this ticket we should also write these energy recordings to prometheus.