2 ms·
When I see these issues in PG, a lot of it seems like incidental complexity. Also I don't believe there's an "impedance mismatch". It's usually a problem of in
by 3cats-in-a-coat 3y ago
When I see these issues in PG, a lot of it seems like incidental complexity.
Also I don't believe there's an "impedance mismatch". It's usually a problem of incorrect abstraction, or incorrect framing of one. For example SSD rarely modify content in-place, it streams data over new sectors to distribute write wear and tear, as each cell has finite write cycles in its lifetime.
Also a system that combines in-place mutation, with immutable sharing, cheap snapshots (deep copy) and easy garbage collection is copy-on-write. So it'd seem this makes more sense as a baseline foundation to step on when implementing MVCC in RDBMS. But there we go.