|
|
|
|
|
|
by stingraycharles
1294 days ago
|
|
|
Out of curiosity: > It scans those columns and knocks out any blocks (technically granules containing by default 8192 rows) that do not values that match the filter conditions How is that not just a sequence scans? Of course it pre-emptively filters away entire blocks that do not contain the data, but indexes typically work differently: they’re calculated upon write, so that they can be queried really fast. Is there a detail missing here, e.g. like bloom filters being used or something else that makes it different from a regular sequence scan? |
|
|