bug summary can definitely answer most of the stats we need rapidly, but the fixed-upstream stat may be harder to get without a table scan. An open question is whether a table scan + bugsummary would be cheaper at all than the current tablescan.
The current scan does a lookup per bugtask for related bugtasks. So one way to rephrase that would be to pull back more raw data and get a group-and-count of the.
The privacy clause is 1 second hot; the related bugtask check another 1 second hot.
The bug join is ~ 2.5 seconds (0.014 * 190K) - a table scan of just bugtask is 1 second hot.
If we denormalise duplicate and latest_patch_uploaded onto bugtask we may be able to do a much more efficient query (even though the table will be getting wider).
bug summary can definitely answer most of the stats we need rapidly, but the fixed-upstream stat may be harder to get without a table scan. An open question is whether a table scan + bugsummary would be cheaper at all than the current tablescan.
The current scan does a lookup per bugtask for related bugtasks. So one way to rephrase that would be to pull back more raw data and get a group-and-count of the.
The privacy clause is 1 second hot; the related bugtask check another 1 second hot.
The bug join is ~ 2.5 seconds (0.014 * 190K) - a table scan of just bugtask is 1 second hot.
If we denormalise duplicate and latest_ patch_uploaded onto bugtask we may be able to do a much more efficient query (even though the table will be getting wider).