/*Return all dirty aggregations in last X days*/
DECLARE @Days AS int
SET @Days = 1 --number of days to go back
SELECT DS.DatasetDefaultName,
CASE SDAH.AggregationTypeId
WHEN 0 THEN 'Raw' WHEN 20 THEN 'Hourly' WHEN 30 THEN 'Daily' ELSE 'Unknown'
END AS Type,
SDAH.AggregationDateTime,
DATEADD(HH, (datediff(HH, getutcdate(), getdate())), SDAH.AggregationDateTime) AS [LocalTime],
SDAH.DirtyInd,
SDAH.AggregationCount
FROM Dataset AS DS INNER JOIN
StandardDatasetAggregationHistory AS SDAH ON SDAH.DatasetId = DS.DatasetId
WHERE SDAH.DirtyInd <> 0 AND SDAH.AggregationDateTime > DATEADD(DD, -@Days, getdate())
ORDER BY SDAH.AggregationDateTime DESC, DS.DatasetDefaultName ASC, [Type] ASC
What does it mean and what should we do to avoid these dirty aggregations?
Having dirty aggregations usually indicates a performance problem, but sometimes can be poorly written procedures. If you see one or two dirty aggregations in a row, and then a bunch of successful aggregations, do not be concerned. If you see 10 or more dirty aggregations in a row, and there is no recovery, then you've got issues. Main point is, the DW should always recover from dirty aggregations.