Saturday, 5 April 2008

Bucket number and SIFT table

The first column of the SIFT table is always named bucket. After the bucket table are the set of key columns that named in the following format f<field>. The <field> represents the Field No. in the table. After the f-columns are the s-columns, which represent the SumIndexField in the associated table. The precalculated totals are stored the s-columns.

(Click to enlarge)

Every bucket number in a SIFT table has a bucket number that corresponds to its SIFT level. The following example shows the relationship between SIFT levels and bucket find in the SIFT table.

As shown in the example, bucket 0 only contains 1 record, which is the grand total for the SumIndexField. Please take note that only 1 record will exist in bucket 0 for each SIFT table. Bucket 0 is not activated by default because it will cause performance issue during update. I have activated the SIFT Level 0 in this example for demo purposes.

(Click to enlarge)

No comments: