Update on partitioned table

Question

I have a transaction table with columns like id, date, product, value, etc. The table has been partitioned based on date column and have around 5m records per partition.

I have to develop a ETL(SSIS) build to populate an additional derived column and logical query would be...

        Update    transaction 
        Set xxx= value 
        where id =100

What changes should i do in the above query to use benefits of partition elimination, partition parallelism etc?

Answer 1

0

If i keep simple update then it takes hours to complete the update hence looking for a method to force optimiser to use partition.

answered Feb 23 '10 at 11:14 AM

kardile
301 ● 14 ● 14 ● 14

(comments are locked)

Answer 2

0

What is the derived transformation you are doing? Does it depend on other data from the database or just other data in the row?

As far as I know, the best idea would be to just update it naturally, and let the query optimiser use the partitions as best it can. But, if the data only depends in other data in the row, then you may be best off with a computed column instead.

answered Feb 23 '10 at 09:18 AM

Matt Whitfield ♦♦
29.2k ● 56 ● 63 ● 87

The derived value is based on "product" that is a part of row and other data(from table). This derviated value req. a week delay and should be calcuated a week after the original record insertion in the transaction table.

Feb 23 '10 at 11:12 AM kardile

(comments are locked)

Update on partitioned table

Follow this question

Related Questions