rcooke (6/7/2010)

---

i like the hash approach - to save time you could just use the konesans checksum algorithm add in to calculate the crc value. Combined with the conditional split it works a treat.

da-zero (6/8/2010)

---

Nice article. Although I would have used a temp/staging table to store my updates, instead of using the OLE DB Command. That way you can use a set based update instead of an update per row.

Adam Aspin (6/8/2010)

---

Really interesting and well-written article, thank you.
As a matter of interest, is it any faster (or in any other way better or worse) to create the hash value as part of the T-SQL Select statement, using code based on something like this:

SELECT HashBytes('MD5', FirstName + LastName) FROM [AdventureWorks].[Person].[Contact]

Of course you will have to use all the fields that you will be hashing on, and cast all values to (n)varchar, but I wondered if you had tried this, and if so, which appraoch gave better performance?