I selected the correct answer because I thought 'myothertab' was from another tab, as in a different query tab. The question seemed to be about whether a temp table...

No comment!!

PS - I recall it...

The foul language generator is definitely the most humorous. The best bad design I've ever seen was designed by a sales manager and had rows for dollar sales by...

Cyclone (5/17/2010)

---

Do you see any problems with this approach?

Also, I checked the length (using Len()) of the Hash Value and the max length I have so far is 20. I...

I see these functions in Oracle queries all the time and only partly understood what was going on - thanks for explaining it the SQL Server way! Now I...

Look, the question of what exactly to use for an algorithm is a far off tangent. The main idea of my article is to use the narrow xref tables...

I surrender; your business users are infintely more badass than mine if they're analyzing these kinds of statistics in your data warehouse.

Mike C (4/16/2010)

---

I take exception to Kimball's point about the efficiency of comparing one (binary) string value, etc., for this reason: your natural key for a dimension might be an...

tarquin (4/16/2010)

---

Compute the hashes and store them in the dimension table

Fact data that has been written to a datawarehouse typicall does not change i.e. is not updated, it is flagged...

You may be using an older version of SQL Server. Sorry,...

tarquin (4/15/2010)

---

The idea is to use the hash value as its leading column so the cardinality of the index is very high and matching is very quick. This is all...

In my book any likelyhood = a certainty (at some point down the line). Do you have anything in place to catch collisions should they occur or are you just...

rmd0 (4/14/2010)

---

One issue I have with this solution is that MD5 *will* have collisions with moderately "wide" tables, especially when bit columns are involved. In fact, I would see as...

tarquin (4/14/2010)

---

I need to use multiple columns in the hashindex and so use a checksum over hashbytes such as:

checksum(hashbytes('MD5', column1), hashbytes('MD5',column2)) - this works pretty well but is still not...

Mike C (4/13/2010)

---

Nice article! I implemented something very similar for loading data warehouses and datamarts a while back. One thing to keep in mind is that you can...