Re: Trying to decide whether to support DB2 or Oracle

> I don't know what the product plans are in terms of the requirement for
 the "C"
> compiler. But there is another side of this that I respectfullly submit
 you are
> completely ignoring: performance. How would an interpretive SQL stored
 proc
> perform vs. an IBM SQLPL "compiled" stored proc, for a reasonably complex
 stored
> proc? I really don't know ... I'm just asking. But I'll bet that the
 compiled
> stored proc would tend to perform better.

As I've stated earlier, for my tests, MSSQL 2k performed the best, inserting about 280,000 rows through stored procedures. Next was Oracle 8i followed by DB2. As for "interpretive SQL stored proc", I would hope that it *wouldn't* be interpretive at all. The RDBMS' know the syntax - hell, it's their syntax. They should compile it internally, build access plans (where possible), etc. It should be as close to native code as is possible.

> In this industry, there are two sides to every story ... there is always a
 cost.
> In this case, ease of use vs. performance maybe?

This doesn't have anything to do with "ease of use" in my opinion. Ease of deployability yes, ease of use, no. This is not just academic whining - this is a small ISV trying to produce high-performing code for multiple RDBMS'. It seems that I should have totally blown off the performance and isolation problems, and gone with dynamic SQL. But since SQL is even different (slightly sometimes, greatly other times) across all the RDBMS', I would have been in a similar boat. It just shouldn't be this way.

> Nothing's perfect. If you try
> hard enough, you will be able to find fault with any database, even
Oracle.

I have found many, many, many problems with Oracle. please see my earlier posting on this topic. The *only* RDBMS I haven't had any appreciable trouble with is MSSQL 2k (one of our required platforms). It has been a breeze (unless you count the lack of before triggers ;-)

Everything has faults. I agree. But some are sucking chest wounds, some are abrasions and some are bruises. I contend that the C-compiler requirement for SQL Stored Procedures is about as useful as whisky in a Morman church.

Marc Received on Tue May 08 2001 - 02:12:29 CEST