DB2 compared with Oracle doc. Comments?

[Guest_imported]

Has anyone read the following document in which Oracle claims to have superior performance and scalability over DB2:

http://technet.oracle.com/deploy/performance/pdf/CWP_9IVSDB2_PERF.PDF

?

Although this is written by Oracle guys and I've seen completely opposite comments made by third party guys, I must admit that they make a convincing point with the section on "multi version read consistency". Apparently, shared locks aren't taken in Oracle when reading data and readers don't block writers, which is exactly what happens in DB2 when isolation levels above cursor stability are used.

Can anyone come up with a counter argument to this? If Oracle's "multi version read consistency" performs and scales so much better than DB2's concurrency architecture, how come Oracle doesn't scale or perform better as a whole, according to most third party DB comparison documents?

 

[blom0344]

Hello Db2meister,

You raised a very interesting subject here.
With our firm we are considering migrating from ORACLE to DB2 with the warehousesystem. I'm basically working with DB2, but I am wondering whether this will not cause a lot of heartache considering the very complex concurrency model of DB2 (4 levels of isolation to be considered) The warehouse system typically sees an enormous amount of read write activity in just a few tables. With ORACLE the system sometimes slows down, but the is a mere inconvenience. What will happen with DB2, when complete tables are going to be locked? Does ORACLE work better with small scale applications with large numbers of read/write activity?
I think ORACLE rightfully claim that their concurrency model is more transparent and SHOULD provide better performance for small-scale deployments. My experience is that DB2 offers better scalibilty and stability.
I hope this thread attracts a lot of feedback from others... T. Blom
Information analist
Shimano Europe
tbl@shimano-eu.com