Sybase vs Oracle vs Teradata for >50GB

I'm looking for EXPERIENCES in very large databases (>50 GB) running on symmetric multi-processing systems (SMP) or massively parallel processing systems (MPP). I'm also interested in OPINIONS on the different architecture approaches.

I have all the marketing info and have been following the comp.databases.{sybase | oracle} sybase vs. oracle discussions.

Now I'm looking for results from anyone who has actually done performance/scalability testing or architecture analysis on the new parallel releases. Are there any magazine or journal articles comparing features/listing issues for large smp databases?

Does each database scale well--is there near linear speedup as you add processors? Is there a limit to the size the database can grow to? (Experiences are helpful for the size claim. Vendors say infinity.)

The initial applications will be decision support (large read-only table scans). But eventually they'll be batch and OLTP.

How well are data partitioning, locking, logging, and versioning handled by these databases on smp machines? What other database issues are key for performance on SMP machines?

Thanks in advance for any info....

/andrew Received on Fri Apr 16 1993 - 22:37:05 CEST