Overlooking Problems with Oracle's Exadata

The premise that Oracle Exadata V2 is going to siphon off all of the existing implementations on older SMP platforms to the new Exadata platform fails for two reasons...

Neil Raden, Contributor

December 15, 2009

3 Min Read
InformationWeek logo in a gray background | InformationWeek

I can't quite figure out where IDC's review of Exadata V2 ends and Stephen Swoyer's opinion begins, but in the latter's article "Why Oracle's Exadata May Attract CXO's," some questionable suggestions are offered. At one point, Swoyer writes, "many of the largest data warehousing or OLTP systems in the world continue to run on Oracle," and adds further that, "many of the biggest combined DW and OLTP configurations also run on Oracle." While this may be true, the key word here is "run." I know of no "combined" OLTP and data warehouse instances that actually perform. There are a myriad of reasons for this, such as:- Workload management is dramatically different. OLTP is small transactions, many of them write; data warehouse is bulk load and very large (comparatively) queries. Activity is much less predictable and rather clumpy compared to OLTP - The physical layout of the tables for performance reasons is different - There is usually a need for aggregation in data warehouses - There are different optimization strategies for OLTP and BI What is a combined OLTP and data warehouse, anyway? An OLTP system typically serves an operational purpose, but a data warehouse combines data from many systems. Does a combined system merely provide analytical access to the data of one OLTP system? That doesn't sound like a very good idea. I hope I live long enough to see the dream of combined databases actually realized -- data entered and stored at its logical location and used and reused for any valid purpose without replication. It would be fantastic, but it's sort of unrealistic. Data warehouses of yore served as big bags of data, with lots of downstream data marts, cubes and other structures that served to satisfy the need the data warehouse was meant to serve. In other words, many of these "large" Oracle data warehouses are simply holding pens. But no one sees data warehousing this way anymore. Whether you have 100 GB or 100 TB, that database has to provide deep analytics on its own. Large Oracle databases aren't doing that. Competitors are. Therefore the premise that Exadata V2 is going to siphon off all of the existing implementations on older SMP platforms to the new Exadata platform fails for two reasons. First, if your existing database vendor can't cut it, you are not likely to cast your fate with their newest promise. Second, Oracle has absolutely zero credibility in massively parallel platforms. Does anyone remember Oracle PQO for the SP2? That was the last time Oracle was going to win over the data warehousing market with a highly scalable platform. It didn't work. And RAC is a terrible choice for data warehousing. Exadata will have to prove its ability to handle BIG DATA warehouses, and the Oracle database will have to have the functionality for analytics that its competitors already exhibit.The premise that Oracle Exadata V2 is going to siphon off all of the existing implementations on older SMP platforms to the new Exadata platform fails for two reasons...

About the Author

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights