Oracle ups the storage ante
Content Copyright © 2013 Bloor. All Rights Reserved.
Also posted on: The IM Blog
I don’t normally write about storage. In fact, I never write about storage, at least in hardware terms. Indeed, about the only time I have ever written about hardware – of any sort – has been with respect to the FPGAs in Netezza (IBM). However, Oracle’s announcement of the ZS3 “application engineered storage” is leading me to break the habits of a lifetime.
According to Oracle the ZS3 outperforms all of its competitors – mostly EMC and NetApp but others such as HP and IBM are also mentioned – on standard storage benchmark tests. I wouldn’t know about that. The reason I don’t write about storage is because I don’t know anything about it. So, why this article?
Because ZS3 is integrated with Oracle Database 12c and this has important implications for database performance and management and, unfortunately, for analysts such as myself who specialise in database management systems.
On the performance and management side, the point about the “application engineered storage” is that the ZS3 storage knows about the database and 12c knows about the storage. In particular, they exchange metadata so that there is now a feature called Oracle Intelligent Storage Processing (OISP) which, briefly, provides dynamic autonomic capabilities for database to storage tuning, significantly cutting the amount of administration that task otherwise needs. Secondly, a feature called Auto Data Optimisation (ADO) supports the concept of tiered compression, whereby the storage and database, between them, can determine (based on currency and usage) whether data should be left uncompressed for performance reasons, or if it should be normally compressed or if it should be deeply compressed (for archival). And, finally, ZS3 supports Oracle’s hybrid columnar compression.
Now for the bad news. It was already the case that analysts needed to know at least something about hardware, thanks to the likes of Teradata, Netezza and, more recently, Oracle Exadata. In the short term it means that you can’t just compare DB2 and Oracle in terms of their software – which I have done in the past – even with a little bit of cognisance of the extra facilities that Exadata adds. Instead, a broader approach is going to be needed in future. Moreover, if this announcement of ZS3 should start a trend – if, for example, IBM should start offering DB2-specific features in its storage, or HP do something similar for Vertica – then that’s going to make the analyst’s job much harder. It’s difficult enough getting your head round all the software that’s out there: adding hardware expertise is a whole additional ballgame.
In the meantime, well done to Oracle for raising the ante. Anything that can cut administration time is going to be a good thing and if the performance improvements the company is claiming are anything like the way it improved in the America’s Cup then this has to be bad news for competitors, especially those in the storage space.