Oracle bests IBM in
grid"> You might ask whats so darn compelling about Oracles version of grid, considering that IBM has long been in that space. The answer boils down to two words: marketing and manageability. Oracles marketing pizzazz speaks for itself, as anybody who breathed in the dry-ice-fog premiere of 10g at last falls OracleWorld can attest. As far as manageability goes, Oracle has it all over IBM. As Bear Stearns analyst John DiFucci noted in a recent conversation, IBMs database group is a slave to two masters, answering both to customers and to IBMs mainframe business. IBM has delivered a type of clustering technology for some time, but it just doesnt compare to Oracles solution, which, according to early users, shows that Oracle has finally gotten a handle on the oft-bemoaned excess of knob-turning needed with its products.OPSnow, there was a hugely disappointing technology. Users complained, for example, about a locking procedure that relied on actual storage. Every time a user had two processors, if somebody placed an order or changed inventory in the system, the other processor couldnt access the data to check the inventory level. OPS required the first processor to write the data to the disk: a time-consuming process that made the locking procedure laborious. With these types of faults, OPS well-earned its poor market reception. Oracle9i RAC, while an improvement, suffered from its predecessors bad reputation. Burned users were hesitant to adopt grid technologies from Oracle, with due cause. As a result, Oracle9is adoption was sluggish. Some 2.5 years after its release in 2001, Oracle9i had seen a 55 percent market penetration. Next page: Oracle 10g RAC is looking at a good reception
The enabling underpinning of grid, of course, is RAC (Real Application Clusters) technology. Oracle9is version of RAC was a vast improvement over the earlier incarnation of this technology, OPS (Oracle Parallel Server). Oracle touted 9i RAC as providing a high-availability and horizontally scaling cluster that didnt need undue customization to achieve scalability. As such, it was a big leap over OPS.