Oracle goes down on Sun
The consensus I've been hearing is that Oracle is buying Sun to improve its cloud computing capabilities. Certainly its Stellent content management system could have done with a boost yesterday morning when Oracle's corporate site found itself unable to cope with the traffic volumes hitting its news pages looking for information. CMSwatch analyst Adriaan Bloem reports that Oracle's servers "were failing for several hours [Monday] morning to bring up their 'pretty page' (from a template) that states the application server is failing to actually serve the news."
Oracle later managed to put up a static page recommending that site visitors call 1.800.ORACLE1, a strategy that Bloem described as "more than a single point of failure — it's like a fail whale, with Jonah failing inside." I haven't heard exactly how Oracle's call center staff handled callers seeking to download its press release or find out co-ordinates for its morning investor call, but I have a mental image.
The PR people at Clickability alerted me to this tale, as it happened to coincide with the on-demand content management vendor announcing it has strengthened its infrastructure with a second data center. Clickability, which says it has maintained 99.99% uptime while serving up to half a billion pageviews a month over the past ten months, prides itself on its ability to handle traffic spikes on behalf of its customers — many of which are in the media business. Fortuitously, its press release happened to mention yesterday how the provider's "SaaS-based platform has enabled many customers to handle surges in traffic effortlessly and avoid the risk of losing page view revenue or potential prospects due to slow performance or downtime."
Clickability attributes this 'just-in-time' scalability to "the power of its massive multi-tenant architecture," a topic on which I've had several run-ins with Oracle over the past couple of years. Oracle believes that SaaS delivers better performance to large enterprises if instances are tuned to the specific needs of individual industries and customers (an architecture that, coincidentally, requires a separate database license for each customer). Perhaps it believes it can tune its on-demand infrastructure even more finely if it owns the underlying hardward technology. But as Bloem notes, successfully scaling capabilities such as dynamic content delivery "is not just a matter of buying a Sun server." Nor am I fully convinced that Sun has the answers Oracle believes it is looking for, given the inauspicious circumstances that surrounded the initial launch of Sun's own cloud platform three years ago. But perhaps Oracle thinks it can do better than that — so perhaps the failure to come will be yet more spectacular.