eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
Who’s to blame when a Web site crashes? The site owner? The hoster? The supplier of the back-end software?
The blame game doesn’t seem to have started – at least not publicly – in the case of MySpace.com, the latest mega-Web site to go down.
MySpace is one of the largest sites on the Web. As of February 2006, there were more than 64 million registered MySpace users, with 260,000 new registered users signing up each day. The site was garnering 23 billion page views per month, according to Media Metrix data.
A power outage in the company’s Los Angeles-area datacenter took the entire MySpace site down completely for hours, starting on the evening of July 23. For hours after that, performance of the site was erratic.
“Apparently (it was) a single source of failure scenario for a systems with millions of users – yikes,” said Peter O’Kelly, an analyst with the Burton Group. “I suspect it was simply inadequate risk management planning on the part of the MySpace operations team. I doubt it was something related to Microsoft software problems.”
A MySpace spokeswoman would not comment when asked about the company’s disaster recovery plans. She also declined to specify how many datacenters the company runs, where they are located or whether MySpace has any kind of mirroring capabilities.
MySpace officials did provide some details about the company’s back-end infrastructure, however, during the course of a couple of Microsoft conferences earlier this year.
According to information presented at the Microsoft Web Summit ’06, as of February 2006, MySpace had three datacenters running 2,682 Web servers and 650 database servers. The company also is running 90 cache servers and 150 disks in its storage-area network.
At the Microsoft Mix ’06 conference in March, Microsoft and MySpace officials touted the fact that MySpace was running on Microsoft technologies on the operating system, database and development platform fronts.
MySpace was an early adopter of SQL Server 2005, MySpace officials said in March. The company also bet big on the ASP.Net platform. By relying on Microsoft’s architecture, MySpace was able to reduce the number of servers from 246 to 150 (in one or more of its unspecified datacenters), company officials said.
When asked whether MySpace was taking advantage of SQL Server 2005 failover or database mirroring capabilities, a Microsoft spokeswoman said neither of those technologies is provided by default, and it was up to individual customers how, when and if to make use of them.
Database mirroring was dropped from the base SQL Server 2005 product, and later provided as part of SQL Server 2005 Service Pack 1, which Microsoft delivered in April 2006.
In a “What’s Next” slide from the Microsoft Web Summit, Microsoft officials highlighted peer-to-peer replication, eliminating single point of failure and database mirroring, designed to reduce failover down time as being in MySpace’s plans. MySpace also is committed to upgrading to Longhorn Server and Internet Information Services 7.0, due in the latter half of 2007, according to Microsoft officials at that conference.