Google Fault-Tolerance Simply Too Good

Dec 3, 2003 • 10:11 pm | comments (0) by twitter | Filed Under Other Google Topics
 

As many people know, the www-sj data center at Google simply died (I don't recall how long ago, let's just say 4-8 weeks ago). At first I thought to myself, "Oh, routing problems or something... that's the whole purpose of data centers." Then after a week I thought, "Oh, they are simply taking that data center down. That's the one that was in their main office, and they need the physical room to move in the IPO crew."

But now... many weeks later, it still resolves within their DNS servers. If it was dead, why would it remain in DNS? So now I'm thinking that someone hit the power strip for that data center with their toe on accident (maybe it's under a secretary's desk or something). Google built an ultra-efficient fail over system with their multiple data centers around the world. That way the end user has no idea when something goes down. It looks to me like it's too efficient. Google doesn't even know it's down. :)

Previous story: DigitalPoint Selected as "Coder Author"
 

Comments:

No comments.

blog comments powered by Disqus