LM10

For them, even a millisecond down time would count as the page being down. We don't know how many of those they've had, but I'm pretty sure it has been down since 2003. We just don't notice because it's back within no time.

They go to the top schools, and recruit the top guys from the top schools. Imagine how a company would be that had the top students from yearly classes at MIT, Stanford, CMU and Berkeley working for them. That's Google.

1. What sort of volume of logging data do you'll have to peruse through on a daily basis?

2. What are some of the most challenging incidents you have faced while trying to maintain uptime?

3. What level of uptime do you attempt to maintain? How many "9s"?

4. What series of checks would you follow if something was wrong? What would be the first response strategy?

5. How do you run a website so well that people basically use it to check whether or not their Internet is up?

6. How often do you have to interface with the Information Security folks and what sort of incident response activities do you delegate to them?

There's only so much difference between a DDoS and a surge of legitimate users. Reddit often causes DDoS's of love on small websites that get linked here. Long story short, all popular sites are under some form of DDoS, it's the way that they handle the traffic that makes it awesome.

I remember reading somewhere that in the early days of Google, a data center caught fire. This was when Google was still barely 2-3 years old IIRC. Even then, the site had already been set up so that the persistent backups kicked in and there was practically no downtime.