Hacker News new | past | comments | ask | show | jobs | submit login

Still, sooner or later, the data center will be hit by a natural disaster, a DoS attack, a network problem, or the like, and you'll have to be ready to move to a different one to get your service back online. Or you'll have to reboot your server to apply a critical kernel security update, in which case you need to be ready to fail over to a hot standby. So, since relying on a single server with high-uptime hardware is penny-smart and pound-foolish, might as go with a cloud-style architecture with commodity hardware.



I use to be fascinated with datacenters and would masquerade as a customer prospect to get a tour and see all the cool gear. I was asking one engineer about what they're plan was for a tornado (this was at ThePlanet in Dallas TX way back when) and they basically scoffed at the question. A week or so later one briefly touched down about 1/4 mile from them, I wonder if they thought about me when the sirens were going off hah.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: