Why do servers “go down” for routine maintenance? Is there not a backup where traffic can route?
In: 6
The infrastructure, both hardware and human, for high availability is expensive and it gets more so the more uptime you try to commit to. Being up 99.9% is massively cheaper than 99.99% is massively cheaper than 99.999 and when you get into something with quad 9s like say, Visa, the cost is astronomical.
For most companies, going down briefly every now and then really isn’t a big deal and they don’t bother. Their users will check back in a bit with little or no ill will or negative impact. For uptime critical companies, like say amazon, google, facebook or Visa, they pour huge resources into preventing any such gaps.
There’s more nuance, some changes are easier to do with failover (running thr site or service on a backup while the main is upgraded and then upgrading the backup once the main is back online) than others and many don’t require any downtime at all. But the jist is that for a lot of things it just isn’t worth the cost.
Latest Answers