It is always DO vs. Linode because both are relatively stable, relatively good for their price points. Tbh, the only real reason I use Linode over DO is the fact I've seen higher restarts on my DO nodes and the fact I've attracted DDoS's at a size Linode shrugs about.
Re:Digital Ocean NYC2 Droplets Down
by anderspetersson on Tuesday July 29, 2014 @06:15PM
(#8105335)
How do you do this on the DNS layer? Very short TTL and switching on failure detection? Do you use both providers at the same time when both are up via DNS round robin or similar?
Re:Digital Ocean NYC2 Droplets Down
by hadoukenio on Tuesday July 29, 2014 @06:29PM
(#8105385)
There's nothing stopping you from adding multiple A records for a domain. In fact, that's what a lot of people do:
dig a www.google.com
www.google.com. 300 IN A 74.125.237.178
www.google.com. 300 IN A 74.125.237.177
www.google.com. 300 IN A 74.125.237.176
www.google.com. 300 IN A 74.125.237.179
www.google.com. 300 IN A 74.125.237.180
Browsers these days will resolve and try each one, on a first-come-first-served basis. If one fails, it will try the request on the next server.If you have your own IP-space you can use BGP to route IPs to different locations.
If you're relying on DNS then short TTLs are pretty reliable, but there can be extended outages due to erroneous client-caching.
In the age of DevOps, it's not hard to configure different providers. I have been using Linode and DO for a while now. I noticed DO go down last week, checked my site, and it was UP because it was being served via Linode.
I flinch every time I hear someone say to me that their service has redundancy because they use multiple Amazon regions. Multiple providers people! And preferably multiple providers in multiple states, paid for by mutually exclusive credit cards!
Redundancy doesn't mean having multiple hosts by the same provider, within the same data center, within the same rack, provided by the same power strip.
tl;dr: Plan for outages.