The Linode Part of My IT Strategy
At the moment, I have 2 Linodes: a production server in one data centre, with a backup/test server in another.
What I am trying to work out is whether I really need the backup/test server, when:
1) I can do all my testing on local servers (to which my primary Linode is backed up daily.)
2) I am doing less client hosting on my own server - setting up separate Linodes for clients on their own accounts.
3) If something really bad happened to the host my production server is on, I would have to make loads of DNS changes which would take time so:
4) It might be quicker to start from scratch with the replacement host.
5) It could take just as long getting my backup up to speed as doing (4).
What do you think, folks? Dump the backup (can't really see the advantage of it now) and spend the money on upgrading my primary?
Comments on this would be much appreciated. I'm trying to cut costs - time more than financial. One less box to look after is rather attractive, especially with the burden of new ones set up for clients.
Cheers
M
2 Replies
DNS changes don't necessarily take long. You could use short TTL values. You could convert as many subdomains as possible to CNAME records, so you only need to change one A entry and everything else just follows. If you manage multiple domains, you can automate the task with the Linode API. Or even better, you can simply switch IP addresses between the broken linode and the new one.
If you want to do the start-from-scratch route, consider my example: I keep a list of every package I installed and every configuration file I customized. I feed it into a home-brewed bash script, along with the firewall entries and the exact rsync command to restore my data from the backup. When I provision a new server, I just upload and run the bash script. 5 minutes + the time it takes to restore the backup, and it's ready.
It pretty well confirmed what I was already thinking.