Do you host your main website on other companies servers? From what I can tell it's not a bad idea incase your own servers go down and you can't be contacted. We've still got alternate means (e.g: Twitter, which we heavily advertise in our emails) however it's nice being able to use the main website.

Do you? And if so, who are you with- (we're about to go with HostRelic due to their low storage, low price & 99.989 uptime report last year)

I hope the best way is to host your own website, many customers do a traceroute to see the incoming providers to your network.

---> From what I can tell it's not a bad idea incase your own servers go down and you can't be contacted.

Think about redundancy in case of your server crashed.
You can also install from outside of the newtork where your website is a monitoring system like nagios that will send you alert if your server/website crashed.

Yeah, it was mainly in case the website goes down- that was they/we can still get in contact should anything *hope it never happens* goes wrong.

Never host your company's website on the same server you host your clients, always keep them separate. Where you host it really doesn't matter as long as it's a reliable and stable provider. This could be your own company or another company.

I suppose you can host your own site on a remote VPS too. It's definitely not a bad idea to host support.yourdomain.com remotely, so that communicating with the customers while a major incident takes place will not be a a problem.

Really good idea that! Most of it is done via email as opposed to ClientExec's support system (I don't feel it's quite there yet- thus why I'm looking forwards to CE 4.1)

Thanks though!

That's also an excellent setup . So you could have something like:

yourcompany.com = Server1

support.yourcompany.com = Server2

yourcustomers = Server3

and so on.

If you are so concerned about website availability and don't want to face any downtime, then I think you should also consider load balanced dedicated servers. Means, if one server is out of service, then the whole system will still work. The load balancer will recognize the failed node, and mark the crashed system. The total performance of the load balancing cluster will be reduced, but services will still be provided.

Normally it is called as a server mirroring & it's a process of channeling data from the Master server to the Slave server so that an identical copy of data exists on the Slave server. In other words server mirroring can also be described as utilizing a backup server that duplicates all the data of the master server. After the implementation of mirroring, if for any reason, the master server fails, the slave server can immediately take its place without any downtime.

A load balanced situation like Paul suggested is great, but keep in mind that the site should be located in different physical datacenters and ideally, different networks.

Also, you should have a backup in place should your DNS severs go offline. This will allow you to re-route your site through an alternate machine.

For us, our site is within our network, but we have two duplicate sites located elsewhere (one in England). If the datacenter has issues, we can quickly modify the DNS on the domain name and point to the other location.

For ultimate protection, using a LIVE DNS system where you can change the DNS within a few seconds and use a Round-Robin sitution, this would provide the ultimate protection.

We use 5 servers to host our site currently - each one dedicated to a single task of our system. One for site, one for support, one for billing, another for blog etc. Keeping these separate from client sites is also strongly recommended.

Nice setup Conor!.. And very nice tips as well. Talk about redundancy!

I host all my sites to hosting providers ( micfo and knowhost ). It is cheaper this way rather then learning / employing someone to look after my own server.

We've now migrated our main site to RelicHost due to a good friend's personal tracker showing almost 100% uptime (exactly what I wanted!)

Thanks for your help

100% uptime over what period of time? weeks, months, years?

He monitored whole of last year (:

I suppose that Conor is a bit suspicious of such perfect uptime, as I am too. I guess it depend a bit on the precision of the monitoring tool as well. If the checks are done every 30 minutes, some short downtime occurrences can easily be missed.