Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I understand completely how to host my own website but my question (i hope this is the right place to ask it) is how to guard against, say your ISP going down or your router failing, things like that? Does anyone run their own website? Anyone have any advice?

share|improve this question

migrated from Nov 29 '10 at 2:07

This question came from our site for pro webmasters.

Since your main question was advise-related, I've got one: spend your money on a server. Companies like Hetzner or Serverloft offer unmanaged servers at awesome prices.

My main point is that if you'd like to be serious about your website or hosting business, you just have to have a serious approach. In the end those money come right back to you because of your great service which runs smoothly and has a great connectivity.

There really wouldn't be a big reason not to host your stuff yourself if there wasn't the connectivity. Speed is one of the few things that you just can't cut. The least you should do is try paying for it. You owe it to your visitors and mostly to your nerves.

share|improve this answer

I'm guessing that by "ISP" you mean whomever provides Internet into your home or office. Beyond that I'll also offer suggestions for dealing with co-location and multiple locations.

Probably the best way to avoid outages due to down-time in a connection to your home or office, via that ISP, is to use either a colocation service, managed hosting provider, or application provider like AppEngine. These are typically services that are hosted in a carrier-class data center with redundant power, cooling, and Internet connectivity, so they can be much more reliable than, say, a DSL or cable connection into your home.

Depending on the level of service these facilities might be able to survive Networking problems, power problems, and more with little or no impact on the reachability of your application. They will provide redundant routers, connections to multiple Internet connections, power transfer switches and more.

These services are available at all sorts of price-points, so it is rarely economical to try to build such a service yourself for only one or a few servers.

If your application really needs to be able to survive even a major outage, the typical solution is to get services at multiple facilities like the above, and design your services to either be able to fail over from one facility to another, or be able to run from both locations at the same time. This often requires extensive work to design and architect applications to operate this way, replicate data between locations, etc...

I will say that I find many people think about availability well before they really can justify implementing such a thing. If you have no users, it's probably more important to work on getting to the point where you can get users, rather than dramatically increasing the difficulty of getting there to make it able to survive a rare event. So my advice would be to keep your feet on the ground when deciding what level of redundancy and replication to do. Having fully redundant and distributed services is a long and involved process.

share|improve this answer

Normally this would be done with a cisco/sonic router with multiple ISP's connected. There is no way to protect a router from failing unless you have two routers with seperate ISP's (to redundent because routers still have to connect to the server, with another router or through to network cards on the server). Now when you setup your DNS for you IP address, most will allow a backup IP. When that is entered you will beable to get to your server from either ip (as long as all forwarding and firewalls are setup correctly).

share|improve this answer
You could set up a router as a backup and swap when it goes down. This is for least amount of down time. – James Williams Nov 28 '10 at 18:24
Normally? In my experience I've never come across a multi-site HA operatior hidden behind a SPOF. – symcbean Nov 29 '10 at 10:11

You could use an external Cache service - is free and the basic account would work - where your site would "appear" online even when it is not.

To make sure it stays up however - you would need an ASN and the ability to be multi-honed or use a DataCenter that is.

share|improve this answer

The best solution is multi-site with a transparent failover mechanism and capability of operating in split-brain mode.

Since you can't sensibly change the software running on the clients - transparent failover means round-robin DNS - perferably with location-aware (CDN) weighting.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.