Disaster Management for Web-Hosted Data

(Technology Infrastructure Series)

In case you’re taking seriously guarantees about uptime, reliability, or backups advertised by website hosting companies, you should know that most guarantees of service are an idealized concept, especially if you use a low-cost web hosting service. Now, this doesn’t mean you should avoid low-cost web hosts. What you should do is give a little thought to the “what if’s” that may arise, and what you can do before they arise to minimize the pain when they do.

In this article, I’ll go through a few situations you might want to consider, and some options you can use to reduce your risk.

In case you’re taking seriously guarantees about uptime, reliability, or backups advertised by website hosting companies, you should know that most guarantees of service are an idealized concept, especially if you use a low-cost web hosting service. Now, this doesn’t mean you should avoid low-cost web hosts.¹ What you should do is give a little thought to the “what-if’s” that may arise, and what you can do before they arise to minimize the pain when they do.

Prepare for Data Vaporization

The first thing is to accept that, even with top-dollar sites, “Life happens”. No amount of post-disaster frustration will bring back your data if it has vaporized, and the compensation — typically the waiver of a month’s hosting fees at most — is usually meaningless when compared to the time it has taken to configure your site and enter your data. So, no how much you pay, nor how good the “guarantee” seems, you should always take care to maintain your own set of backups of crucial content, databases, and software configurations.

Some sites make backups easy by providing administration dashboards such as cPanel. If they do, you should take advantage of the user-friendly interface, and schedule periodic backups.

But even if you use these services, you should also go in through an FTP client and physically create a local tarball (assuming your webserver is a Linux machine) of your entire home directory and download this onto your own computer — Windows or Linux, either is fine. This is the only way to ensure that you have everything, including server logs, configuration settings, etc., stored elsewhere.

Handling Backups

Date your backups. Keep at least the last two. And retain at least one snapshot from every 6-12 month period, going back indefinitely. That way, if you don’t notice something has gone missing for some time, you’ll still have a reference to go back to. If this sounds prohibitive, remember: storage space is cheap. You can now buy a terabyte drive in an external enclosure for under US$75 (55 GBP).

Plan for Mid-Stream Disruption

Once your site is up and running, every connectivity hassle with your hosting company will be magnified a thousand-fold as you imagine frustrated users unable to connect and giving up on your site. To give yourself the greatest flexibility in finding a speedy resolution, I’d recommend studiously avoiding the “bundled” domain + hosting offers that are so common and heavily advertised.

Buy your domain name registrations separately from your hosting service contract, being sure to keep the two companies separate as well. The key is flexibility as an operating principle.² In the event of repeated problems with your hosting company, it is often simpler and cheaper (and less damaging to your users) to simply take your site-hosting business elsewhere.

If you’ve registered your domain name with a different company than your webhost, you can simply open an independent hosting account with a third company offering similar features and a similar software stack. While the original site is still configured, migrate your content and those critical configurations from one site to the other, testing as you go. Once the new site is ready, simply redirect your domain name traffic from the host with the problems to the site with the newly selected host, and you’re ready to roll again.

Since domain name redirection happens with the company you registered the domain with, you can see why it is important to keep this an independent third company. That way there is no conflict of interest, and minimum likelihood of additional complications. Once your web traffic situation is taken care of, you are free to continue to pursue a resolution with the previous company, or just cut your losses (chalk it up to “cost of doing business”).

The Concept of a “Warm Site”

If you’re truly in the mindset of disaster prevention, or if your website represents a critical operating asset that you simply cannot afford to have go down, then consider taking these precautions one step further.

To combine low monthly costs with seamless service for your users, consider opening that second hosting account before the problem happens and keeping it configured and current, ready for disruption. In this way, you’ll have a reasonably low cost duplicate server that serves as a “warm site” — ready to go at any time. In the event of disruption, simply redirect your domain name to the warm site, and your users will be slid over to the prepared backup while you resolve the issues with the primary site. You should know, however, that it takes a while for domain redirections to propagate through the system. For a level of redundancy that provides “flick of a switch” transitions, you’ll need true mirrored servers, i.e. an enterprise level hosting service.

An Ounce of Prevention…

If you’ve ever had the unenviable task of migrating hosting services under pressure, you’ll know that in the data business, a pound of cure would be a bargain — real life remedies can get much more costly. So, start those backups, consider those what if’s, and start taking steps toward regular, sustainable disaster management of your data.

Footnotes