Fail-Over Website Hosting Caching Engine

Two Copies of each Website

You have the option of maintaining a full copy of each site, but then you are responsible for 2 copies and their databases. Because this is a nuisance and keeping databases in sync is a programming exercise for each site, we built a caching engine to keep a static copy of each site.

Also, since most websites are only static or semi-dynamic, the cached version should be adequate in most instances.

The Globi Caching Engine

We built our own caching engine that scrapes all website content. It’s rules are to follow up to 1000 links per site. Any object older than 3 days is checked for a change, and after 30 days everything is re-cached anyway. At worst, a cached copy will be 30 days out of sync with the live site.

Your cached website will be accessible via cache.yourdomain.com

For those pages which are dynamic on the live site, the caching engine will issue a warning on posted data, informing visitors to use phone or email as communication methods currently.

per Example

To see the cached copy of THIS website in action, see cache.globihosting.com.

SEO Considerations

In order to prevent spidering of cached content on a cache.subdomain (duplicate content issues), all cache.subdomain requests will return a robots file which tells all bots not to spider anything on this subdomain.

In the event that a primary server goes down, DNS will point to the caching engine within under 5 minutes. In this case, you will not end up at cache.yourdomain.com but instead at the regular www.yourdomain.com. Here the robots file is non-existent as we don’t want to stop spiders when there’s a fail-over situation.

Conclusion

In the big-picture, this is not intended to replace any of the live sites. The whole idea behind the caching engine is to have a mostly correct website viewable and usable in the event of an outage of a primary server.

Primary server outages hardly ever happen, but when they do, most outages last from 5 minutes to an hour or two, depending on the problem. Our fail-over solution will serve the cached site within under 5 minutes, and automatically revert to the live site when it comes back up.

Will you even notice when this occurs?