Disaster Prevention and Backups at Asgemi
I remember reading earlier this year about a blog service that was deadpooled after the drives that housed their entire database were overwritten. Yikes! As a result, everything was lost, and they closed down.
I’m sure that everyone involved, from the blog developer to the bloggers themselves were heartbroken. Prior to launching the site, it was on my mind about what we should do at Asgemi to avoid something like this from happening to our community, and I wanted to briefly share with you what’s going on behind the curtains.
Asgemi is hosted through a company called Nearly Free Speech, which is highly supportive of artistic expression, but has also struck me as very tech-savvy as well. As they explain on their website, they do automatic rolling backups, and use mirrored RAID 5 (with 4 hot spares per array) and RAID 6 (plus hot spares) to protect live website data, and extensive multi mirroring for MySQL data. Which, incidentally, in the event of a serious catastrophe, the most harmful consequence would probably be the loss of everything since the most recent backup, which is generally about 12 hours for sites and 24 hours for MySQL processes. In practice, the time-to-first-backup for new data (web sites or MySQL) averages 12 hours.
In addition, at Asgemi we also use a secondary service hosted by Rsync.net to backup off site copies of our key databases, all of our custom server source code, and a lot of configuration information every 24-hours.
While there is no one perfect system to prevent a major disaster, what’s key is that there are a number of redundant systems in place to provide backup protection. We think that this is a responsible approach, and hopefully gives you peace of mind that we take the community, and what you add to it very seriously.
Recent Comments