Complete RAID array failures like the one we had last week don't happen often. In reality, given all the precautions we take and the quality hardware we use, they shouldn't happen at all.

Improvements We're Implementing

But it is essential to be prepared and so we are implementing some improvements that will, we hope, both cause these incidents to happen less often in the future and enable us to restore service more quickly when disaster strikes:

• All of our new VPS servers are RAID10, instead of RAID5. This means greater fault-tolerance.
• All of our new managed servers will soon be RAID1.
• We're implementing a more agile backups system, so backups take less time, use fewer resources and can be restored from much more quickly, perhaps as much as 5 times faster.
• We will continue the practice of keeping at least one idle server of each type in service on-site and ready to be restored to.
• We will continue the practice of keeping enough spare parts on-site to fix most any problem.
• We are working on strategies for keeping multiple copies of your data on- and/or off-site.

Changes in the Backups System You Should be Aware of

The more agile backups system will require some changes that you should be aware of:

• We won't back up files larger than 2GB. If this presents a problem for you, please email support.
• We will only back up certain directories. We won't back up the entire account from / down any more.
• We won't back up any files named Data.fs.*, only the Data.fs. All the large Data.fs.* files sap a lot of the agility from backups.

More Information

You can read more about the specifics of the new backups system in the Backups FAQ.