The 3 Rs of Server Backup Reliability: RAID, RAIN and Replication
When it comes to server backup reliability, you can never have too many copies of your data. Although they have certainly gotten better over the years, disk drives do fail. So do controllers. So do data centers. Securing data requires a multi-layered approach that uses RAID, RAIN and Replication.
David Patterson, Garth Gibson and Randy Katz of UC Berkeley’s Computer Science Division first described RAID in their 1987 paper A Case for Redundant Arrays of Inexpensive Disks (RAID). They listed five ways to improve storage speed and reliability by using multiple disks. To increase read/write speed, RAID stripes data across several disks allowing multiple head to simultaneously read and write data.
Because striping data across multiple disks also multiplies the opportunity for data loss, RAID takes two approaches to protecting data: storing multiple copies of a file on separate disks or calculating parity data which can be used to reconstruct data in the event of a failed disk.
There were originally five RAID levels but many more variations have evolved over the years. The most common types of RAID are:
• RAID 1 is purely concerned with reliability and stores identical copies of data onto multiple disks to protect data in the event one drive fails.
• RAID 3 uses a minimum of three disks and stripes the data across two disks, using the third to store the parity data.
• RAID 5, the most common, stripes data blocks and parity data on the same drives rather than dedicating a drive to the parity data.
RAID 0 has no redundancy and stripes a single copy of the data across multiple disks without generating any parity information. There are also the hybrid RAID methods RAID 10 and RAID 01 which combine the features of RAID 0 and RAID 1.
The most recent addition is RAID 6 which stores two parity blocks per stripe. While RAID 5 can survive the loss of a single drive, RAID 6 can survive the loss of two. This means that even if a second disk goes down during a rebuild, all the data is intact.
Remote Offsite Backup Moving Beyond RAID
RAID is a good, proven method of preventing data loss and it should be used. However, it isn’t enough.
Disk capacity has grown much faster then I/O or network speeds, lengthening the time it takes to restore a failed drive. In the 1980s, RAID could restore a disk in minutes; today’s multi-Terabyte disks may take hours. In addition, controllers, racks, switches, data centers, power grids and Internet segments can also go down, preventing data access.
Zetta takes a multi-layered approach to server backup data redundancy using RAID 6 so even if two disks in an array go down, none of the data is lost.
In addition to RAID, Zetta uses Redundant Arrays of Independent Nodes (RAIN). Instead of relying on a single storage array to safeguard your data, every single byte is stored on a completely independent storage node. With RAIN, if any component in one of the RAID 6 arrays goes down, complete access to data is still available.
Finally, Zetta replicates data between two data centers at opposite ends of the country (each of which using RAID 6 and RAIN) so if a disaster hits one area, data can still be accessed from the other data center.
There are still some scenarios that could cause data loss: an alien invasion such as in War of the Worlds or Independence Day; the Mayan apocalypse depicted in 2012 or the rise of Skynet from the Terminator franchise. But outside the realm of science fiction disaster movies, Zetta Data Protection has you covered.
To see what our customers have to say about Zetta's server backup reliability – read some case studies or start a free trial today.