As a Storage Administrator and Consultant working on a lot of different storage roll outs for different requirements, I’ve seen a lot of points of failure and disaster relating to enterprise shared storage. However one scenario that struck me the most is a misconception on storage.
I've responded on a few occasions where a client facility is unable to access their shared storage. After a quick investigation, I found out that their RAID set has died due to failed drives. Right then and there, the first client reaction was – “I thought we were safe with this storage? I thought we had redundancy? We have a backup, isn’t that the RAID?
This month’s blog is a quick look at what RAID or Redundant Array of Independent Disk is and what it is not. I won’t discuss full technicalities about it, but a simple clarification for everyone who owns one.
RAID is a protection. Not a backup, not an archive but a protection. And like any protection, it has different levels. Depending on the level, the minimum number of drives you need is 2 in order to create a RAID.
Another thing that a RAID offers is performance. Not needing to go too technical but the reason why it offers performance is because you group independent disk together to work as 1. It’s like having 4 dressmakers working together to come up with 1 dress.
So to some it up, a RAID is a level of protection that provides performance. A RAID is not in any shape or form a backup, or an archive.
If a disaster has breach the level of your RAID protection, then obviously your data is compromised and in this case you lose data. Again, a RAID is not a backup or an archive of data.
Now that we’ve established that a RAID is for protection and performance, lets discuss the levels that we normally use at Digistor:
RAID 0 – Normally called as a Stripe RAID Set, this offers zero protection but maximum performance due to the number of drives you have on the set. In a production environment, you can have this type of RAID for you to render files or cache. Only because render files can easily be re-rendered or cached are temporary files, you don’t need that protection.
RAID 1 – Normally called a Mirror RAID, offers the best protection but minimal performance. Not to mention costly as you end up buying 2 drives for the performance of 1. In a production environment; you can use this RAID for databases, metadata or anything that will only need minimal bandwidth like small files. Important files can be stored in there as if you have 1 failed drive, another one is the exact replica where you can still work on.
RAID 5 – This type of RAID offers performance and protection. Data is distributed to all the drives in the RAID set which provides performance. Protection is via a distributed parity on all the drives on the set. This RAID level allows you 1 failed drive and still have the data in tact and assessable. But don’t wait for a 2nd failed drive as that will render your RAID5 incomplete and you lose your data
RAID 6 – This type of RAID is like a RAID 5 but with double parity. This increases the protection as it can handle 2 disks failure. While performance on the RAID 5 is better than the RAID 6, this is often a preferred RAID set for larger disk capacity as it takes a long time to rebuild a single failed drive and the chance of a 2nd drive failure may be around the corner.
Those are the 4 most commonly used RAID Levels we use in Video Production. Different storage requirement may need a specific RAID Level or even a combination. The main thing to understand is that a RAID is not a backup or an archive.
Digistor has 20+ years of experience in the Industry specializing in storage solutions. If you have a specific storage require, talk to us about your needs.