Before the sodas were digested our detectives got called back to Hotel Network.
A brutal crime had taken place. A hard drive was found dead in one of the storage bays. Foul play was suspected. Was it murder or suicide? Or was the death caused by ‘natural causes’?
Preliminary reports suggested a clear case of overwork – the poor guy’s spindle had burst, stopping data flow and causing instant death.
Curiously, Hotel Network wasn’t affected. It continued to function. How did it manage that?
“Hard drives have two major drawbacks. They’re slow and they fail”, declared the OS. “I would be stupid to rely completely on them.”
“You obviously use backups.”
“Obviously, detective. I don’t need to be told how to run my shop. But one doesn’t have time to run around looking for hard drives in Hotel Network. Data must always be present and ready for use.”
“You mean to say, even if drives are slow and unreliable, they still have to be fast and reliable?”
“Well put, detective. That’s what we do here. Ever heard of the story of the father who taught his sons that one twig can break easily, but twigs tied together are harder to break?”
“How on earth do you manage that?”
“Easy. I create an army of hard drives working together. By taking away their independence, I make them stronger and faster.”
The slave driver!
Imagine watching a movie with your friends, from a file stored on a hard drive. Just when you reach your favorite scene, the drive fails. Your friends pass disgusted looks all around, except in your face.
You mumble a few apologies, curse the drive manufacturer, and walk to (or drive or fly or log on to) your backup area, take out the backup disk – and insert the backup. Then, you connect another backup drive (you no longer have a backup of the backup, remember?) to copy this disk. After the copying is done you disconnect the new backup and take it back (or drive or fly or upload to) to your backup area.
Then you get back to your movie and play. Your favourite scene isn’t your favourite anymore. Your friends have already updated their social network statuses.
Studies have shown that the typical drive failure rate is about 2% to 5% on average. This kind of performance hit is acceptable to most consumers – that’s why computers and laptops come with single drives. But it’s not acceptable for mission critical work where your life, or your pay check, depends on it.
In video production, time is money, and it is critical to have your expensive color grading artist work on the grade rather than wait for an hour or more for data to be replaced.
It is also important for a DIT on a set where everyone needs real-time feedback of what is being shot. Nobody has time or the money to wait around while you hunt for that backup disk.
The writing is on the wall – one needs a system of backup that is real-time. If one drive fails, another drive must take over and the system must continue to function. What kind of system does this?
Enter RAID, or Redundant Array of Inexpensive Disks (Some substitute Independent for Inexpensive, but I’ll stick to the original acronym).
A bunch of disks in RAID is called an Array, so we often refer to them as RAID arrays. In simple terms, if JBOD is a bunch of goons then a RAID array is a special forces unit – highly organized and prepared. A drive in a RAID array is expected to fail at any time. Nobody is indispensable.
Let’s get one thing clear – RAID is not intended to be a replacement for data backup. You still need to copy your data to other drives or the cloud or whatever to keep them really safe.
The commander of the RAID array is a slave master. He is called the RAID Controller. RAID controllers can be hardware or software based. In the case of Hotel Network, the OS was controlling the RAID array directly.
Could this have been the reason for the drive’s untimely ‘failure’?
“You can’t suspect me!” remonstrated the OS. “I’m the Operating System! Drives fail. That’s the reality of it.”
The detectives felt it was necessary to investigate Hotel Network’s RAID system in greater detail. Which RAID was it?
Links for Further Research:
- PC World Article on Drive Failure