White Paper - RAID (s)
January 12, 2004
By Jon Golden
I think some clarification of wants vs needs vs RAID levels is in order.
In our video world there is a need for massive amounts of storage and speed to deliver that data to our NLE's: But having large amounts of storage means that any drive loss can spell doom on your project. So regardless of what kind of drive/raid/storage solution you decide upon you MUST have a BACKUP plan that works (and you have tested) in place.
If your needs are just lots of storage, then I would recommend buying large Firewire drives and hanging them off your system. Remember to think seriously about your backup procedures. A dead drive is lost time/data = money.
BUT... If your needs are to greatly improve Data Transfer Speed and/or you need some protective Redundancy then you need to consider a RAID array...
RAID is defined as Redundant Array of Independent Drives. Raid arrays come in 2 forms: Hardware or Software (simulators) For the sake of this discussion I am only considering hardware Raid controllers as I would never ever trust a software Raid scenario in my heart of hearts (sorry software vendors)
There are a variety of RAID levels (levels of redundancy/performance) Currently IDE/Firewire solutions seem to only support Raid 0,1,5. SCSI supports 0,1,5, and above.
I have tried to summarize this in English and techlish:
RAID 0 - If you just want flat out SPEED Raid 0 is the way to go. Lots of vendors offer this solution.
TECHNICAL: RAID Level 0 provides no redundancy. RAID Level 0 splits or stripes the data across drives, resulting in higher data throughput. Since no redundant information is stored, performance is very good, but the failure of any disk in the array results in complete data loss. Level 0 is only used to increase disk performance.
RAID 1 - If you want some level of redundancy but cannot afford Raid 5 then go with Raid 1(mirroring) you do not gain any speed but you do have the advantage of having an instant 2nd copy of your files. But if you are NOT using some form of backup that works for your environment then this might be a waste. Since, if you write bad data out, it is bad in 2 places instantaneously AND in the very small possibility that the controller goes bad you will write garbage to both drives thus rendering them both useless... Again lots of vendors offer this solution.
On our Xserve's we run Raid 1 on 2 drives, and then we use Carbon Copy Cloner to perform a scheduled clone/rsync/backup every night to a third drive, we also perform a backup to tape... Apple does not support R5 on Xserve's, although I see where FWB software does... I just do not wish to have FWB drivers installed on my Xserve boxes.
TECHNICAL: RAID Level 1 is usually referred to as mirroring. A Level 1 array provides redundancy by duplicating all the data from one drive on a second drive so that if either drive fails, no data is lost. This is a good entry-level redundant system. Most newer motherboards have this option built-in today using the Promise Technology IDE RAID chip. The downside to this is that the cost per megabyte of disk storage is twice that of a single drive as two drives are needed to store the same data.
RAID 5 - SPEED, and Redundancy: Raid5 is the way we go with all critical systems. Our office file servers are all R5. I have lost single drives in many of these systems at various times and have NEVER (knock on wood) lost any information. Drives can be pulled when their indicate they might go bad and replaced on the fly (w/o stopping the array or usage) and the array re-built. Depending on the controller (SCSI) you can also tune the read write characteristics to better suit your READ or WRITE needs... This can be almost as fast as RAID0 if the right drives and controllers are used. And provides complete safety of your data. In addition some controllers (SCSI) support extra spare drives which will automatically be added to the array in the event of a drive failure and depending on how many drives are being used you can lose as many as 2 drives at the same time w/o loss of data. Many people offer SCSI Raid 5 but it is expensive and you need to choose a vendor and/or your components (Do It Yourself) with more care... as controller/OS/software issues can make or break you...
FOR DIY'ers www.infortrend.com is the OEM that makes many of the actual hardware controller boards you will find inside the pre packaged manufactured arrays. For any SCSI Raid Arrays you might purchase I would recommend Adaptec controllers as they are the ones most likely to be OS 10.3.x >> onward compatible. I personally use a Adaptec 39160 controller in my FCP machine it will allow me to hang as many as 30 drives off of it..
I did a cursory search for a "relatively" inexpensive (USD$2-$3000) RAID 5 FIREWIRE solutions for the Mac and came up with:
They seem to have a 1TB R5 limit... So currently for arrays larger than 1TB I believe you will need to go with a Fibre Channel Solution from www.apple.com - Xserve RAID - 1-3TB (4-14 drives) or a Fibre Channel or SCSI/SERIAL ATA solution from other aftermarket array makers like:
www.raidinc.com (4-16 drives)
www.medea.com (4-10 drives)
TECHNICAL: RAID Level 5 stripes data at a block level across several drives and distributes parity among the drives. No single disk is devoted to parity. This can speed small writes in multiprocessing systems. Because parity data is distributed on each drive, read performance tends to be lower than other RAID types.
The actual amount of available storage is about 75% to 80% of the total storage in the disk array. The storage penalty for redundancy is only about 20% of the total storage in the array. If one disk fails it is possible to rebuild the complete data set so that no data is lost. If more than one drive fails then all the stored data will be lost. This gives a fairly low cost per megabyte while still retaining redundancy.
An easy formula is 4 drives of 120gb == ~360gb of storage. 5 drives of 120gb == 480gb of storage. you must have minimum of 3 drives in a RAID 5 configuration. i.e. lose a ~drives worth of space.
For an even more complete course on hard drives and RAID check out:
RAID levels 6--> 15 if you are looking into this arena write me offline...
Jon Golden, is a Virginia based still photographer http://www.jongolden.com that does some FCP video work, but has a background in Computer Science, and has consulted IT in the publishing industry for many years. He has been playing
copyright © Jon Golden 2004