tmon
02-17-2003, 10:03 PM
Two of the biggest fears an editor must live with and respect:
1. A workstation C:/drive failure, with no spare drive available, and no Norton Ghost backup. All I can say here is: Just do it!
2. A RAID drive failure.
RAID Level 0, while offering optimal read/write speeds, provides for no redundancy of precious data. A single drive failure means that all of your data is immediately unavailable. Also, making a quick diagnosis as to which drive is at fault may be problematic. If you are an editor/producer who faces tight deadlines, I would strongly recommend some kind of tape backup or perhaps parallel IDE storage protocol. One can choose to believe that multiple drive platters, spinning at 10,000 RPM and getting pounded for hours and hours on end won't fail when you least expect it to, but you do so at your own peril.
While RAID Level 3 provides for slower read/write times, its main benefit is data redundancy. Add a hot swappable spare, robust, intelligent software, and you can be back in business relatively quickly.
The case for RAID Level 3 was made for me last Thursday evening.
I was deep into a 15 minute video, perhaps around 40% done, when our T[2] froze up and I heard a beeping sound from one of the RAIDs. A red light appeared on one of the drive positions, indicating the specific drive failure. Because the RAID had a spare and proper software installed, the RAID began to rebuild itself immediately. Over 300 GB of data (much of it belonging to other ongoing projects) was restored to full working order in under 3 hours. I didn't have to redigitize ANY of the multitudes of source elements needed for the project.
Waking up a bit earlier on Friday, I picked up the edit session where I left off and was able to make my 2PM deadline (sure, I could have used another hour to tighten the audio mix, but I got it close enough to satisfy the client).
It is very doubtful that I could have made my deadline if I was living on the edge with a single RAID Level 0 array and no backup. If I remember correctly, it took me over three and a half hours just to capture all of the source elements to the video drives. I am glad that we have the RAID Level 3 system that we have and I can't thank my engineer enough for recommending this system!
Unfortunately, for my personal system, I probably can't afford these relatively expensive RAIDs that we use at work. I'd love to hear what others are doing to deal with this type of nightmare scenario. IDE drive RAID backups? AIT backups? What are your contingency plans? Your data recovery protocol? How long would it take to get back to where you were before you suffered the drive failure?
***************
Our RAID vendor's company name is "Symation." The RAID we are using is called "Symation 'Raptor 8'" http://symation.com/
We have two 450 GB arrays striped at RAID Level 3, mounted as one huge 900GB RAID Level 0 RAID Device using the Supermicro P4DC6+ onboard U160 controllers
These RAIDs give us redundancy AND hot-swappable spares.
Our Supermicro P4DC6+/2GB RDRAM/dual Xeon 2.0 CPU and "Symation" RAID set up is providing 4 layers in real time (no waiting for the background renderer). With five or more layers, T[2] has to go into background rendering. We are getting 155MB/s throughput with disktest.exe/T[2]autoconfig utility.
1. A workstation C:/drive failure, with no spare drive available, and no Norton Ghost backup. All I can say here is: Just do it!
2. A RAID drive failure.
RAID Level 0, while offering optimal read/write speeds, provides for no redundancy of precious data. A single drive failure means that all of your data is immediately unavailable. Also, making a quick diagnosis as to which drive is at fault may be problematic. If you are an editor/producer who faces tight deadlines, I would strongly recommend some kind of tape backup or perhaps parallel IDE storage protocol. One can choose to believe that multiple drive platters, spinning at 10,000 RPM and getting pounded for hours and hours on end won't fail when you least expect it to, but you do so at your own peril.
While RAID Level 3 provides for slower read/write times, its main benefit is data redundancy. Add a hot swappable spare, robust, intelligent software, and you can be back in business relatively quickly.
The case for RAID Level 3 was made for me last Thursday evening.
I was deep into a 15 minute video, perhaps around 40% done, when our T[2] froze up and I heard a beeping sound from one of the RAIDs. A red light appeared on one of the drive positions, indicating the specific drive failure. Because the RAID had a spare and proper software installed, the RAID began to rebuild itself immediately. Over 300 GB of data (much of it belonging to other ongoing projects) was restored to full working order in under 3 hours. I didn't have to redigitize ANY of the multitudes of source elements needed for the project.
Waking up a bit earlier on Friday, I picked up the edit session where I left off and was able to make my 2PM deadline (sure, I could have used another hour to tighten the audio mix, but I got it close enough to satisfy the client).
It is very doubtful that I could have made my deadline if I was living on the edge with a single RAID Level 0 array and no backup. If I remember correctly, it took me over three and a half hours just to capture all of the source elements to the video drives. I am glad that we have the RAID Level 3 system that we have and I can't thank my engineer enough for recommending this system!
Unfortunately, for my personal system, I probably can't afford these relatively expensive RAIDs that we use at work. I'd love to hear what others are doing to deal with this type of nightmare scenario. IDE drive RAID backups? AIT backups? What are your contingency plans? Your data recovery protocol? How long would it take to get back to where you were before you suffered the drive failure?
***************
Our RAID vendor's company name is "Symation." The RAID we are using is called "Symation 'Raptor 8'" http://symation.com/
We have two 450 GB arrays striped at RAID Level 3, mounted as one huge 900GB RAID Level 0 RAID Device using the Supermicro P4DC6+ onboard U160 controllers
These RAIDs give us redundancy AND hot-swappable spares.
Our Supermicro P4DC6+/2GB RDRAM/dual Xeon 2.0 CPU and "Symation" RAID set up is providing 4 layers in real time (no waiting for the background renderer). With five or more layers, T[2] has to go into background rendering. We are getting 155MB/s throughput with disktest.exe/T[2]autoconfig utility.