insanityfw
Ban Me Please!
THOM. Love the idea of mirrored writes!
One new spec...
- Thread starter Jannard
- Start date
ThomSteinhoff
Member
Actually I like the idea of a second Red Raid mirroring the first better than 4-5 drives raided together in a single monster red raid, which does allow for internal redundancy but who is going to swap out the bad drive in the field? Server raids work because they have swapable bays--but how much overhead is swapping bays going to add to the red raid case?
I like the idea that Red Raid is as essentially sealed as can be, stackable, semi-indestructible. If it fails, it fails (you have a live backup in the form of a slave red raid) You label it bad, kiss the good backup, and grab another one--you are not stopping everything to do open heart surgery in the field to swap out the bad drive to ensure it works--you just move on. Also, hot swap bays add a level of mechanical and connection fragility that I would rather avoid.
Incidentally, in this slave scenario, when you grab the new one and hook it into the slave position, it should immediately get to work giving you another backup of what has been shot already and you are back in business.
Also, if you pause midstream to take off the slave to pull off and verify shot footage and run for a bit with no live backup, when you connect again it should re-sync the missing footage, bringing up to date you backup.
It would also be great if you could plug the slave backup into a laptop, copy it off to cheaper, redundant backups, and somehow flag the footage as "captured" on the red raid, so when it returns to the camera environment it can tell the master to recover the space of captured footage at the same time it is catching up with what it missed in the interim.
This may sound like overkill, but I am trying to work out a workflow in my head that is not too complicated but is also safe, easy, repeatable and doesn't require dozens of red raids.
If this mirrored strategy, I can also envision a workflow of 4-6 red raids, where they always work in pairs once the two are "exposed" or filled with data, they are replaced with another two. then one of the exposed drives is emptied and only when the save is verified they can both be wiped.
I like the idea of a rotating slave better but it may be more complicated to understand.
I like the idea that Red Raid is as essentially sealed as can be, stackable, semi-indestructible. If it fails, it fails (you have a live backup in the form of a slave red raid) You label it bad, kiss the good backup, and grab another one--you are not stopping everything to do open heart surgery in the field to swap out the bad drive to ensure it works--you just move on. Also, hot swap bays add a level of mechanical and connection fragility that I would rather avoid.
Incidentally, in this slave scenario, when you grab the new one and hook it into the slave position, it should immediately get to work giving you another backup of what has been shot already and you are back in business.
Also, if you pause midstream to take off the slave to pull off and verify shot footage and run for a bit with no live backup, when you connect again it should re-sync the missing footage, bringing up to date you backup.
It would also be great if you could plug the slave backup into a laptop, copy it off to cheaper, redundant backups, and somehow flag the footage as "captured" on the red raid, so when it returns to the camera environment it can tell the master to recover the space of captured footage at the same time it is catching up with what it missed in the interim.
This may sound like overkill, but I am trying to work out a workflow in my head that is not too complicated but is also safe, easy, repeatable and doesn't require dozens of red raids.
If this mirrored strategy, I can also envision a workflow of 4-6 red raids, where they always work in pairs once the two are "exposed" or filled with data, they are replaced with another two. then one of the exposed drives is emptied and only when the save is verified they can both be wiped.
I like the idea of a rotating slave better but it may be more complicated to understand.
ThomSteinhoff said:
Those are some great ideas Thom. Possible? I have no idea.
But I'd like to expand on one point above, and that is the issue of verifying footage after transfer. It may not always be possible to review transferred footage before moving on. And 'moving on' often means overwriting your REDDRIVE / REDminRAID.
I wonder if it would be possible to include a tool in REDCINE which quickly scans through the metadata of each frame to verify that everything was transferred safely? Sort of like a data integrity check. I realise that the best method is to manually watch each shot and check file sizes, but you would almost need a dedicated crew member just to QC real time playback, especially @ 3+ hrs of 4K per REDDRIVE.
Perhaps Graeme or Rob could come up with a cool little REDCODE script that scans through 320GB of footage in, I dunno, 5, 10, 15 minutes while visually displaying a new 'screen grab' every 5 seconds or so. 10 minutes of potential downtime on set would be worth the wait IMHO - just tell your AD it's a trade off for not having to check the gate all day. :happy:
glenn chan
Well-known member
There's a utility called Spinrite which will write data onto the drive and then try to read it back. Some drives show errors on this the first time around. The Spinrite tool really shows how sketchy the drives themselves can be. Perhaps you could test every magazine with Spinrite. While that would take hours, it would be reasonable since you will likely recycle drives.
Or on the other hand, perhaps the camera could implement a higher level of error correction.
2- I know that Jeff Kreines is looking at the same problems, although Kinetta proposed a lot more drives. When you have that many drives in there, heat and vibration become a problem (Sorbothane solves the vibration?). Ideally, the camera would be ok if you dropped it, if the temperature suddenly changed, etc.
I know that hard drives don't like either. If you don't cool the drive, it can get up to 85 degrees centrigrade and stop functioning; that happened to me. A hard shock to the drive will also jostle the read/write head around and it won't work temporarily.
With a RAID, you kind of make your problems bigger in those aspects.
Or on the other hand, perhaps the camera could implement a higher level of error correction.
2- I know that Jeff Kreines is looking at the same problems, although Kinetta proposed a lot more drives. When you have that many drives in there, heat and vibration become a problem (Sorbothane solves the vibration?). Ideally, the camera would be ok if you dropped it, if the temperature suddenly changed, etc.
I know that hard drives don't like either. If you don't cool the drive, it can get up to 85 degrees centrigrade and stop functioning; that happened to me. A hard shock to the drive will also jostle the read/write head around and it won't work temporarily.
With a RAID, you kind of make your problems bigger in those aspects.
taubkin
Veteran
glenn chan said:
Well, we have to accept it, that it's the weak link of the chain. While RED made everything possible to get a new, "revolutionary" sensor, and a new "revolutionary" compression, they are still using off-the-shelf tech for their drives. The VERY good news, is the modular nature of the camera, so when something better comes around, it will be possible to upgrade it, or the flash module inside the camera, etc...
Looking at the industry as it is now, it's very clear that nobody has a perfectly solution for tapeless production: Large tape compression like HDV, large compression for XDCAM, small running time and expensive P2, etc...
But as something better is still around the corner, we will have to settle for low(er) reliablity...
alacritymedia
Active member
Flash for safety?
Flash for safety?
At the risk of sounding like a Panny rep pimping P2 there is just nothing like the warm fuzzy feeling of solid state storage on board. Sure it costs a fortune but you can reuse it thousands of times. Who's ready to spend $5K for a 64GB RedFlash (this is a guesstimate only) that would hold +/- 30 minutes of 24p/4k RedCode? Unless you said yes to the preceding question then the RedDrive or a tether is all we've got for the foreseeable future. So...
If your project is going to have the camera getting jostled on a regular basis rent a bunch of RedFlash modules. If your project is going to have the camera getting knocked about only occasionally then get one RedFlash for those shots and use the RedDrive(s) the rest of the time. On features, (particularly during the first few months the RedDrives are in the field) I would have a crew position for data management. Back in the day when I used to 2nd AC I basically loaded film mags, slated and did camera reports all day - is this really that different?
My workflow relies on a data management person who works out of a nearby van or trailer with a rack of hard drives. Making lots of back ups, reviewing footage for problems, running the circle takes through RedCine for on set dailies, even doing some rough editing.
Open your mind. The new paradigm is not just dodging the costs of film stock and lab work, it is about using the digital tools to short circuit the laborious and time intensive process film requires. The Director, DoP, scripty, producers, etc can now step into the van and see what they have while they can still do something about it. Check the density, continuity, eyelines, etc. Is there a huge potential for abuse and actually slowing down the day? Obviously. This working style requires a set of ground rules set up by the producers and the director during pre-pro, no doubt. If done right though it should cut post costs, reduce the need for pick ups, etc. more than enough to make up for the cost of one more person on the production crew to manage the data. To bring this discussion full circle - by having a dedicated crew member managing the RedDrives you should be able to limit the repercussions of lost data to acceptable levels. (ever had someone open the door of the dark room while you were unloading mags or had film jam in the gate? *hit happens.)
Blair S. Paulsen
RedOne #19
Flash for safety?
At the risk of sounding like a Panny rep pimping P2 there is just nothing like the warm fuzzy feeling of solid state storage on board. Sure it costs a fortune but you can reuse it thousands of times. Who's ready to spend $5K for a 64GB RedFlash (this is a guesstimate only) that would hold +/- 30 minutes of 24p/4k RedCode? Unless you said yes to the preceding question then the RedDrive or a tether is all we've got for the foreseeable future. So...
If your project is going to have the camera getting jostled on a regular basis rent a bunch of RedFlash modules. If your project is going to have the camera getting knocked about only occasionally then get one RedFlash for those shots and use the RedDrive(s) the rest of the time. On features, (particularly during the first few months the RedDrives are in the field) I would have a crew position for data management. Back in the day when I used to 2nd AC I basically loaded film mags, slated and did camera reports all day - is this really that different?
My workflow relies on a data management person who works out of a nearby van or trailer with a rack of hard drives. Making lots of back ups, reviewing footage for problems, running the circle takes through RedCine for on set dailies, even doing some rough editing.
Open your mind. The new paradigm is not just dodging the costs of film stock and lab work, it is about using the digital tools to short circuit the laborious and time intensive process film requires. The Director, DoP, scripty, producers, etc can now step into the van and see what they have while they can still do something about it. Check the density, continuity, eyelines, etc. Is there a huge potential for abuse and actually slowing down the day? Obviously. This working style requires a set of ground rules set up by the producers and the director during pre-pro, no doubt. If done right though it should cut post costs, reduce the need for pick ups, etc. more than enough to make up for the cost of one more person on the production crew to manage the data. To bring this discussion full circle - by having a dedicated crew member managing the RedDrives you should be able to limit the repercussions of lost data to acceptable levels. (ever had someone open the door of the dark room while you were unloading mags or had film jam in the gate? *hit happens.)
Blair S. Paulsen
RedOne #19
im.thatoneguy said:
It is enough to add a single third parity drive to the two working drives for redundant operation. This is similar to RAID 5, but this cannot rebuild itself or anything complicated like this. So the electronics to calculate the parity is nothing, it can be put into the RED body as well. If one drive fails, there is no hot swappable drive or anything, so you may want to continue shooting with a fresh RED RAID so that you do not risk loosing another drive. And the software that receives the data from the RED RAID does the reconstruction of the data from the two intact drives. The Kinetta planned to use such a thing I think.
In this way you do not need 4 drives for RAID 0-1 and you do not need a complicated electronics for RAID 5.
im.thatoneguy
Well-known member
baro said:
... are these systems available? I would love to have one for my PC. RAID 0 speed with RAID 5 protection for <$200 Sign me up.
Speaking of HD tech, there was a interesting article from Cringeley back in October describing a fascinating new hard drive tech coming out next year:
http://www.pbs.org/cringely/pulpit/2006/pulpit_20061026_001143.html
robust, cheap, fast access - the RED of hard disks?
http://www.pbs.org/cringely/pulpit/2006/pulpit_20061026_001143.html
robust, cheap, fast access - the RED of hard disks?
filmmaker1977
Carbonite Member
the weight is actually my concern too..Blue said:
Nik Manning
Well-known member
evinsky said:
I think you got it! Nothing left to say.
bastards... :beer:
you guys should make it clear that 60p at "all" costs is what you want (I would)
32min. of 4k for hand-held/steady making the package lighter, and two hard to fill Red Raids for dolly and tripod sounds like non-stop shooting with the best of both worlds for almost eveyone's needs, just with a little on the fly organizing of the shot schedule in your head. If you are shooting all steady/handheld then you might need two solids, but thats only like going back to P2... anyway, I shouldn't be on this thread, just makes me want to break into your houses, lol :beer:
you guys should make it clear that 60p at "all" costs is what you want (I would)
32min. of 4k for hand-held/steady making the package lighter, and two hard to fill Red Raids for dolly and tripod sounds like non-stop shooting with the best of both worlds for almost eveyone's needs, just with a little on the fly organizing of the shot schedule in your head. If you are shooting all steady/handheld then you might need two solids, but thats only like going back to P2... anyway, I shouldn't be on this thread, just makes me want to break into your houses, lol :beer:
Last edited:
mike the beginner
Well-known member
What about the balance of the camera when using s35mm red zoom lens and just using red flash. Does anyone think that will be a concern? The weight saving might be say1lb - 1.5lbs max guessing. Red zoom lens below 7lb say matt box? Will it be front heavy?Balanced on your shoulder? I dont know so maybe someone could give an opinion with the new design of red rail. Has the balance improved?
The idea behind the red drive in raid form sound good, the only downside i see is that you are twice as likely one of the drives fail. If the drives are in a sealed unit there goes $1,000. I must say though i have been using computers for ten years now and have yet to suffer a hard drive failure. I suppose once you start doing video editing perhaps the situation changes?
Will red give us a similar good deal with the red flash..i hope so.
Michael
The idea behind the red drive in raid form sound good, the only downside i see is that you are twice as likely one of the drives fail. If the drives are in a sealed unit there goes $1,000. I must say though i have been using computers for ten years now and have yet to suffer a hard drive failure. I suppose once you start doing video editing perhaps the situation changes?
Will red give us a similar good deal with the red flash..i hope so.
Michael
evinsky
Well-known member
Y.Goh said:
If anyone can make a pro-caliber EVF it's Red, but the limits of existing technology tell us that even with a 700:1 contrast ratio an Apple 30" LCD can't show the whole 11+ stops of recorded latitude Red can achieve. So be realistic in your expectations and they probably will be exceeded.