FW800 RAID or Internal HHD's for DV Editing?

[D*I*S_Frontman]

I have a DP 1.25 G4 FW800 system w/2GB RAM, Superdrive, and the original 120GB HD. I want to get set up for multiple editing projects and I would like input on the best strategies for HD configurations.

I have two additional FW800 cards yielding a system total of three FW800 buses installed, so I could use external FW800 drives in a RAID 0 config, each with its own dedicated bus. That is one idea I like, because I could swap out drives and edit different projects back and forth with no backup/reloading issues (maybe using a Wiebetech BayDock 800 dual system with removable drives, each drive on its own separate bus). I could also use the raw HHDs themselves as my archival storage medium--revisiting a project for a client as easy as slapping the caddied drives back into the FW800 housing.

The other nice thing about this idea is that with a FW800 CardBus adapter I could use the same drives on my Pismo in a pinch (Yes, it would be a slower RAID daisied on one bus, but see Barefeats.com regarding the speed of FW800 drives daisy chained through the Lacie CardBus unit--very respectable). Oh, and before I were to do that I would bring the Pismo up to 900mhz via the Powerlogix upgrade (already maxed out the RAM @ 1GB).

The downside to this idea is that it is a LOT more expensive than just jamming my tower with drives.

If I decide to just install three more drives in my tower, just how hard is that? Is it something I could do easily enough if I were swapping out drives for different projects? I am no hardware genius, but I am not afraid to do it myself if I have some good instruction. If anyone has install tips or animation links, please post them here for me. When asking this in the past people have told me that it is a hassle to swap out drives--not something you want to do on a frequent basis.

For speed's sake, if I went the tower only route, should I get three larger drives (say 200GB each) and stripe them all @ RAID 0 via OSX for maximum throughput?

Discuss. And thanks for you insights.

 

[jamall]

If you go with the internal option you will need to get an ATA/133 PCI card, as your tower can only handle a total of 4 ATA devices (including the optical drive), so bear this cost in mind. Xlr8yourmac.com and barefeats.com have recent articles that may be of interest. It's a LOT easier to unplug a firewire device than swap drives in and out of a tower, so the extra expense may be worth it. Hope this helps.

 

[D*I*S_Frontman]

Much easier. Yeah, that's what I thought. I am just trying to discern if it is worth the cash for that ease.

 

[LethalWolfe]

FW800 is uber-overkill for DV. RAID-0 is uber-overkill for DV. Internal drives or any quality FW400 is more than fine for DV. I also strongly advise you *not* to use HDD to archive video. A HDD crash/corrupted file is much more likely than tape becoming unusable. Buy some archive/master quality tapes and dumpt your finished projects onto that. Plus, depending on what you do and how much you do, HDDs can start filling up pretty fast.


Lethal

 

[D*I*S_Frontman]

Overkill? Not if you are running multiple video streams it isn't. If you want to be sure not to drop frames, a bit of overkill is good insurance.

It's not just about archiving the final output--tape would be fine for that. I am talking about archiving EVERYTHING associated with a given project in exactly the form you left it in the day you rendered the final version. No recapturing media from the raw footage tapes and reattaching it.

Drives are getting so cheap that it is a viable back-up media cost-wise. It seems to me that a drive or a set of drives used for only one major project would have a miniscule failure rate if they were properly stored after project completion and never used again. This would have been absolute lunacy a few years ago, but when a good 120GB HHD is around $120 (and only getting cheaper all the time) it begins to make sense. I could back the cost into my final invoice for clients.

Immediate turn-around if someone wants to tweak a program a few years later (changing the credits, etc) is a productivity boon. Imagine having to find all of the old source tapes, recapturing all of the used footage and making sure it is all properly reconnected--then making you minor changes. Several hours worth of work vs. 30 seconds to slap in the old drives and return to the exact moment you finished the project last.

 

[ClimbingTheLog]

Originally posted by LethalWolfe
A HDD crash/corrupted file is much more likely than tape becoming unusable. Buy some archive/master quality tapes and dumpt your finished projects onto that

It's all about redunancy - you can have a tape go bad just as likely as a hard drive. Actually, a tape on shelf will theoretically degrade faster than a hard drive platter.

So, if it's really important, make two tapes and take one offsite. Or setup a RAID-1 or 10 or 01 mirror or do a daily backup with a hotswap unit and take it offsite.

 

[LethalWolfe]

Originally posted by D*I*S_Frontman
Overkill? Not if you are running multiple video streams it isn't. If you want to be sure not to drop frames, a bit of overkill is good insurance.

It's not just about archiving the final output--tape would be fine for that. I am talking about archiving EVERYTHING associated with a given project in exactly the form you left it in the day you rendered the final version. No recapturing media from the raw footage tapes and reattaching it.

Drives are getting so cheap that it is a viable back-up media cost-wise. It seems to me that a drive or a set of drives used for only one major project would have a miniscule failure rate if they were properly stored after project completion and never used again. This would have been absolute lunacy a few years ago, but when a good 120GB HHD is around $120 (and only getting cheaper all the time) it begins to make sense. I could back the cost into my final invoice for clients.

Immediate turn-around if someone wants to tweak a program a few years later (changing the credits, etc) is a productivity boon. Imagine having to find all of the old source tapes, recapturing all of the used footage and making sure it is all properly reconnected--then making you minor changes. Several hours worth of work vs. 30 seconds to slap in the old drives and return to the exact moment you finished the project last.

Still talking about DV, by "multiple" do you mean a few streams or do you mean a few dozen streams? Unless you are having technical difficulties (or have crappy drives) you will have to render way before you will be dropping frames due to the HDD not being able to keep up. I'm sure there is a limit to what the drives can handle it would probably take a project specifically designed to bring the drive(s) to it's knees because I've never heard of anyone needing anything other than 7200RPM IDE drives to edit DV. The choke point in external FW drives are the harddrives themselves. FW800 will get you more bandwidth but the 7200RPM ATA drive sitting in the enclosure isn't fast enough to use it. Heck, it's not even fast enough to max out FW400.

Keeping your projects permantly on-line only makes sense, IMO, if you don't deal w/a lot of media and/or you will be making relatively frequent changes to the project. If you only had an hour or 2 worth of media than that wouldn't be bad. But if you have 5 or 8 or 12 hours of media then you are looking at only putting one or two projects per drive. Heck, to pick up some extra $$$ I've cut some weddings for a local guy and he usually gives me 4 tapes of footage. I've done 4 weddings for him and it's about 52 gigs per wedding, so over 200gigs total (not including the project file, render files, gfx, etc.,). So just four projects is going eat up an entire 250gig drive and cost well over $200 for the HDD+enclosure. You could spend, lets say, $260 for a HDD+FW400 enclosure to keep the whole project on-line or you could spend around $50 on master quality DV tapes ($30 for non-master quality tapes).

You could cut down your excess media by using FCP's "delete unused media" function, but media management is still a weak point in FCP so this function doesn't always work as it supposed to. And it is a royal PITA to repair if it screws things up. Most people tend to avoid it if at all possible.

Like I said, under certain circumstances I could see keeping a project on-line all the time, but keeping everything on-line isn't a very viable solution yet, IMO. Tapeless aquisition<sp?> and storage is definetly coming down the pipe but it is going to be a while before it replaces tape (at least on a large scale). I agree that the chance of failure on a HDD used for just one project and then shelved would probably be closed to zero, but everytime I looked at my shelf-o-HDDs all I would see is wasted $$$.

I wrote this post thought the day so I apologize if it doesn't flow together very well.


Lethal

 

[D*I*S_Frontman]

Well, perhaps i could leave two or three bigger projects per stored drive set. Thanks for the advice.

Any other insights from the MacRumors faithful?