[Neil Sadwelkar]

The application my client has in mind is to have a large off-site RAID storage to backup camera files and other exported media to be available online at the new location. About 50 TB is their projection of what they would need in about a year’s worth of shoot. So, knowing that these estimates are often off, I’ve doubled it. They want to be able to plug in their MacBook Pro and copy off drives on to this big RAID. Sharing is not essential hence a NAS kind of setup is not being considered.

So its really a choice between a Thunderbolt storage or a SAS storage. For SAS they would need to provide a Thunderbolt-PCIe box with a SAS card inside it anyway, hence Thunderbolt to reduce the complexity. LTO was considered but the backup data needs to be ‘live’ and they want to avoid having to read off tapes.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

So it looks like the rack mount ones from Netstor or Maxx are the only viable ones.
Any idea if the Netstor NA333 16-bay will work with 8TB drives? I couldn’t find any documentation on this at their site.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

Bru, seems to write some error correction and redundancy information along with every file. So, when one is backing up a large number of small files, Bru is slower compared to a small number of large files.
The extreme comparison is when we backup 2 TB of Arriraw (which is file sequences, 7 MB per file) compared to 2 TB of Red raw.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

Just to ensure that the tape drive and everything else in the system is functioning normally could you try doing a backup with something else? Like Bru-PE for instance. Its free for 30 days, and if a copy goes through without any error you can rule out your drive or system being the problem.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

I’ve seen a few old Avid Unity setups where the editors are simply not using the Avid bin and project sharing facilities that only unity provides. Instead there are different editors working on different projects seldom sharing any sequences. If that is your workflow, then a Unity/ISIS is overkill. Any shared storage will work fine.

ISIS or Nexis, or some other Avid capable solution will always be costlier than a standard shared storage. At the moment a Nexis Pro with 16 TB shared storage runs £ 10,000 (US$ 14,000). Plus about £ 1000 ($1400) for the switch. If you’re in a country with import duties, these will run up the price by another 25-30% in some countries.

So, examine the editing workflow, check out the extent of sequence sharing. If all 8 can work off separate partitions with the occasional copying of media from one to the other, then just a shared storage will do fine. And for the same price as a Nexis you’ll get two to three times the disk storage. Alternately, you could even do a ‘hybrid’ with critical shared media off a Nexis Pro with only 32 TB storage, and the remaining 170 TB of standard shared storage partitioned one per user with only unshared media on it.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

Here’s a transcript of my system about AppNap

Last login: Fri May 13 10:57:46 on console
xxxxxxxx:~ xxxx$ defaults read NSGlobalDomain NSAppSleepDisabled
2016-05-13 20:56:58.371 defaults[62761:409156]
The domain/default pair of (kCFPreferencesAnyApplication, NSAppSleepDisabled) does not exist
xxxxxxxx:~ xxxx$ cd /
xxxxxxxx:~ xxxx$ defaults read NSGlobalDomain NSAppSleepDisabled
2016-05-13 20:57:24.917 defaults[62807:409528]
The domain/default pair of (kCFPreferencesAnyApplication, NSAppSleepDisabled) does not exist
xxxxxxxx:~ xxxx$ defaults write NSGlobalDomain NSAppSleepDisabled -bool YES
xxxxxxxx:~ xxxx$ defaults read NSGlobalDomain NSAppSleepDisabled
1
xxxxxxxx:~ xxxx$

And my Energy Saver settings…

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

Between USB3.0 and Thunderbolt, 4-bay RAIDs you’re likely to see better speeds with a Thunderbolt RAID.

I haven’t done a side by side comparison with the same box, but from observation, and running Blackmagic Speed test on new RAIDs, I’ve found the G-Tech 4-bay, and Promise 4-bay deliver consistently 400 MB/sec plus read and write speeds. Whereas USB3.0 G-Tech and Stardom RAIDs over USB 3.0 rarely go over 240 MB/sec.

Cost of Thunderbolt RAID boxes are likely to be higher.

While on the subject of speed, I just bought a Samsung Evo 1 TB SSD. In a small USB3.0 enclosure this little thing delivers consistently over 400 MB/sec speeds. 1 TB is not great capacity, but its enough as a ‘scratch’ drive for PrPro or FCP X. Or even a Cache disk for Resolve.

Another observation I have is that is some enclosures, over USB 3.0 the speed difference between RAID0 and RAID5 is only marginal, so maybe you should take a look at the safety of RAID5.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

On the exact same system, I rolled back to Bru-PE, and backed up 1.74 TB off a 2 TB LaCie Rugged drive over USB 3.0. Composition of files was similar. Large number of file sequences from an Arri Alexa camera and many DPX file sequences of graded material. over 170,000 files.

This time the backup took under 6 hrs and verify, another 3 hrs or so. Well within previously observed time frames.

I think Bru-PE 3.1.18 is doing something that drastically slows down backup and verify timings.

Have started another LTO and will do one more after that. Let’s see if this is repeatable.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

I’ve used IBM, Tandberg and HP LTO drives, not Quantum. But I don’t think there’s any difference between drives, they all do the same thing.

Internally, an LTOP drive is usually SAS. If there’s a Thunderbolt interface to the case, its probably doing a Thunderbolt to SAS conversion internally. So there too you won’t see a difference in speed or performance.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Neil Sadwelkar]

What kind of data are you going to back up?

Is it a huge number of small files like in the case of ArriRaw/DPX/OpenEXR which are file sequences. Or a smaller number of large files like in the case of Red Raw,/Quicktime ProRes/Sony F65 Raw/Canon C300 MXF?

I’ve used both PreRoll Post and Bru-PE. These are two different things and you don’t necessarily need both. Both have their pros and cons.

And then there’s Yoyotta ID LTFS which is an offload cum LTO backup software.

The cost of these three is comparable.
PreRoll Post is $499, or $399 if bought with an mLogic LTO drive
Bru-PE is $499 and there’s a 30-day full trial so you can figure if it works for you
YoYotta ID LTFS is $514 after conversion from UK pounds
Yoyotta has a demo too.

So my suggestion is to get an LTO drive, gather about 2 TB of footage, back it up and check for yourself which is easiest to use and works stably.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India