EditShare vs Facilis TerraBlock

Be sure to check out small tree too. We're working on avid support now, but we can certainly do all the other pieces. We even have 10Gb to the desktop solutions.

Steve Modica
CTO, Small Tree Communications

Graham,
unless you have 10 Gig cards in your MAC computers, you will not be able to work with your Discreet/Autodesk/Flame on EditShare. Standard GigE is not fast enough for those large uncompressed files. EditShare offers 10Gig ethernet cards (as does Small Tree), but standard (less expensive) EditShare systems can't support these uncompressed files. Facilis can because they are using fibre, which requires a fibre card in each client computer. If you decide to use Small Tree (or EditShare with 10Gig ethernet ) you will need a 10Gig card in each client computer to achieve these faster speeds. This is not a matter of "which company is better", it's just that regular Gigabit ethernet is no where as fast as 10Gig ethernet. And 10Gig ethernet costs more money.

Bob Zelin

This migrated from a previous thread - it's not really a TB vs ES post, but I recently did some testing of our Editshare system that I thought I'd share with the COW...

---

I'm an editor and our technology manager at a small post facility in Portland, OR. We do mostly advertising and corporate vids for an international ad agency - everything from Cinema & Broadcast TV to web videos. We finish everything that's non-broadcast in-house - and most of our broadcast work is offline edited here, online in LA or a Flame across town.

We used to be all Avid/Unity (4 seats), then added FCP stations as we grew, and HD became the norm. In 2007, we ditched the Unity and purchased an EditShare system - mainly for it's cross platform capabilities - a Unity for FCP? With project sharing? Same server runs Avid & FCP? Hallelujah!

(I looked at TerraBlock in 2007, and if memory serves, they only offered Volume Level Sharing, and did not have the same functionality with Project sharing, and may also have not been Avid compatible - don't remember exactly, but ES was definitely the winner at the time.)

Honestly, I was a bit worried about the ethernet vs. fibre bandwidth. We demo'ed an ES system, saw demos at NAB, etc… With DNxHD and ProRes available, we made the call that we will never plan to share Uncompressed HD media. We have a "Finish Suite" with a CalDigit local storage if we need to do Uncompressed, but really, we never do - ProRes HQ has been fine for us. We will occasionally work in DPX format for in-house finishing, so it's nice to have that bandwidth available if necessary, but still - bandwidth on the SAN was always in the back of my mind as a potential issue.

Our system has 24 TB raw storage, 20TB usable with RAID 5, using 48 500GB drives, and 2 EditShare Chassis - (The "Storage" series, a master and a slave, acts as one). Each chassis also has 16 available drive slots, which we could populate if we needed to increase storage (and would also increase aggregate bandwidth).

We have a 10Gb Switch from the chassis, which breaks out into 24 Gig-E ports for the clients. We have 16 ports wired with Cat6 cable straight to the workstations - no extensions, hubs, or additional switches. Each line runs directly from the CPU to the switch. (Some of these lines are "extras", available, but not in regular use - but it can be quite helpful for an assistant to be able to plug in a laptop and connect while in the room with the editor. Producers can scroll through footage, etc…)

When using DAVE SMB as the connection method, our workstations have read / write times in the 60-70 MB/sec range. When using AFP, we were getting reads in the 90's - but there were some other issues with using AFP, and in EditShare ver 5.5, SMB is the only protocol allowed. I think I've read that v6 made AFP improvements, and is once again an option (for FCP users, anyway. Avid always needs SMB.)

Anyway - we have been growing like gang busters, and adding workstations left and right, and I'm always wondering when we'll hit a bottle neck on bandwidth, so I decided to run some tests and push it to find the limitations. (There is no per seat license from EditShare, but for Macs, you DO need a copy of the networking software DAVE, which is about $80 per seat - so adding workstations is cheap, but not free.)

I started with running all 12 of our FCP workstations, all looping a 5 minute 1080p23.98 ProRes HQ sequence and outputting via KONA cards at full HD - all 12 hitting the same media, so using only one of EditShare's 10 RAID arrays. (All systems were set to report dropped frames on playback.) No problem… so I added 4 more workstations (2 towers & 2 laptops), opening up the ProRes QTs in QT player and having them playback on Loop. No problem. No dropped frames in any FCP system, and smooth playback in QT player.

16 streams of 1080p24 ProRes HQ with 2 tracks of audio - no problem.

I took all of the FCP workstations and added another track of video, added another file (same format, but from a different RAID array), and cropped it 50% for split screen effect - no rendering - just realtime playback. Added 12 more streams. 16 + 12 = 28 streams - no dropped frames on any system. Honestly, I was pretty surprised at this point.

OK - let's add a 3rd stream to the FCPs. Added V3, dropped another PR HQ file in, set the opacity to 50%. At the 5th system, I finally started getting dropped frames - and FCP crashes, actually. This was kind of surprising - but maxing out the bandwidth caused "application crashed while using KonaVideoOutput" messages to pop up on a few systems. After restarting, and trying the loops again on various systems, I figured out that it was capable of about 32 - 35 streams of 1080p23.98 ProRes HQ Stereo - under these test conditions, remember.

Normally, we're all not hitting the same media on the same drives - we're spread across ten different 2 TB arrays, and we're also never all working in 1080p ProRes HQ - offline editing is usually done at ProRes LT, or various camera native codecs - XDCAM, DVCPro, etc… And we also have an additional 32 drives we could put in the systems, which would increase aggregate stream counts.

So - 32 streams! Better than I thought it would be - but what does that mean as far the limits?

Did we hit the bottleneck at the switch? Or the RAID arrays? Would ES software v6 increase that? Using AFP? A newer ES Server with faster hardware? More drives? etc… Honestly, I'm not sure - but I'm pretty happy with where we are, and much more comfortable with our system than I had previously been.

Certainly, if you need Uncompressed HD or DPX playback, then a gig-E system is probably not right for you. But is Facilis Terrablock the only other option?

I chatted with an ES tech today, and asked about options for increasing bandwidth if needed. Not to sound like a salesman, but… We have a basic "Storage" series system, but they also have an "Xstream" system, which is more easily expanded, and if you need more than 24 users, it is possible to have multiple switches and bridges, etc…, and they also have the options of running 10Gb-E to the workstations via fibre.

(I see Facilis has done some of the same things - with multi-user write, fibre & ethernet options, and even mention "project sharing" in their literature, but I have not seen how the system works lately, so I can only comment on our EditShare system.)

So, there it is - 32 streams of HQ, but your mileage may vary.

Anyone have a TerraBlock and love it? Tell us about it!

David Jahns
Joint Editorial
Portland, OR