Andrew Richards
Forum Replies Created
-
[Craig Alan] “If a student pulls the plug on a thunderbolt drive or it looses power will sparse images survive?”
Probably the same riskiness as removing any storage without unmounting. Not a guaranteed failure, but it can damage the filesystem if it happens while the storage is busy writing. But at least with Thunderbolt RAIDs there is little reason to be unplugging them in the first place. That’s another +1 for the RAIDs.
[Craig Alan] “Students did loose unsaved work. But that should be a lesson learned. “
Won’t happen with FPCX- it saves every action as it happens automatically. There is no unsaved work (assuming it is functioning normally).
[Craig Alan] “Are large raids more prone to failure?”
No, if anything they are better vetted than a generic FW drive. I’m not positive, but I imagine the Pegasus uses “enterprise” SATA drives that are also better vetted and typically more reliable.
[Craig Alan] “I could opt for less storage – raid 10 is supported on these. I assume that’s about half the total storage which might be fine for now. I could also (I think) partition each of the four drives into two. That would give me eight partitions, enough for 8 classes. It would no longer be a raid, but as you point out, fast enough for most projects. “
If you want protection, use RAID-5. That will yield a bit under 6 TB for you. If a drive dies, replacing it will allow the RAID to repair itself. Two dead drives is a dead RAID-5. You can partition the RAID set into as many as 16 partitions (max supported by GUID table). Remember- you are not addressing specific physical drives with a RAID, the OS sees just one logical storage entity. Partitioning is a logical division on top of that. No loss of speed since you are still using all the hardware even when partitioned.
But if you like the sparse image method, there’s no reason to partition the RAID. Just set read-only preferences on the root and it won’t even show up in FCPX. Only the mounted sparse image they have permission to write to will mount and be addressed by FCPX.
[Craig Alan] “I want to be able to do all this logging in and out each period from the ARD computer – possible? “
Oh yes. You can log out all the Macs in the lab in one stroke if you like. Or reboot them. Or shut them down. Or sleep them. And that just scratches the surface of all ARD can do. It is very useful. Logging in is something that has to happen by hand since it requires authentication.
[Craig Alan] “It really needs to be just the server not used for other tasks like attendance and printing worksheets and so forth?”
You can do userland tasks on a server, just be sure it locks its screen with a password and doesn’t get turned off or put to sleep. If the server needs its resources for its work, then you want dedicated hardware. But for what you are doing it won’t be breaking a sweat.
Best,
Andy -
Carey nails it.
Best,
Andy -
If only I were going….
Best,
Andy -
[Oliver Peters] “Really? The check-out/check-in process makes it almost function like Unity and Interplay. I’ve been working with it a couple of years now as an editor.”
Really. Like pulling teeth. It didn’t help they tended to have really bloated bins that made check-in take a long time.
[Oliver Peters] “Custom bin entries show up as metadata in FC Server if you assign it that way. You have to search through the productions branch and not the assets branch.”
They wanted markers to map to Annotations (which can be done with some really crazy scripting and even then it requires a rather brittle series of hoops to be jumped through). What little access you have to FCP project metadata in FCSvr wasn’t condusive to their work.
[Oliver Peters] “However, I would agree that it’s not the best DAM when compared with other products.”
Not for editors, but it was the king of value for money and automation.
Best,
Andy -
[Andrew Kimery] “This was my reaction when I started testing it for possible deployment at the last place I worked. The integration I’d hoped for just wasn’t there which was very disappointing.”
And for the record, I loved FCSvr. But I’m not an editor. I made it do some pretty killer stuff with automating archiving and publishing to the web.
Best,
Andy -
[Richard Herd] “Thanks! I get about $5,000 per student @ 30 students.”
Do you anticipate the students sharing media across their projects? Like they all pool their shoots and edit from the pooled media? Or would this be demo media for teaching that they just all need access to?
Best,
Andy -
[Richard Herd] “I applied for and will probably get a grant for 5 iMacs i5s and 5 FCPXs. I was hoping to make all the media available to all the computers all the time. Is there a real way to do that, currently?”
Yes, but there is a catch. If you uncheck the box for copying files to the Events folder and do all ingest externally (not via FCPX), you can use any shared storage system that supports your codecs and the client load (aggregate bandwidth). See the SAN forum for that. Then when you import the files into the event, FCPX just links to the media in place a la FCP7. In this case, Projects and Events are still stored on direct attached storage (Movies folder or other local drive). If you wanted to get fancy, you could do home folder syncing with OS X Server and leave the Projects and Events in the user’s Movies folder (default location), but that could get unwieldy since render files are stored with Projects (another thing Apple really needs to fix). Then when they sit at any machine and log in against Open Directory on OS X Server, the contents of their home folder is synced to that machine and will show them the same thing they saw the last session on any other machine. A lot of render files would mean a lot of network traffic, hence the warning.
Xsan would let you put Events and Projects on shared storage via SAN Locations, but Xsan is expensive and complicated and not for the novice (though a lot less expensive and complicated than it used to be). Even for a workgroup as small as yours, you are still looking at minimum $25K just for equipment ($1,500 Qlogic 1400 switch, $850 SANlink Thunderbolt HBAs and cables for each client, twin $1,600 Mac mini Servers for MDCs, additional Gig-E switch for metadata traffic, fiber optic cables, and of course a qualified Fibre Channel RAID). The price goes up if you want more than a dozen usable terabytes or more switch ports for more clients. Then there’s the cost of a qualified installer’s time.
Either way, you’d be spending five digits to get some form of shared storage that can support editing.
Best,
Andy -
[Oliver Peters] “But it does take care of moving the media and relinking the files in FCP 7, so that there is a very cohesive environment.”
Cohesive as far as linking to media, but disjointed for metadata and feedback. The editors I worked with found the check-in/check-out process added a lot of friction to their workflow, and they lamented that the places they naturally dealt with metadata already (markers and in the bins) did not translate at all into FCSvr even if they did tolerate the distraction of check-in. For them, FCSvr was a burden they wound up avoiding whenever they could.
Where FCSvr did shine was in automating producer approvals and other tasks external to FCP. In this respect, it saved editors a lot of time and effort. They could and did embrace that. But as far as the core functionality of metadata and media management, there was (is) a gulf between FCP and FCSvr. CatDV and its “magic door” address that using Apple Events and XML, but FCSvr never did.
That’s why I see so much potential in FCPX as a true client. Just by using FCPX, the editor is creating gobs of valuable metadata that would be so much more valuable if it were available to more than one editor at a time. If FCPX were directly interacting with a central Events and Projects database, then the whole thing becomes frictionless. Elegant, even.
Pie in the sky, sure, but it looks damn tasty in my imagination.
Best,
Andy -
[Oliver Peters] “The stupid part is that FCP 7 with FC Server works (worked) EXACTLY that way.”
I disagree. FCP7 had no idea FCSvr existed. You had to check out your FCP project file from FCSvr and then open it in FCP7. FCP7 and FCSvr could be united atop shared storage and FCSvr could manipulate the project files it was handing out to allow for edit-in-place, but FCP7 never had any idea FCSvr was there doing that. FCP7 just know it had a project file and that file linked to such and such media on the SAN.
If FCP7 had been directly addressing FCSvr, you would be searching for clips on the server from within FCP7’s Browser pane. Not in a separate Java app that you then had to drag your clips into FCP7.
What FCPX’s CoreData store model for Events and Projects opens up is the possibility that you could have FCPX as the client proper, with no separate sidecar app. The keywording in FCPX would be hitting the server directly, not waiting for a save to a file that would then have to be uploaded and parsed by the server. Shared FCPX Events would be like having a live shared Browser window in FCP7, and shared Projects would be like having a live shared Timeline window in FCP7. This is utterly different from the way FCSvr worked with FCP7, and in my opinion, it would be much more powerful.
One of the biggest problems with FCSvr in my experience was that editors saw using it as a chore, they did as little as possible in order to avoid spending time in it, and they really just wanted to have the metadata they were creating in FCP (markers, bins, etc) to “just show up” in FCSvr. But since FCP7 didn’t talk to FCSvr at all, that could not be achieved with anything approaching a practical workflow.
It would be cool if Apple built such a thing, but it seems extremely unlikely that they would.
Best,
Andy -
[Walter Soyka] “Jeremy seems to have a lot of confidence that future updates to FCPXML will allow third parties to deal with this. I’d love to see that, and I think that third-party developers are offering some really innovating solutions (like VirtualMXF), but I also think that splitting DAM duties between FCPX and a third-party app reduces the power of having a DAM in an NLE in the first place.”
From where I sit, FCPX is begging to be the client in a client/server setup where the heavy lifting of storing and serving media and metadata is on the server-side and FCPX is only mounting what the editor is working on (while everything else is a quick search away). The way it sits now, FCPX has a glass ceiling for volume. Fine for small-scale stuff, but a busy shop would find itself working around it with Event Manager X, the sparse image workaround, or SAN Locations on Xsan.
I’ve noted before that OS X already has the frameworks to enable this. FCPX interacts with its Events and Projects via CoreData SQLite persistent stores. But what if it could also read and write to a centralized PostgreSQL database? Live sharing of Event and Project databases would enable some killer workflow efficiencies!
[Walter Soyka] “I hope that this is an area that Apple intends to develop themselves in the future, because this is one of those maddening close-but-not-quite features.”
YES.
Best,
Andy