Film-Tech Forum ARCHIVE: Cinema server

This is topic Cinema server in forum Digital Cinema Forum at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=16;t=003797

Posted by Justin Bacon (Member # 10638) on 09-15-2019, 02:32 PM:

What's the advantage of using a stand alone server on the rack vs one built in the projector it's self?

Posted by Frank Cox (Member # 6258) on 09-15-2019, 03:37 PM:

Having just changed from a stand-alone to one that goes into the server, I can see three advantages to the stand-alone:

1. You don't have to have the projector fully powered up to ingest a movie.

2. The CRU drive slot is built into the stand-alone server so it's a lot easier to insert and remove the drives. The "built-in" just has a USB-connected boot to insert the drive into. It's a more flimsy arrangement so you have to hold the boot down with one hand while inserting the drive with the other hand, and the boot fits more snugly so you have to peel labels off of more of the distribution drives to get them to slide into the boot.

3. Doing stuff with the new server is slower. You have to wait a lot longer when ingesting trailers, for example. Everything just drags along. It takes a whole minute or two for the machine to get ready to start ingesting when you insert a trailer drive, and another minute between each trailer, and another minute or two afterward before you can remove the drive. Ingesting features isn't quite so draggy, but it's still slower overall.

When it comes to actually playing a movie, it's pretty much the same thing either way, though. And I suppose it's nice to have everything together. The control screens are all too long vertically for the height of a web browser window on a laptop, though, so you do have to do a lot of scrolling up and down that really shouldn't be necessary.

My server is a GDC SR1000 and my old server was also a GDC, so others might be different.

Posted by Mike Blakesley (Member # 26) on 09-15-2019, 04:22 PM:

quote: Frank Cox
The "built-in" just has a USB-connected boot to insert the drive into. It's a more flimsy arrangement so you have to hold the boot down with one hand while inserting the drive with the other hand, and the boot fits more snugly so you have to peel labels off of more of the distribution drives to get them to slide into the boot.

There is a "desktop" version of the drive bay. It's heavier duty, and I have it tucked into the space between the projector base cabinet and the lamphouse. It's connected by a cable to the server unit and has a power switch on the backside. You still get the slower operations, but the drive bay is a lot easier to handle, and it just looks and feels better overall. I think it cost about $80. Maybe Mark G. might chime in here with a link to it if he's paying attention to this thread.

To answer the original question.... I don't think there are any advantages to the in-projector media block, except it is a lot easier to change out bad hard drives since (at least on ours), they are in an external unit.

Posted by Steve Guttag (Member # 268) on 09-15-2019, 04:27 PM:

Note, the speed of ingest wasn't a function of standalone or not, that was how the various companies allocated resources and what processing/data paths they had to move content.

The Dolby DSS servers could pretty much run full speed ingest/transfer while shows were running but they were pretty unique in that capability.

The IMS servers seem to be increasing their ingest speed with each revision but nothing like the DSS servers.

For GDC, you really have different things at play going between a "box" server like the SX2001 to the SX3000/4000 and now the SR1000 that will affect speed as they are completely different from each other in most respects other than the UI of the SX2001 and the SX3000/4000 were essentially the same. The SR1000 is a complete departure as they transition to web-based UI.

To add to Frank's list. With box servers, you don't have to shut down the server before the projector, requiring some form of user interaction with the server UI or automation.

Now, in terms of processing power for showing the movie, the IMB/IMS systems are continuing to increase. You can see this in what SERVERS are capable of running the new High Frame Rate stuff (setting aside a license, the ability to handle it). The latest servers can generally handle 60-120fps whereas the previous generations were either a solid NO or were dicey with it (could handle specify data throughputs but would choke on higher ones).

HDMI 2.0 is typically supported on the latest servers (ICMP, IMS3000, SR1000) but only HDMI 1.4 was supported on previous efforts.

Mike, I'd say it's a wash on changing out hard drives. You can have either of them out in seconds and spend most of your time taking the drive out of the special carrier and putting the new drive into the carrier rather than the physical act of pulling/inserting the drive.

As for CRU, they make rackmount versions (most expensive), desktop versions (still expensive but VERY solid and MUCH higher(several times) than $80 and the basic "sleeve" that is in the under-$100 range but it is something that can get damaged if allowed to drop. By and large, with the new servers it is all USB3.0 for local ingest.

Posted by Mike Blakesley (Member # 26) on 09-15-2019, 08:27 PM:

I'm sure you're right on all counts, it's been about 4 or 5 years since we got that desktop drive bay...but whatever it cost, it was worth it in the convenience vs. clunkiness of the "boot" option.

Posted by Steve Guttag (Member # 268) on 09-16-2019, 06:56 AM:

List price on the table top version RTX115DC-3Q is $339.00. The rackmount version start at a list price of $489 for a single bay eSATA version and go up from there. However, the "Move Dock" aka "sleeve" remains low cost. $52.50 list for the USB 3.0 version and $74 list for the version with both USB3.0 and eSATA.

Mike is right about the table top version though. It is substantial in weight.

Posted by Scott Norwood (Member # 30) on 09-16-2019, 07:03 AM:

Interesting. I've never heard of the RTX115DC-3Q before now. It seems like a good idea, though. It does seem slightly expensive for what it is; I have the internal receiving frame in one of my desktop computers for making/copying DCPs, and that went for about $100 a few years ago.

Posted by Steve Guttag (Member # 268) on 09-16-2019, 07:11 AM:

Talk to ANYONE that has to have custom metal work done for things like frames. You'll easily spend more on that than the product it houses. Check out the JNIOR thread and Bruce can tell you all about it. A large portion of the cost of making a rackmount version is the metal case in the quantities they would be talking about. How many table-top frames do you think they would sell each year?

The Sleeves are relatively low-cost metal and without an esthetic concern. It just has to give the parts a place to attach to. Likewise for a slot for a computer. I wonder how much of what they use was designed/made by others and they put their face on it and the bits and pieces. Personally, I don't know how they make the Move Docks so cheap.

Posted by Mark Gulbrandsen (Member # 72) on 09-16-2019, 07:26 AM:

quote: Justin Bacon
What's the advantage of using a stand alone server on the rack vs one built in the projector it's self?

None for the most part and the industry is moving away from seperate rack mounted servers. Full size servers also draw A LOT more power... 300 watts as compared to 50 watts for an IMB/IMS server. The plug in can save 4000 watts over full sized units on a big plex. And with the environment heading the way it is, saving power these days may seem trivial but it is is important and something that may be regulated in the not too distant future.

Mark

Posted by Harold Hallikainen (Member # 5405) on 09-16-2019, 12:35 PM:

As pointed out above, power available to run a server in the projector is pretty limited (I think there's a maximum of about 75 watts available). This may create a processor speed limitation. It also encourages the use of solid state drives since they take a lot less power. But, processor speed continues to increase at the same power consumption level. I think a big thing is the bandwidth available between the server and the projector. With an internal server or IMB (as opposed to dual HDSDI), a lot more bandwidth is available to handle higher resolution, higher frame rates, and stereoscopic images. In general, one box costs less than two, so integrating everything into a server that goes into the projector is more cost effective. I THINK USL was the first to do that by adding a server board to the IMB. The CMS-5000 now uses a single multicore processor to handle the server functions, IMB functions, network audio (QLAN), etc. All that and consuming less than 75 watts.

Harold

Posted by Leo Enticknap (Member # 534) on 09-16-2019, 03:41 PM:

One significant disadvantage is local storage capacity. Most IMSes (servers that go into the side of the projector) have an onboard RAID of three 2.5" drives. Barco now approves a model of 4TB drive for use in their ICMP, which gives you a maximum content capacity of a little under 8TB in a RAID5. As far as I'm aware, the other manufacturers approve 2TB drives as the biggest, meaning 5-6TB maximum capacity (though Dolby only approves an insanely expensive model of SSD in the 2TB size).

For a standalone, rack mount server, however, you will likely have more drive bays and larger capacity drives at lower cost to choose from. As an extreme example, there is an (admittedly unofficial and unauthorized) modification to the DSS200 you can make (swapping out the 4-drive RAID controller with the 8-drive variant of the same model), that will allow you to use up to eight drives and 16TB of onboard RAID storage.

So for a venue that does lots of festivals, rental work, or any other usage pattern in which you rarely play the same movie twice and/or need to ingest stuff a long time before the show and leave it ingested, an old school server could still make sense.

Posted by Steve Guttag (Member # 268) on 09-16-2019, 08:40 PM:

The IMS3000 now has a dedicated NAS option for just under 11TB of storage. With it also returns the option of transferring content while the server is off, providing that your ftp source knows how to tell when the IMS3000 is off so use the NAS dedicated NIC for it.

Harold, did the CMS predate the IMS1000? I thought Doremi was first there.

Posted by Harold Hallikainen (Member # 5405) on 09-16-2019, 09:34 PM:

Oops! It looks like the IMS-1000 was indeed out before the CMS-2200.

Doremi IMS-1000 on 5/4/13 - https://web.archive.org/web/20130504065233/http://www.doremilabs.com:80/products/cinema-products

USL CMS-2200 on 7/1/14 - https://web.archive.org/web/20140701060330/http://www.uslinc.com/

Good memory, Steve!

Harold

Posted by Dave Macaulay (Member # 813) on 09-17-2019, 01:28 AM:

A chassis server takes more power, yes. The high capacity NAS box for a GDC SX3000 server takes a fair amount of power as well, I think.
There are advantages for a chassis server.
A Doremi SV4 IMB system has a DVD/CD drive (data only, these will not play "normal" DVD or CD content) and a SATA CRU ingest bay. It can also use 4x4TB drives for a RAID capacity of roughly 12TB. This is pretty large, and you have the possibility of swapping in a second preloaded RAID set in roughly 7 minutes (drive swap time plus chassis shutdown and restart times, the projector and lamp can stay on). For something like a film festival with many features, content can be preloaded on a second RAID set during off hours or in an offsite SV4 (or DC4).
The 3'5" drives in these are generally more durable than 2.5" drives typical in internal servers.
3.5" disk drives are usually available with larger capacity than 2.5" drives (although there seems to be a limit in drive capacity: 8TB drives are available but I don't think any server has larger than 4TB approved drives).
SSDs are taking over though, with low power draw and high capacity regardless of "legacy" drive size - and with new packaging that is often much smaller than those.
I expect to see more SSD based internal servers, and probably approved SSD drives for older internal and chassis servers.

For 99% of installations, an internal server will be the logical and economical choice.

Posted by Marcel Birgelen (Member # 6801) on 09-18-2019, 03:56 AM:

The move from stand alone servers to the recent "integrated media servers" was more or less a two-step process.

There was a need for more bandwidth and better security between the media block and the projector/imaging part, primarily for stuff like HFR, so the first step was to move the media block into the projector. Ironically, HFR is currently possibly one of the most underused features of DCI.

The second step was to miniaturize the server, which was already based on standard PC components, into something that fits into the slot designated for the media block.

The trend of moving playback hardware into the display hardware is something that's steadily progressing all over the place, even in your living room. You used to have all kinds of playback devices like VCRs, VideoDisc, DVD, BluRay players, Digital Set-Top-Box, those are all progressively being replaced by the streaming apps running directly on your SmartTV...

Personally, I do like a bit of separation in functionality. So, I prefer having a dedicated playback device like a "server" and a projector that doesn't do anything else than displaying what it's being fed via a high-speed digital interlink, although the future will clearly be, hooking up your IMS via 10GE fiber to some central NAS.

Posted by Mark Gulbrandsen (Member # 72) on 09-18-2019, 08:14 AM:

quote: Marcel Birgelen
There was a need for more bandwidth and better security between the media block and the projector/imaging part

Bandwidth yes! But the whole security thing is still a bunch of hogwash. As long as the content is encrypted no one is going to be able to steal it.

Mark

Posted by Marcel Birgelen (Member # 6801) on 09-18-2019, 10:38 AM:

Well, it depends, because it's far easier to put some kind of capturing device on an SDI link than to capture it from an internal bus.

While the first also carries close to zero risk of inflicting any serious harm if things go wrong, the other one will potentially destroy a pretty expensive part of hardware.

Posted by Gordon McLeod (Member # 33) on 09-18-2019, 11:17 AM:

Taping the SDI would be no value as the signal is encrypted which is why it requires the enigma board

Posted by Marcel Birgelen (Member # 6801) on 09-18-2019, 11:24 AM:

Well, yes, it is encrypted, but you do have the signal. Point is: It's harder to get to the signal on an integrated bus. Hollywood even doesn't want you have the encrypted signal, they're that paranoid.

Also, encryption schemes have been broken in the past, especially due to weaknesses in the key exchange or key storage mechanisms. That's also how stuff like Blu-Ray got hacked or how HDCP can be circumvented.

Posted by Mark Gulbrandsen (Member # 72) on 09-19-2019, 07:54 AM:

No one has yet to decrypt a stolen signal... for that matter has anyone even stolen a signal? There is no reason to at FIPS level 140-2 and hackers even know that. And DCI has already moved that up to FIPS 140-3.

Hollywood is paranoid and ALMOST ALL OF their movies which are leaked out are stolen through people getting screener Blu-Rays from industry people that worked on those films, and those members viewing the films voting for awards. Sheesh! The way it's headed, pretty soon, we will all have to load in a new KDM every day just to run the content.

Mark

Posted by Steve Guttag (Member # 268) on 09-19-2019, 10:02 AM:

They encrypt content that is already in home release. What is the point. The image and sound are out there then.

Posted by Mark Gulbrandsen (Member # 72) on 09-19-2019, 12:49 PM:

Exactky Steve!