Film-Tech Forum ARCHIVE: current preferred server (6/2015)

This is topic current preferred server (6/2015) 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=002326

Posted by Scott Norwood (Member # 30) on 06-12-2015, 09:19 PM:

So, what is everyone's preferred server now that the DSS200 is NLA?

The GDC has the fatal flaw of the non-redundant boot disk. The Doremi has a 3-drive RAID array implemented in software for everything, including the OS. I have not worked with the Sony system. What else is left?

Posted by Steve Guttag (Member # 268) on 06-12-2015, 11:20 PM:

Ahem...the DSS line isn't NLA yet...it is still available until the end of the month (just ordered some last week).

Doremi on their 2K4 and Show Vault CAN have a 4 drive RAID though still software implemented. Doremi dabbled in a hardware RAID on the IMS1000...it has had such issues that on the IMS2000, they went back to a software RAID.

I don't know anyone that is putting in the traditional GDC server (SX2000AR) anymore and any paranoid person about the OS drive can have a clone drive at the ready...what does a HDD cost nowadays? I believe the SX3000 (Which is what they sell most of now) is solid state on the OS and external RAID drive for content. They just upped their external RAID options too with an "Enterprise" storage unit which can be used with the SX3000 or as the Library system for an LMS.

There is USL's IMS that uses 4 SSD in a RAID too.

Posted by Brad Miller (Member # 2) on 06-12-2015, 11:20 PM:

GDC server...non-redundant OS and 3 drive software raid via standard 3.5" drives.

GDC IMS thing...I believe this is an actual hardware raid, but again only 3 drives and they are "laptop" drives, so that means even more expensive to replace.

Doremi DCP2K4/ShowVault...you can put 4 normal 3.5" drives in this model, but it's still a software raid. The OS is held on an SSD chip, but those chips have failed.

Doremi DCP2000...3 drive software raid. OS via SSD chip as above.

Doremi IMS 1000...3 drive hardware raid, but they chose the wrong raid controller and basically every one fails. "Laptop" drives too.

Doremi IMS 2000...back to the 3 drive software raid nonsense with "laptop" drives.

USL IMS thing..."laptop" drives. The good news is that there are 4, but the bad news is they are solid state. I've major concerns for the use of a solid state raid array. I don't recall if it is software raid or a true hardware raid.

Christie IMS...external NAS. Continual failures.

Sony...I'm not even going to go here.

Where does that leave us?
3 drive raid = dealbreaker
Software raid = not making me happy
Laptop size 2.5" drives = no thanks

I would hesitate to say "preferred" server, as that seems to imply ideal. So let's go with "least sucky" and say the DCP2K4/ShowVault.

Second option, although I haven't had a chance to test it yet, would be USL with non-SSD drives loaded.

Third option forces us to a 3 drive raid since that's all that is left, so the GDC IMS thing is the least painful of those options.

Posted by Steve Guttag (Member # 268) on 06-12-2015, 11:22 PM:

Brad,

On the GDC SX3000, you can go to 4-drive on their PSD (laptop drives) or go with an Enterprise grade storage unit with up to 5 drives, though three are standard.

Posted by Brad Miller (Member # 2) on 06-12-2015, 11:24 PM:

I stand corrected.

I guess in the least sucky order then.

1. Doremi DCP2K4/ShowVault (real server)
2: GDC IMS thing (assuming it actually is a hardware raid)
3. USL IMS thing (with mechanical drives loaded)

USL could be tied or beat GDC if they have a hardware raid and/or their GUI isn't awful like the GDC interface.

Posted by Marcel Birgelen (Member # 6801) on 06-13-2015, 12:29 AM:

Time to roll your own server then? [Smile]

I guess it's about time to focus on a stable IMB/IMS implementation with reliable NFS client and a stable NFS server/NAS solution to go with that.

Something like FreeNAS looks quite promising. It's using FreeBSD as core OS, which is usually rock solid and has a stable NFS implementation. It works fine with the Christie Solaria IMS, but both the soft- and hardware of that IMB have been nothing but trouble.

You can put the FreeNAS software on any kind of standard PC server you like. Out of the box Dell, HP or Supermicro server, with HW RAID, redundant PSUs and almost anything you want.

A beefed up machine could possibly serve multiple screens, although you should do some serious performance testing and you're creating a nice big fat Single Point of Failure this way.

Posted by Carsten Kurz (Member # 5396) on 06-13-2015, 06:59 AM:

quote: Brad Miller
Sony...I'm not even going to go here.

Then I'll do:

6 drives hardware Raid6 with dual drive failure redundancy, one additional spare drive with automatic rebuild function prefitted in every server. That is, threefold failure redundancy. System SSD. UPS standard item for every server. All this already in their entry level systems.

Ah no, all this NOT if you order a Sony with a Doremi ShowVault and IMB-SA...

BTW - adding a fourth drive to a doremi only increases capacity, it doesn't offer more redundancy. Plus it slows down writing/ingest considerably.

Still, if I'd buy a 'real' server (not IMS), a ShowVault +IMB is the right thing to get.

That said - we need an IMB for a NEC 900C. What do I do? The IMS1000 seems to be out of the question now, and we can't wait for an IMS2000. GDC is not actually sold/supported here. I'm open towards a USL, especially as this system will be transported, and the SSDs are beneficial there (don't care about capacity or long-term stability, this will not be a daily operation). But the USL server is also not sold over here, and I hear it still has HDMI issues.

@Marco: It is obvious that Doremi/Dolby goes the WebGUI way for their discrete server systems as well. That means, on their IMS it is only WebGUI, on the classic server it will be the classic GUI for the old buffs, and the IMS-type WebGUI for everyone who doesn't hate it. Which is not a bad choice, to have both options.
The basic concept of the WebGUI is good, I think, placing all control functions in a more customizable front-end, instead of having it in OS dependent specific apps. It offers more options for future development and ergonomic improvement.

As a matter of fact, I don't choke on the IMS GUI. Maybe Dolby will streamline the Doremi WebGUI a little bit more towards their liking - the IMS2000 could be the first to show this. And then the WebGUI for the classic Doremi would follow. At least there is hope.

- Carsten

Posted by Marco Giustini (Member # 4544) on 06-13-2015, 07:14 AM:

quote: Brad Miller
they chose the wrong raid controller and basically every one fails.

The RAID card has been replaced on all machines. Hopefully this works. Still, the Doremi IMS is #99 on the list! [Smile]

I am still puzzled that Dolby are completely dopping the ShowManager software - and Doremi haven't been developing much lately.

I thought Dolby would bring the best of both world in the new Doremi systems but it seems a little late now?

Posted by Dave Macaulay (Member # 813) on 06-13-2015, 07:37 AM:

Hardware RAID is a bit of a red herring. You put another CPU in charge of the array but it's still a software controlled system. The RAID hardware (if m/b integrated) or card are hardly failure proof. The drives - and their failure rates - are identical to a "software" RAID.
Shows are going to be lost occasionally, regardless of what hardware you pick.
On several hundred screens since conversion, the many lost-show problems I've seen don't have a glaringly common cause.
I can only think of one screen down because of the Doremi "software" RAID: the site ignored a single drive failure and then a second drive failed.
A few Doremi OS flash chips have failed. That was a hassle a few years ago but the current software simplifies replacement.
Servers like the Christie IMS with NAS content storage allow you to use high reliability storage hardware: the downside so far is that the IMS hardware and server GUI have "issues".
The Doremi 2K4/SV4 with 4 drives advantage is storage capacity: 6TB if you use 2TB drives (who needs that??). It uses RAID5, with the same single drive failure tolerance.
I don't think a central on-line content storage model for a multiple screen venue is workable. A single point of failure - very bad idea. I've seen several TMS failures... it's a major PITA to handle ingests and scheduling manually but it can be done.

Posted by Mark Gulbrandsen (Member # 72) on 06-13-2015, 08:40 AM:

I'm glad that I did two things. GDC and NEC. My combined screens (335) have less troubles then some people have with twenty screens of other manufacturers products. The 50 or so GDC TMS systems I built have also been bullet proof with only two hd failures among them in nearly five years.

Oh, and on the non redundant GDC OS drive there is plenty of room in the drive bay to bolt a back up OS drive all ready to go right next to the one that's there (they are serial number coded to the server). GDC will also sell you a brand new server with a backup OD drive already in place. But you can copy your existing drive if you want to but be sure to copy all the drive because it is divided up into several smaller areas. If the OS drive fails you merely move the SATA connectors over to the new drive then reboot and remake your show line up and run your movie. You don't loose the content! This can all be done in less then 5 minutes if your server is on rack slides. Let me see you beat that reloading the Dolby OS and ALL of your content on your DSS XXX!

Oh, and I still have yet to have a GDC OS drive fail. Some are over 5 years old now. When they do fail they won't be missing any shows either.

Mark

Posted by Steve Guttag (Member # 268) on 06-13-2015, 09:48 AM:

We all have our experiences, likes and dislikes.

For me...the Dolby DSS line represents the most reliable servers in the industry. The ONLY servers I've had to replace to get shows back on the screen was one where the power supply make the guts look like a road flare went off inside (Supermicro power supply). The other was an old DSP100 that locked up on boot up (has since been repaired). That is two servers over however many hundred we have in the field and one of them was over 5-years old.

Have there been show interruptions? Yes, the software has not been perfect but lost shows, none in our world other than the aforementioned ones.

Dolby's GUI again is simply the best. It is intuitive. I've trained people that don't know how to operate computers on it and they pick it up fast (seriously...one guy, 74, did not have the skills of "click" or "double-click." He runs his theatre very old school...right down to paper tickets off of the roll. In fact, there are no printers of any type I've seen in the theatre...not even a fax machine.

Can you teach someone the other UIs? Absolutely...but never as fast and as consistent as Dolby.

I think the thing I hate most about the DSS server is the 6:30 boot up time (faster on the DSS220). Since I use automations, the lack of macro capability on it is of no consequence to me and it does have real relays so you can control things like masking and lights out of the box.

But if you need to bank on the server running that night...nothing beats a Dolby DSS server, particularly the DSS200 though the DSS220 starting in System 4.6 has been reliable too.

Need a cheap TMS...well if you have three screens or less (and most have tested up to four without an issue) Dolby does that out of the box for no extra $$$. There is no learning a different system.

Another big plus on the Dolby DSS line...you can kill the power to them without corruption unlike GDC and Doremi.

I also like the Jupiter Client but there are times it is very advantageous to work on a system from a support standpoint without having to share the UI with the end-user.

The USL GUI seems to mimic a lot of the Dolby GUI so it is one that we're looking at. But I have no hard data on its reliability or shortcomings.

GDC's GUI is a mixed bag. I think their SMS screen is just fine and I like the whole touchpanel aspect of it. I also like that via VNC you can remotely control it and set up shows (even for a manager this can be handy rather than handy). Having to go through the "Control Panel" screen to get to the Content/Ingest stuff seems like an unnecessary step.

I have been successful training non-computer literate people on the GDC GUI though it took about four times as long to be proficient as with the Dolby GUI. But like most things, once you get used to them, you get proficient/fast with it.

I don't like that you can't get to the configuration pages once the show starts. It is an unnecessary hindrance for at least some of the things. If that stuff was moved over to the Admin screen, then it would still be password protected but also available. Those things that absolutely can't be changed once a show starts, one could leave there. In fact, the swap the config button and the content button in the control panel and you'd have something that makes more sense to me. Then, the user has everything they need right there on the SMS screen where they live.

This business of having to shut down the server before the projector (IMB) is silly. Yes, I've automated it so that the automation shuts things down but it is not intuitive to the end user. To them, the IMB is part of the projector. Since there are IMBs out there that have figured out how to survive and instant power off (e.g. the CAT745), it is time to step up your game. Make your system robust .

Failure rates on our GDC systems do not mimic Mark's accounts. The HFR IMB has had a rather large failure rate in our world and it is essentially the SX-3000 but with a conventional OS in a conventional server chassis. We don't have many SX3000s out there but they have not been problem free. I do not want to give the impression that they are dropping like flies but in comparison to Dolby, the hardware has been nowhere near as reliable. Likewise for their earlier HDSDI based systems...they have been mostly reliable and as Mark said, they kept their UI throughout their models and I commend them on that. If you know GDC, you know ALL GDC models.

I don't like that GDC's logs are encrypted such that only they can read them. I can only imagine that when they pull logs they are getting some sort of sensitive data in there that needs a level of security to them. Christie series 1 projectors did that too. However, if you have the right access levels and know where they are stored, you CAN get to the individual log files.

Doremi I have the least experience with but I can say without hesitation, the UI of the IMS1000 is an abomination. Seriously? A black background with warning messages with red lettering? Did someone flunk GUI school on that one? And your system, based on a web GUI requires a shutdown before turning off the projector? Seriously? So is it reasonable that there will be a web interface at each projector or is the manager just carrying around their phone on the booth network? Or does one shut all of the servers down from a central location and then shut the projectors off?

What Doremi has always been good at is being first and nimble with the changes in the industry. They were the first on the IMB, first on the IMS...got problems with captions, no problem, WE'LL generate them as burned in right here on the server! Want to display them but the show didn't have the font needed in the package? No problem, we got one of those too! Doremi has a reputation of being able to play the "difficult" CPLs.

I wonder how this will change under Dolby. A big problem of Dolby and the DSS line is the lack of nimbleness. It can take months to years to getting feature implemented. When I ran into the subtitle font thing, I was quoted DCI rules rather than "yeah, we can come up with a workaround for that."

Doremi does still have the HDSDI based servers for those that are in the 2K world and want reliability. I dislike the PICe cable thing on their IMB solution...big, stiff and limits you to under 15-feet. I prefer the Dolby or GDC Ethernet based solution.

Christie and Barco both have IMB/IMS type solutions but if you go that way, you are pretty much committed to their respective projectors...which can pose problems, from a uniformity standpoint. Both are new guys to the server world and both have had some serious growing pains. The Barco Alchemy, when fully operational, will bring 4K alternative content to the table (at faster than 30Hz). For us, this could be a big feature since this has come up several times in several locations. But again, that requires a Barco projector. What if the existing projector is a Christie or NEC?...they have to throw out a projector to put in a server? Seems silly to me. And Christie's IMB...100% failure rate in our world...fortunately, never sold one and have replaced them with Dolby DSS200s when possible.

Posted by Dave Macaulay (Member # 813) on 06-13-2015, 10:04 AM:

The PCIE cable limit is 7M, about 23 feet.

Posted by Steve Guttag (Member # 268) on 06-13-2015, 11:48 AM:

Okay...23-feet...switch to an Ethernet based solution (or even fiber) and go out to whatever you want, pretty much, with far cheaper cable. That was the first IMB out the door...time for an update.

Posted by Harold Hallikainen (Member # 5405) on 06-13-2015, 11:55 AM:

Wow! Great conversation! I'm really learning stuff. One idea that popped up a bit up the thread is the unreliability of a single server driving several screens. I wonder if RAID could be replaced or supplemented by RAIS (https://en.wikipedia.org/wiki/Redundant_Array_of_Inexpensive_Servers ). You'd have a local "cloud" providing capacity, speed, and redundancy. You are still relying on a network switch, but perhaps several could be used to provide redundancy there too.

Harold

Posted by Marco Giustini (Member # 4544) on 06-13-2015, 12:11 PM:

Harold,

Not familiar with the product, but isn't that what Cinemeccanica is doing?

quote: Mark Gulbrandsen
Oh, and I still have yet to have a GDC OS drive fail. Some are over 5 years old now.

Mark. Hard drives do fail. It's not a matter of IF, but of WHEN. If you design a system where the OS is stored on a single HDD, you make a mistake, at some point ALL your systems will be off screen for the time it takes to replace the HDD.

It is a fact, not a speculation.

Posted by Steve Guttag (Member # 268) on 06-13-2015, 03:48 PM:

I like how he says that with that oh-so-polite British accent! [Big Grin]

Posted by Carsten Kurz (Member # 5396) on 06-13-2015, 07:29 PM:

quote: Harold Hallikainen
One idea that popped up a bit up the thread is the unreliability of a single server driving several screens. I wonder if RAID could be replaced or supplemented by RAIS (https://en.wikipedia.org/wiki/Redundant_Array_of_Inexpensive_Servers ).

I don't think one should increase system complexity by using distributed storage. As a matter of fact, I do not even see a real issue with existing RAID systems locally connected - and I don't think anyone here does. Nothing beats that in terms of guaranteed realtime performance. Hardware RAID , software RAID, both have their issues, but in general, there is little to complain as around 10 years of regular digital cinema operations prove.

They talk about increasing HFR/4k bandwiths beyond 500MBit/s. How do you support that with traditional network storage hardware. Yes, you can go 10Gig, but where is the benefit cost wise over traditional locally connected storage?

Even if a distributed system creates a different level of redundancy - it still remains a single point of failure for a whole multiplex - be it on the switch or software complexity/central management level.

Now who wants all his screens to go down the same moment? And for what? To have a bunch of storage servers in a central location as opposed to individual servers in the booths?

We should be glad all established servers actually run as solid as they do. The manufacturers should strive to keep that level of robustness for their future products, not create additional weaknesses.

- Carsten

Posted by Mark Gulbrandsen (Member # 72) on 06-13-2015, 07:53 PM:

Marco,

No argument at all. Of course it's gonna die some day. But as I mentioned there is room for a backup OS HD in there and it takes all of 5 minutes to slide the server out move the cables over to the new drive, reboot and make up your show. Basically zero down time at all. And you don't loose your content like you do with a Dolby.

Mark

Posted by Brad Miller (Member # 2) on 06-13-2015, 09:01 PM:

Mark, we used to do that with GDC servers. In fact I believe I was the first one to suggest it. Doesn't matter though, the issue is unfortunately as the software is updated, your backup drive sits there on outdated software. Likewise any change in the system (perhaps a new cue or something) isn't updated. In the end by the time the backup drive is utilized, it's so far out of date that you aren't really saving the next show.

Since you apparently haven't been through the process yet, when the drive does die and you purchase another one from GDC, you then have to sit and step through all of the upgrades to get to where you need to be, so even that isn't a "simple" fix. It all takes time.

Quite literally of all the servers, a corrupted OS can be remedied the fastest on a Dolby with a clean install disc. That's something nobody else offers. Drop it in and voila, it's just like a new server again. (Note, by "just like a new server", that also means no content loaded.) There's a catch to everything! However the Dolby is still the fastest to recover from. Doremi is second, as you need to update the software, firmware and security manager, but you do get to save the time of not having to reload the content (like with GDC...but GDC takes longer to upgrade than Doremi).

I do have a great idea for you Mark. Since you are the only person in the universe that has never had a GDC fail, you need to offer "Mark G's extended warranty and tech support for GDC products". I would gladly sign up and you wouldn't have to do anything since any GDC you are involved with is apparently impervious to failure. [Razz]

Posted by Steve Guttag (Member # 268) on 06-14-2015, 01:26 AM:

Content may be lost on the Dolby DSS but you can play your feature right off the CRU drive...always could since day-1. In fact, play it off the drive and transfer it to the server at the same time so it will be there for next show.

The sucky things about recovering from Dolby are the multiple long-winded reboots on them. You have the config reboot to get all of the IPs and screen number set just so, then the serial cue ingest and reboot. And all of this is after the reboot needed after updating/installing the server.

I just did some GDC updates...time wise, they are on par with Dolby if not a little faster (depends, the DSS220s update/boot faster than the DSS200s) and the GDC boots up WAY faster than Dolby.

And as for Mark's comment about losing content...it is a RAID 5...you should only lose everything if you lose two or more drives simultaneously (or if you run an Install disc for a CLEAN install). But if you are trying to get rid of a nagging problem...a clean install is the best way for any system...not just servers.

Posted by Tim Sherman (Member # 569) on 06-14-2015, 02:27 AM:

Don't forget on the Doremi server you can plug a flash drive into the front usb and run a terminal command to back up the boot os ssd. If your ssd ever fails, open up the server and unplug the os ssd usb cable and plug in the flash drive to boot.

Posted by Marco Giustini (Member # 4544) on 06-14-2015, 06:27 AM:

quote: Mark Gulbrandsen
But as I mentioned there is room for a backup OS HD in there and it takes all of 5 minutes to slide the server out move the cables over to the new drive, reboot and make up your show. Basically zero down time at all.

Cool.

So the server fails, the customer calls you and you ask Scotty to transport you straight on site to slide the server out and swap the cable. Total down time: 5 minutes, no shows lost, patrons clapping you from the auditorium while the curtains open and you disappear in a glittering of light to the infinity and beyond.

I also guess all your servers are in very accessible positions, clean and without a tangle of cables on the back.

Also, as Brad mentioned, I assume that the other HDD managed to keep up with the upgrades wirelessly over the years.

==

I feel server manufacturers should do more to make the servers more robust in case of failure.

Dolby should make a 'reinstall OS only' feature on their servers. Just the OS partition. It doesn't sound very complicate to do.
Also, we need a "backup all" button. Everything can be backed up but it's not straight forward. They also need to introduce a way to unconfigure the server without losing shows, serials and GPIO - or at least allow us to easily restore those. You can back up serials and GPIOs at the moment, not shows.

I appreciate that the reason for the server to malfunction may be the show database (it's just XML files though...) but I'd like to try and I could repeat the procedure with no shows if it still does not work. A script to try and fix a broken database would also be a good thing.
Finally, we should be able to extract logs from VNC and save them on the content partition, which can then be remotely accessed.

Doremi should move the OS on the RAID - maybe keep the SSD but allow the server to boot from RAID if the SSD has failed. That would give 100% redundancy. Doremi should also allow the customer to restore some sort of ISO on the SSD, in case something wrong happened, even though that has never been the case.

The emergency USB is a good idea, but again it implies that the customer is happy to open the server. Also implies that the USB stick is up to date with the software. Doable but I don't feel it's the best backup plan here.

Posted by Mark Gulbrandsen (Member # 72) on 06-14-2015, 10:02 AM:

Nope, my customers all know how to do it themselves. To send a tech some times 600 miles to move two cables over is pretty stupid. As stupid as sending the same tech just to relamp their projector.

Brad, no real need to keep that drive updated. It's still going to boot up and play back even with older firmware versions. Although I do upgrade that drive also once a year. I have a number of customers still running on 7.8! There is no real need for DCI firmware YET unless you are getting VPF's. Studios are still issuing kdm's for those running 7.8.

Also note that GDC also offers a clean OS install, or just an upgrade. No need for a CD either. A USB drive is actually preferred. Or one can also do it remotely. GDC upgrades are now few and far apart. Only two or three times a year. SX-3000 has only had four updates in its life span so far.

Mark

Posted by Marco Giustini (Member # 4544) on 06-14-2015, 10:47 AM:

quote: Mark Gulbrandsen
my customers all know how to do it themselves

Wow, always thought that Cinemas are run much better there than here. We constantly receive phone calls by panicking managers who were just shown how to create the playlist and cannot distinguish between the server and the projector. Clearly all your customers are very well trained and skilled in both projection and IT.

Clearly the cinemas you look after have a competent person available 24/7 to keep the down time to 5 minutes - which means that there must be several trained projectionists/managers at the same site.

So every year you have to update the server, then open the server, swap the cables, replace the server and update it again.
Then repeat to go back to the original disk - or do you rotate the HDDs every year so they wear out evenly?

You're right, it seems the most practical and the most sensible thing to do, I can't understand why the other manufacturers went for a different route.

We should tell Dolby to go back to the Flash Disk solution - Improved, by placing another flash disk somewhere else in the chassis.

Posted by Steve Guttag (Member # 268) on 06-14-2015, 10:56 AM:

How many of your GDCs are on rails? The only one that seemingly supplied rails was Dolby. And of those, how many people made the cables and secured them such one can slide the server out without yanking the cables all to heck? I know WE did on the Dolbys but I've not seen others do it. It took a bit of work too to put the rails into various pedestals or racks as the rail placement often didn't fit in with the rest of the rack's needs or how the pedestal was constructed.

As for GDC 7.8...I agree...up until this year. This year, we are getting SMPTE packages with greater frequency...at some point your customer is going to get a package that won't play/ingest. 8.01-build200 (current) should be stable..build 186 was mostly stable...just found a problem with it this past week after having months of glitch free life.

On Dolby 4.9, the ability to remotely save serial and GPIO I hear is coming (believe it when I see it).

With Dolby, you have to think like they do..."Everyone should buy Dolby products." As such, everyone will by a Dolby TMS like the DSL100 or DSL200. With one of them on your system, it keeps a database of all of your shows and cues and everything. Thus, if you run an install disc on a server...once you configure it to the correct auditorium, all of your shows and cues and everything will be put right back where you had them in no time without ANY effort on your part. All you need to do is move the content. No problem, use the Copy Clips feature to pull the content from the "best" source on the network (Library if possible). And in no time, you'll be back up in full running capacity. See how easy it is? Your only mistake was NOT buying DOLBY EVERYTHING because you wanted things like "VPFs" or wanted a "Real Automation" or anyone of a number of reasons to not buy Dolby everything...for that you will not have the niceties of a DSL backing up the database. However, if the DSL can do it, then it should be doable period! Wouldn't it be nice to be able to have that sort of storage off the server in case of emergencies?

Remember too...if you do things the DOLBY way, the DSS line was by far the most pre-configured out of the box. Their only big flaw in the plan was to presume Theatre 1...thus if you bring a server back from scratch and it isn't Theatre 1, it will stomp on Theatre 1 until it is reconfigured. After an Install disc, it probably should have come up asking what auditorium number you wanted it to be.

Posted by Leo Enticknap (Member # 534) on 06-14-2015, 11:03 AM:

quote: Steve Guttag
For me...the Dolby DSS line represents the most reliable servers in the industry.

Dolby's GUI again is simply the best.

+1. Although I've only spent serious time with one other server model with which to compare it (Doremi DCP2000), this is another vote for the DSS servers and Dolby's front end. From a hardware standpoint they are thoughtfully designed (e.g. the components that are most likely to fail and need replacement, principally hard drives, fans and power supplies, being very easily replaceable), and the Show Manager front end does everything you need it to and is very clear and simple.

There are two features that the Doremi has that the Dolby does not, that I miss on the Dolby, however:

- The ability to specify a pause point in the DCP that will pause automatically. So if you've got a 10-second DCP of a logo slide to display while someone is speaking, you can set a pause point in Doremi Cinelister, such that it automatically pauses midway - you don't have to remember to press the button.

- No DNS capability on the DSS200, meaning that it has to have a direct, static IPv4 address to the outside world if you want to ingest content over the Internet.

Also, if Dolby were to provide off-the-shelf packages of serial automation commands to drive commonly found non-Dolby equipment to which it could be hooked up (e.g. AP20), that would make it even better.

That having been said, these are minor niggles compared to all the good points of the Dolby machine, and I'm still scratching my head as to why it's being discontinued. I really hope that at the very least, Dolby produces a version of Show Manager to run on Doremi hardware. The Doremi pseudo-Windows style front end is simply abysmal. The projectionist who uses our Doremi the most is a hugely knowledgeable, experienced and unflappable festival veteran, and she loathes it as much as I do!

Another vote for Dolby is in their customer service. If you open a ticket on their website and flag the issue up as potentially show-stopping, the phone will ring within minutes - guaranteed.

Posted by Marco Giustini (Member # 4544) on 06-14-2015, 11:17 AM:

quote: Steve Guttag
On Dolby 4.9, the ability to remotely save serial and GPIO I hear is coming (believe it when I see it).

Finally!!

+1 on "ask me what auditorium I am" during install. May I also suggest to ask for confirmation before wiping everything? I once made the terrible mistake of leaving an install disk in the drawer: the system was about to be wiped but then it regained consciousness. I rebooted to make sure the system was ok... and it was not ok anymore [Smile]

Re. the backup on other servers, what if the current database is causing trouble? You'll have to unconfigure all the servers while offline to wipe all the copies of the database. If you're just reinstalling for a failure, fine. If you're chasing some weird issue and you want to wipe, then you don't want the same database to be restored on your server!

Posted by Steve Guttag (Member # 268) on 06-14-2015, 12:15 PM:

A couple of points.

Regarding the Pause cue...how handy are you with solder or crimps? Use the GPIO port on the DSS server to define one of the GPOs to a toggle called "Pause" and then tie that to the "Pause" GPI pin. When the cue fires...it self pauses and awaits your command to restart. Not exactly rocket science but "automation" is NOT the DSS' strong suit. If you have an automation system...again, no biggie...define a cue for Pause, have the automation "Pause" the show.

True that on the lack of DNS (also means thinks like the NTP pool are not very viable). BUT it isn't too hard to get the IP of something "out there" on the web. Just ping to the web address and the ping will return its IP so it isn't the end of the world for just setting up a site for FTP like it would for how the NTP Pool works. That said, Dolby is so strict on the NTP thing and how many hops it takes to get to NTP, I use our Cardinal Care computers as the NTP source so NTP is just one hop away. NTP reliability has skyrocketed since. Our NOC systems where the NOC is a VPN tunnel also are quite reliable since the number of hops is minimal.

As a matter of personal preference, I don't like any part of the system to really be out of the building for security reasons. So again, the lack of DNS is a non-issue for us, it never needs to know about things outside of its small little universe.

Marco...I hear 4.9 will also allow (now brace yourself) Ethernet control of external devices like sound processors. What a modern concept! I bet the others will do back flips trying to catch up to this "technology." [Roll Eyes]

No problem on the database at all. You don't have to do anything OTHER than unconfigure the DSL for the database and force it to rebuild. I recently did this when I changed the IP scheme of a site. I wanted it to forget about all of the little IP addresses it once knew and then work with the new ones. I had the DSL200 unconfigure just about everything but content. Pretty quickly, it found everything and all was harmonious.

The Individual servers maintain themselves and are their own islands. So they were unaffected by it all. They still had all of their shows and cues and everything. It was the DSL's job to see what each server had and harmonize ITSELF, not take anything away from an existing server.

As such, your situation can't happen can't happen. If the database is wacked somehow. Take the DSL off line, make sure all of the individual servers have what you want, and then unconfigure the DSL to force it to rebuild the database clean...presto...you are back in business in almost no time.

Again, if you use Dolby everything, they have you covered VERY well. If you stray from the pack...well...shame on you! If Dolby didn't make it, you don't need it. [Wink]

Due to how the serial thing works (And how hokey it really is) there is no way to have preconfigured libraries. They don't even do that for Dolby products. What I have are serial files for the complete systems we use. And since we almost always have an automation (eCNA), all of the commands are identical to the automation regardless of projector, sound processor...etc. It is the automation's job to translate that to actual equipment commands. For those we do maintain a library of various commands as does EPRAD. But again, you'll get no argument from me that DSS's automation "features" are lacking as compared to most all other products. What you really want, Leo, is in the configuration of the server for Dolby to take their logo off for a moment and add other brands to its built-in knowledge base so you can have USL, QSC or Datasat or whatever processor be a choice, just like for the projectors. Remember, on a DSS, the volume indicator is also a remote control. You don't have to preset individual volumes like "5.6" as in other servers. If you have a CP650, CP750 or CP850, the entire range is at your disposal at all times. And what do you do with people like QSC? That INSIST on the silly dB scale for volume control? The volume indicator is part of the Dolby UI. Handy too since it can be used as a remote volume control if you are VNCed into the server on your phone. There is but so much burden you can expect a sound company to accommodate another sound company's product and their "design choices."

Posted by Marco Giustini (Member # 4544) on 06-14-2015, 01:21 PM:

quote: Steve Guttag
Marco...I hear 4.9 will also allow (now brace yourself) Ethernet control of external devices like sound processors.

Gosh, I need to sit down and have a drink now! [Eek!]

quote: Steve Guttag
If the database is wacked somehow. Take the DSL off line, make sure all of the individual servers have what you want, and then unconfigure the DSL to force it to rebuild the database clean...presto...you are back in business in almost no time.

Ehm... I am not sure I share your optimism! Yes, this works, but it's far from being easy. My points:

1. I do not see many DSL's around. As such, Screen 1 is usually the TMS. Unconfiguring Screen 1's database means... you're losing the shows and serial commands for that screen.
Ergo, in case something goes nuts I have to unconfigure both ShowStores and lose all the shows. Not ideal.

What I am suggesting is some more interaction:

"A backup database has been detected, would you like to restore it? Yes/No" - Easy peasy. [Smile]

DNS should be enabled if there is no specific reason to keep it off. Yes, you can ping pool.ntp.org and see what it says but NTP servers do go offline or - as seen a few times - the ISP will block port 43 so you can still ping the server but it won't synchronise. To test remotely you'll have to login as root, disable ntpd, run ntpdate on that server to see if it syncs. If not, repeat till you find one which works.

If you don't know the root password - it shouldn't be known - the only way is to set a different IP and reboot and wait for the UI to come up. Again, not practical nor ideal.

Posted by Mark Gulbrandsen (Member # 72) on 06-14-2015, 02:46 PM:

All of my servers are on rails. I've done this from day one, even with DSS-100's. Its neither expensive nor difficult.
Haven't had any isses ingesting or playing back any content any where that has 7.8. No doubt it will eventually have to go.

Mark

Posted by Marcel Birgelen (Member # 6801) on 06-14-2015, 05:23 PM:

quote: Steve Guttag
True that on the lack of DNS (also means thinks like the NTP pool are not very viable). BUT it isn't too hard to get the IP of something "out there" on the web. Just ping to the web address and the ping will return its IP so it isn't the end of the world for just setting up a site for FTP like it would for how the NTP Pool works. That said, Dolby is so strict on the NTP thing and how many hops it takes to get to NTP, I use our Cardinal Care computers as the NTP source so NTP is just one hop away. NTP reliability has skyrocketed since. Our NOC systems where the NOC is a VPN tunnel also are quite reliable since the number of hops is minimal.

The problem with the "something out there" concept is that often, stuff just gets renumbered. The IPs in an NTP pool are not constants, that's the whole idea of those DNS names anyway.

Keeping your clock aligned isn't obviously just important to start your shows on time, your whole KDM validity and keeping your media block clock in line with UTC is equally important. But it doesn't really matter if the clock is off by a few seconds, so it's a bit counter-intuitive Dolby's NTP implementation is so nit-picky.

I'm not sure if Dolby is really counting TTLs, but many NTP clients are configured to be quite picky when it comes to stratum level and jitter.

The best solution is indeed to use a local NTP source (and let this connect to one or more of those NTP pools), like a router, firewall or other server. Something that's not due to change the IP address without some consideration.

Meanwhile, I'm still waiting for what I would consider the ultimate IMB:

- Cross-platform availability, so no vendor-supplied solution like Solaria or Alchemy.
- Intuitive and well designed GUI, comparable to Dolby's implementation.
- All advertised features and on-board ports (like HDMI) actually do work.
- HDMI that works and actually supports 4K.
- No local content storage, but a rock solid networking and NFS implementation, so you can roll your own storage, using reliable standard solutions.
- Redundant flash for local settings, databases and OS.
- Good hard- and software support that keeps track of recent developments and actually tests their patches before they ship them.

At this moment, it looks like we're further away from any of this than ever.

Posted by Steve Guttag (Member # 268) on 06-14-2015, 05:35 PM:

Oh Marco...why do you not see the error of your ways? In a "proper" Dolby system, you ALWAYS have a DSL!!! Hence, no problem (and hence why it is considered not a problem in the world of Dolby where you always buy the house brand of everything). Your example of being a mini-plex using server 1 as the TMS...we'll that was just an econo thing "we" put out there...we can't put all of the "goodies" in such a thing or why pay for our DSL? (Pssst....you do realize I'm playing a bit here, right? But in seriousness, that is how Dolby as traditionally designed things...for a pure Dolby world and everyone else should bend to their design. And in the analog sound processor days, the world essentially did. Dolby had the patents on the technology used by the industry. It didn't work out that way for them in the DCinema server world, did it?).

BTW...with a DSL, you DO have the option of backing up the Database to a USB stick and loading a saved database back onto the DSL. Furthermore, the DSL does make periodic backups (kind of like Windows "System Restore" points). Here is another thing I didn't know until recently (since I don't have many DSLs myself)...if you have a DSL...the DSL must have network access to all of the components that may need to be controlled via Ethernet. That is, if you have a CP750 and you want volume cues and the volume control to show up on the TMS pages, the DSL100/200 has to have a means to "reach" that CP750. That is why each server has its own internal router. The DSL sits on the 192.168.241.x network but connects to each screen server on the "Theatre Network" NIC...via that server, it can "reach" the CP750, or DFC100 or any of the other Dolby peripherals. Force the local server back to "SMS" mode and it will always find them since they are on the same network as the server. I always figured that the individual servers were what talked directly with the local devices and that the DSL merely kept track of the shows and ran the GUI. That overhead is already at each server but no they decided that if you have a DSL, then IT is what will talk with the accessories. If you set up a routing table for each theatre in your router, you can connect up to just connect your laptop up to the 192.168.241.x. network and "reach" everything. As such, with a "proper" Dolby network, you need only run a single Ethernet line around the complex for both Media and Management rather than maintaining two separate networks like most everyone else does. A beauty in their system is it leaves you with a LOT of IPs available for doing more complex systems since you have 128 IPs PER SCREEN to work with.

As for DNS...yeah, things like NTP POOL is out since its IP changes but there are plenty of "fixed" NTP places to poll from and really you just need one thing to reach there...like the DSL since each server can also be an NTP reference for everything else on the network(s).

If you worked with a DSL system...you'd see that it is pretty bullet proof in terms of databases and getting back on line...I mean it is just about plug and play. Really. However, as TMSes go, the Dolby one isn't what one would call "feature laden." They also haven't updated it in probably a decade. So the only show schedule import it knows is its own and anyone sending a schedule to it MUST conform to their format. (see above post(s) regarding Dolby mindset on other products interacting with theirs).

Posted by Marco Giustini (Member # 4544) on 06-15-2015, 03:50 AM:

eheh, I see what you mean.

My mom used to say 'you have salame slices on your eyes' then [Smile]

Posted by Stephan Shelley (Member # 8587) on 06-18-2015, 05:56 PM:

The DSS 100 and early DSS 200 with X7 boards the operating system was on a separate IDE flash drive. The DSS 200 with X8 boards do not have the IDE flash drive. You could keep a spare if needed with they older systems.

Even with the mini-plex configuration the show database is on all server so if you lose the TMS server it will restore the shows from one of the SMS servers. If you are a single screen then you are out of luck.

Posted by Steve Guttag (Member # 268) on 06-19-2015, 06:18 AM:

I wouldn't call that the database. Each server has the show list already on it. If you yank the Theatre Network cable out of any server (even while the show is running) it will automatically switch back to SMS mode and ALL of the shows will remain on that server (with "X"es on any content not resident on the server). If you add a show in that mode and then plug the Theatre Network cable back in, once it joins the TMS mode...the other servers on the TMS will get the new show too (not the content but the SPL).

With the DSL...it actually has a database of various configurations like serial cues and such and if you plug in a brand new server configured with the same IPs as the one you removed...it will proceed to populate the new server with everything the old one had, including cues and such. It will not move content back on though...that is always a manual job.

Posted by Stephan Shelley (Member # 8587) on 06-19-2015, 08:18 PM:

Once you have the SPLs restored, you can right click on the show title in the build tab and tell it to copy all the missing clips. If they are anywhere on the theatre network it will copy them. Saves doing them one by one.

If one is in a pure Dolby world you don't have serial or GPIO cues once the replacement server is configured it finds the cues on the na10 Dolby processor and such.