Film-Tech Forum ARCHIVE: GDC PSD-300P Storage units

This is topic GDC PSD-300P Storage units 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=001960

Posted by Monte L Fullmer (Member # 2797) on 07-03-2014, 12:09 AM:

Another question arises:

With the PSD-300P units (the one with three, 2.5inch laptop drives) that is connected by one cable to the SX-3000 unit, any reports of durability and longevity with the drives in these units?

We have these in our location, and with being on all the time as with the BARCO they're attached to, do these things needed to be rebooted as with the SX3000, for heard that doing reboots on these RAID units is asking for trouble.

But, it seems with weekly reboots of both the SX-3000 (with shutting it down via the GDC GUI in the TMS) and projector, I still have issues with skipping and freezing content during playback, where one has to stop the playback, shut down the IMB, then power off the projector for a few seconds then turn back on the projector which turns powers back up the IMB. Then, head back to the GUI and restart where one left off.

I wonder on using laptop drives being that of their smaller size than regular 3.5 SATA drives if these smaller drives can hold up for duration.

I definitely miss full sized servers.

thx -Monte

Posted by Matt Olson (Member # 6849) on 07-03-2014, 12:45 PM:

I have installed 32 of them so far running with Barco projectors. I have had only one skipping issue in the last 6 months. What software version are you running I found 265 to be the best 271v2 I have had some TMS issues.

I have never shut off the power to my hard drives have them hooked to a UPS so they stay running all the time. From all my time working with computers in the past I would think the 2.5" drives would last longer then most 3.5". Laptops are designed to be moved around a lot, powered down often and stuffed in a warm case so I don't see them being a issue with this setup.

Posted by Carsten Kurz (Member # 5396) on 07-03-2014, 01:30 PM:

2.5" drives are available for notebooks and as enterprise grade drives. E.g. the Doremi IMS uses the enterprise grade drives. Don't know what GDC uses.

The problem with all consumer grade drives in a RAID is that they do not guarantee a specific response time. In a RAID, the redundancy takes care of lost individual drive data. Consumer grade drives do not know that they are operating under a RAID shield. So, they may try to recover data repeatedly for a long period, keeping the RAID controller busy waiting for them. Special RAID drives give up quicker, to the RAID controller can quickl recover data from the other two drives.

This is usually not a problem in an office environment. If the data is recoverable, the user doesn't care wether it comes after 50ms or 800ms. But in a realtime environment like a cinema server, the data has to be there in time, otherwise playback stutters/skips.

- Carsten

Posted by Monte L Fullmer (Member # 2797) on 07-03-2014, 04:39 PM:

Hello Carsten:

Thx for the very descriptive reply on the functions of a RAID system, for this could be the issue that we face from time to time.

On the identification of the drive itself, how is the drive described in being either a consumer based drive, or an enterprise drive - is it on the label or similar? I do believe the drives are Hitachi drives.

Thx for the help on this - Monte

Posted by Carsten Kurz (Member # 5396) on 07-03-2014, 04:55 PM:

If you have access to the label on the drive, throw the device type at Google and see what comes out. There are not many 2.5" enterprise drives on the market.

I remember that recently I did a backup from an old drive with some defect sectors. It took more than a minute until the system reported a sector as unreadable. While this is most likely a combination of drive AND software trying to read the sector multiple times, you can see what happens with response time when from a set of three drives one of them has a problem.

Now, this response time problem can be masked if the RAID system/controller uses a large buffer memory, reading ahead many seconds of data. But memory costs money - a DCP can have 30MByte/s, video and audio needs to be buffered separately, there is additional need for memory for the RAID stripping/parity process, etc.

- Carsten

Posted by Mark Gulbrandsen (Member # 72) on 07-03-2014, 07:38 PM:

The Hitachi drives used in the PDP-3000 are enterprise type drives. All my customers, close to 30 with SX-3000's leave their PDP's on 24/7. The thing doesn't draw much so it's inconsequential to leave them on. The SX-3000 talks to the raid in the PDP on a somewhat limited basis right now. Note also that the blue drive activity lights do not illuminate until until the SX-3000 has almost booted up... Hint hint...

Mark

Posted by Monte L Fullmer (Member # 2797) on 07-03-2014, 09:29 PM:

quote: Mark Gulbrandsen
Note also that the blue drive activity lights do not illuminate until until the SX-3000 has almost booted up... Hint hint..

Oh, already know all about that one-I see that all the time when I need to service these BARCOS and have to turn the power off.

Our PSD units reside behind the projector on a platform.

Had one drive crash in one unit a few months back - went into CTRL Panel and notice No.2 drive said "no" where the other two had "ok" by them along with No.2 light wasn't on.

Sent me a new drive and the NOC had to do the population after I installed the new one.

If they're enterprise drives, guess the SX-3000 just gets all grumpy at times causing the skipping/image freeze issues...which needs the total reboot.

quote: Mark Gulbrandsen
The SX-3000 talks to the raid in the PDP on a somewhat limited basis right now.

If so, when is the fix or patch coming around the corner to open up this communication to be on an unlimited basis?

thx-Monte

Posted by Mark Gulbrandsen (Member # 72) on 07-04-2014, 05:54 PM:

Monte...

quote: Monte L Fullmer
Sent me a new drive and the NOC had to do the population after I installed the new one.

Why, It's all automatic. You stick in a new drive and the raid goes right into rebuild. Not so on the larger Servers. You have to place the new drive into the raid array both physically ad in the admin panel and then it goes into rebuild. Sounds like your NOC is looking for work and it's pretty pointless in that aspect of the unit. As for drive failure, they are not any more failure prone than the 3.5" drives. The laptop drives generate much less heat too.

Hard to say on the RAID com thing. If you loose a drive the control panel page will show it as a degraded array. If the TMS is properly set up you'll also get a degraded array warning emailed to yourself. You also still have access to those drives in the Admin panel to look at sector read errors and such just as you do on the larger servers.

Are you running three or four drive raids? I recommend four drive.

Mark

Posted by Monte L Fullmer (Member # 2797) on 07-04-2014, 06:37 PM:

Hi Mark - three drive RAID unit.

I know, that using the four drive RAIDS are better .. Monte

Posted by Monte L Fullmer (Member # 2797) on 07-05-2014, 02:00 PM:

Drives in these PSD units here at my location are the HGST C5K-1000 1TB units. I have three of these drives in each of my units.

Made for cinema/video storage, but the HGST site doesn't list them neither as consumer or enterprise drives-but do have a fast transfer rate of 6k Mb/s, if that means anything.

So info to share - thx Monte

Posted by Steve Guttag (Member # 268) on 07-05-2014, 04:59 PM:

They are consumer drives and not very speedy ones at that:

quote:
CinemaStar C5K1000, which offers 1TB capacity in a 2.5-inch form factor, meeting the needs of a growing number of CE OEMs who wish to offer video recording and playback features in a compact design, while still offering greater capacity. At 5400 RPM, the 9.5mm, 2.5-inch CinemaStar C5K1000 is available in 1TB, 750GB and 640GB capacities and specifies a read/write power consumption as low as 1.5W and a quiet operation at 2.4 bels idle.

Probably a bit underspecified for cinema applications. All of the full-size cinema servers are using 7200RPM and faster drives.

I'm surprised GDC didn't go with a faster and more robust drive. Since their drives are external, they don't have the heat considerations that those that have IMS systems have to consider (Doremi, Barco and USL). Since they (wisely) went with external storage...take advantage of that fact and use fast/durable drives.

I still prefer the traditional servers (with the coax cables and internal mediablock as opposed to the "integrated" mediablock. The system is inherently more robust except for the effin' Enigma module imposed upon us by the paranoia of the industry over who is stealing what, how and how much is actually lost. Even for 4K, I'll take four-3G lines over the IMB concept with the audio moving to the freaking IMB.

Posted by Carsten Kurz (Member # 5396) on 07-05-2014, 05:55 PM:

quote:
An exceptional blend of audio/visual (AV) features fine tunes the CinemaStar C5K1000 for video storage applications. HGST SmoothStream technology supports the ATA-7 streaming command set, and SCT provides time-limited error recovery and thermal monitoring capabilities.

They are not exactly enterprise-grade drives, but have some optimizations for realtime media/RAID applications. They are certainly fast enough straight or in a RAID configuration for digital cinema, but the real-world reliability of SX-3000 and storage pod is a different thing and can not be judged from transfer rate or seek time.

The PSD-3000P devices are meant to be mounted to the projector or placed nearby, yet they feature non-locking consumer-grade connectors for power.

In my opinion, that is ridiculous given the effort for robust playout/operation taken in other areas of a cinema projection system.

GDC also offers the enterprise 19" mounted storage option utilizing 3.5" drives. Sure there is a reason they offer both variants. Well, capacity may be the prominent reason.

- Carsten

Posted by Steve Guttag (Member # 268) on 07-05-2014, 07:08 PM:

I'm thinking it is the "$$$$" or insert your country's symbol of currency.

As for IMBs...all of them should have powered themselves off of the storage system if it is external. Shutting down the projector should have no affect on server. If you want, if the projector power vanishes...the IMB can send the system into a hibernate or minimal power (heat) mode but powering it entirely from the projector is just plain bad design.

It seems that the various projector/IMS people are trying to provide workarounds to sudden shutdowns causing the IMS to commit suicide.

Posted by Mark Gulbrandsen (Member # 72) on 07-05-2014, 08:03 PM:

quote: Carsten Kurz
The PSD-3000P devices are meant to be mounted to the projector or placed nearby, yet they feature non-locking consumer-grade connectors for power.

The idea is to be able to easily move the arrays between theaters if you do not have a TMS to do it. And quite literally one can be moved in about 30 seconds and then the projector rebooted to reflect the different raid box. They are not common to just one server SN.

I disagree on the IMB being powered from the storage device. There is really nothing wrong with the way it is being done right now unless perhaps your projector does not have a standby mode... There is little point to being able to boot up an IMB on it's own since it relies on so much communication with the projector also being on just to function, ICP Com, Suckurity Manager, etc won't work without a running projector. The SX-3000 boots in well under 200 seconds so the wait time just isn't really even a factor there if you need to reboot... NEC projectors average 180 seconds boot time so the server comes up right after the projector does.

Mark

Posted by Monte L Fullmer (Member # 2797) on 07-06-2014, 12:18 AM:

True, reboot time for the SX-3000 isn't that bad at all - I'll give it that a plus sign.

From the time I close the breaker in the breaker panel to put power to the BARCOS, til the green light comes on below the TV screen on the TMS is between 90 to 120 seconds....which is a LOT better than rebooting a Dolby DSS200 server, which almost takes 5 minutes from when one turns back on the UPS, that the server, CP850 Processor and the USL monitor amps, til the green light is illuminated below the TV screen to let me try to reconnect to that assigned screen.

To add: We did have the PSD-300u, rack mount units with the 3.5 drives in them, but we had horrible issues with these units (mainly due to the three SATA cables running into the SX-3000 having connection issues since the 300u unit was just sitting on top of the projector, thus it was chosen to do the 300i upgrade with the single cable that can be locked into place on the IMB) where there were times I had to turn off the unit, open the bay door to reseat the drives, then power back up to get content to play normally .. and this was long before the instruction of proper procedure of how to do a correct reboot by shutting down the IMB via the "Shutdown" menu in the GUI, THEN, power off the projector either by the power switch or breaker.

-Monte

Posted by Carsten Kurz (Member # 5396) on 07-06-2014, 10:10 AM:

The PSD-3000u probably had the RAID software running on the SX-3000 itself. Strange concept. The enterprise 19" storage option contains a dedicated RAID controller, like the current PSD-3000P. With all these variants, it doesn't really look like a solid concept.

- Carsten

Posted by Mark Gulbrandsen (Member # 72) on 07-06-2014, 10:14 AM:

Monte... Yes, I had a dozen or so of the rack mount units at first. Some of them certainly had issues. GDC has since replaced them all...

Carsten... The actual raid controller is inside the PSD-3000. At a certain point in the boot sequence the SX-3000 wakes up the raid card and the blue drive activity lights come on so you know the drives are spinning. The drives do not spin up unless the SX-3000 wakes the raid card. There is also the usual software raid available in the SX-3000 if you prefer to run triple external 2.5 or 3.5" drives via separate eSATA cables. Or as you mentioned an external source either supplied by GDC or some other manufacturer. Frankly, I think the idea of these server/IMB's is great because they offer so many options for data input. There are really no hindrances between these vs. a full size server that I have encountered in my installations.

Posted by Carsten Kurz (Member # 5396) on 07-06-2014, 11:46 AM:

I have no problem with IMBs or IMSs, and not with ingest drives being attached through USB, eSATA, whatever. But the playout storage itself is so dependent on solid realtime response, I just doubt it's a good idea to do it the way Christie or GDC implemented it with their IMS. While internal drives may be a potential source of heat problems, they are better to control. Even just a slightly higher probability of hanging, skipping, stuttering is not acceptable in my opinion. There simply needs to be enough safe margin for uninterrupted playout.

- Carsten

Posted by Monte L Fullmer (Member # 2797) on 07-06-2014, 11:54 AM:

quote: Mark Gulbrandsen
The drives do not spin up unless the SX-3000 wakes the raid card.

This explains why I was told not to power down the PSD since it wasn't necessary when doing the reboots.

Looks like when powering down the SX-3000 through the GUI also shuts down the PSD unit, thus making as a whole system shutdown.

When the projector power return, it's also 'wakes' up the GDC system back to normal.

Thx for the info - now it makes more sense on the procedure.

-Monte

Posted by Steve Guttag (Member # 268) on 07-06-2014, 12:45 PM:

quote: Mark Gulbrandsen
I disagree on the IMB being powered from the storage device. There is really nothing wrong with the way it is being done right now unless perhaps your projector does not have a standby mode...

Bullshit...it is just plain stupid. If one shuts down the projector before the server, the server can go corrupt on both the GDC and Doremi. What a silly design. If the IMB was powered by the storage part, the server would essentially remain a separate entity. Then again, I'm not a fan of the IMB system to begin with.

Posted by Mark Gulbrandsen (Member # 72) on 07-06-2014, 03:07 PM:

Steve...

It has never personally been a problem for me or my customers where things become corrupt on the SX-3000. What's there to become corrupt? The data the server has just sent to the IMB to be decrypted? That's about it! It's still fine on the media drives as is the OS! A reboot of both devices takes care of any data connection issues. In fact when you power down an NEC projector with an SX-3000 on board the projector almost always goes down first. For sure the ICP does. Lately, I also have not found it to be a problem on the larger servers either. I believe that GDC has taken care of this sort of thing in SA-SX firmware updates which are likely based on what the SX-3000 is doing. You can teach your customers the shutdown sequence till your blue in the face but someone is going to forget that at some point in time and do it the wrong way anyway. I can't speak for Doremi and their firmware. It is highly likely a huge issue for Qube since they are Windows based. Nor can I speak for how an SX-3000 behaves in a Barco or Christie. But it is a non issue for me. My customers have been powering up their NEC projectors and SX-3000's and powering them back down daily without shutting the server down first for well over a year with zero issues at all. They just flat out work.

Mark

Posted by Steve Guttag (Member # 268) on 07-06-2014, 03:48 PM:

I never had a problem either...but GDC absolutely insists that the server be shut down BEFORE the projector (with an IMB) be shut down or it will become corrupt. My systems are SX-2000ARs and that may be a difference.

I agree, it is a stupid design to depend on people to do things in a certain order. We do have some systems where the automation takes care of it (boot up and shutdown in the right order)...but I would rather leave the servers up 24/7...with or without the IMB being alive.

Posted by Mark Gulbrandsen (Member # 72) on 07-06-2014, 04:59 PM:

Yes, they used to insist on it... And there may be reasons even I am not aware of to still do it but not doing it has not been an issue...

Mark

Posted by Monte L Fullmer (Member # 2797) on 07-06-2014, 11:07 PM:

Okey.. here we go (again), but this time Dolby shows its ugly face with a DSS200 and a CAT745 IMB in a Barco 32B:

Same thing: right in the middle of "Trans 4:Age", feature freezes solid.

Told the assistant manager to reboot the Dolby with shutting down the UPS that it's receiving power from and turning it back on and wait almost 10 minutes for a full reboot. While waiting for a reboot on the Dolby, go kill the power to the projector in the breaker panel. Wait a few seconds and power up the projector.

They were back on screen in 10 minutes ... and I just did a full reboot three days earlier.

What gives, invaded by gremlins or similar?

-Monte