Film-Tech Forum ARCHIVE: Dolby DSS220: ingestion speed through eSATA

This is topic Dolby DSS220: ingestion speed through eSATA 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=001289

Posted by Antti Nayha (Member # 4980) on 11-24-2012, 12:08 PM:

I was told that when you plug in an external CRU dock to Dolby DSS220’s eSATA port, the ingestion speed from a CRU drive is very slow – comparable to ingesting through USB (which, of course, defeats the purpose of getting a CRU dock in the first place).

Can anyone confirm that this is indeed the case? And if so, is there a fix available? It seem like a deal-breaker to me in a single-screen setup without a TMS.

Posted by Richard May (Member # 2627) on 11-24-2012, 12:50 PM:

I doubt it. We have Doremi servers. We just added an ESATA port. We injest from CRU drives and it's MUCH faster than going USB. I don't know why Dolby would be different.

Posted by Pete Naples (Member # 761) on 11-24-2012, 04:25 PM:

On DSS220s it's painfully slow going eSATA.

I've observed USB is actually faster!

Posted by Steve Guttag (Member # 268) on 11-24-2012, 05:39 PM:

I have observed DSS220s faster through eSATA than USB...notably faster actually. Then again, there is no doubt, the DSS220 is SLOW on ingest, no matter which method...though FTP seems to be more reasonable via a TMS. Note, with eSATA you can still play directly off of the drive (done it).

Posted by Antti Nayha (Member # 4980) on 11-24-2012, 07:52 PM:

OK, thanks for the info. That’s a bit weird though – surely the three-drive RAID alone shouldn’t cause such a bottleneck…

Posted by Steve Guttag (Member # 268) on 11-25-2012, 12:20 AM:

A 3-drive RAID combined with the software RAID really took a hit on the speed department. It also means that like other software RAIDS...matching drives (read/write) is more critical to performance. I haven't done a speed test on FTP transfers on the DSS220 but the DSS200 could "smoke" the others in a transfer contest...my guess is that the DSS220 will be as slow or slower than the others now.

Posted by Brad Miller (Member # 2) on 11-25-2012, 01:40 AM:

I've only said it a thousand times:

Software raid = [puke]

3 drives instead of 4 = more [puke]

I will be conducting some speed tests soon and will report the findings, although I do expect all of the software raids to be somewhat equal in their slowness.

Posted by Nerijus Marmokas (Member # 6712) on 11-26-2012, 03:13 AM:

I have heard fro sites in my cinema circuit that some contents are ingesting faster and the others slower (on DSS220) using eSata (not because different size of the content). Haven't done a proper "research" yet, but i probably depends on HDD or even content characteristics, somehow...

Posted by Wolfgang Woehl (Member # 6901) on 11-26-2012, 06:17 AM:

Yes, Brad, you go do that. You might be in for a surprise or two re. software RAIDs.

On the other hand: I wouldn't be surprised at all if Dolby had messed up their implementation/interfacing to the perfectly able software RAID their kernel provides. They seem to be rushing product to market these days. For example their Cat. 745 is beta at best -- Of course without them telling their clients who put up good money for it. Without looking too hard we found 2 showstopper-class/dark screen bugs on it (Dolby confirmed, fixes in the pipeline but only at Q1 2013 earliest).

Wouldn't know about the DSS220. But does the above make me feel good? Like in I'd reccommend Dolby products to my clients? You can bet it doesn't.

Posted by Steve Guttag (Member # 268) on 11-26-2012, 10:25 AM:

I wouldn't consider Dolby rushed at all...if anything...they took their time on delivery (missed many deadlines). The ones we have put in have been near flawless (CAT 745/DSS220). My biggest issues have been interfacing things like the TMS (Cinedigm) which required 4.5.3 software due to some naming conventions. If you want to talk about Beta testing in the field...Dolby has been the best for me for NOT doing that. I have definitely been hit by "others" that have had less than reliable product.

Posted by Wolfgang Woehl (Member # 6901) on 11-26-2012, 12:26 PM:

Steve, The Cat. 745 is clearly not even approaching "flawless". If your impression is "near flawless" you've been lucky so far and that is all. Let's talk again once you get a call from a client with perfectly valid content (DCP or KDM) and a dark screen.

I mean, instead of jumping in here for Dolby you could simply ask what the 2 bugs are, right?

Apart from that: The original poster's question about the Dolby DSS220's ingest speeds and related comments kind of prove my point anyway.

Posted by Steve Guttag (Member # 268) on 11-26-2012, 12:31 PM:

If you wanted to state what the bugs were...you would have already. I can say that we have had them in Barco and Christie projectors, thus far and they have NOT posed an issue despite months of use (we had one go in the week they were released). For them to be as buggy as you claim, you'd think we would have hit upon it. The issues I've had seem to be more projector related (one brand of projector has had issues, the other, not so much...I have since updated the software in the projector in question and, thus far, have not had issues. Note, this projector is also a 4K projector, the others have been 2K.

Posted by Mattias Mattsson (Member # 4308) on 11-26-2012, 06:52 PM:

Cat. No. 745 with v4.5.3(5) chokes on DCP:s containing a mix of reels that are both encrypted and non-encrypted. I.e. the CPL as whole is requiring a KDM, but one or more reels are unencrypted.

This is perfectly legal according to spec and actually quite common over here as local distributors add tagged logos etc.

Posted by Steve Guttag (Member # 268) on 11-26-2012, 09:40 PM:

I believe you but I have not seen it.

Posted by Brad Miller (Member # 2) on 11-27-2012, 12:58 AM:

quote: Wolfgang Woehl
Yes, Brad, you go do that. You might be in for a surprise or two re. software RAIDs.

Wolfgang, all "mission critical" servers have hardware raids. I've never heard of any that use software raids. I also tend to consider the playback server a mission critical item, as if there is a problem, you are refunding the house. The difference in price is minimal, such that once you refund the first show, you could've paid for the hardware raid.

Have you ever heard the term save a dime to spend a dollar? That's exactly what software raids have done for me. Granted Doremi's software raid is pretty stable, but GDC's has proven to me to be an absolute joke with stability.

Also I have seen Dolby (200 and 220), Doremi and GDC on their ingestion speed differences and it will surprise you how much faster the hardware raid is. The reason I don't post just yet is because I want to give some specific examples with and without other network traffic at the time to keep it fair. I also plan on comparing speeds when idle vs. during playback.

Posted by Wolfgang Woehl (Member # 6901) on 11-27-2012, 02:55 AM:

Comparing vendor implementations and interfacing is one thing (and useful, too). A dataset on software RAIDs' principal capabilities would be something very different. And here's me, thinking the general claim was about software RAIDs [Smile]

Posted by Marco Giustini (Member # 4544) on 12-02-2012, 06:53 AM:

DSS220 ingest speed when playing is *very* slow, did not take any speed measure, but my USB drive is barely used when playing back and ingesting at the same time.

All IMBs look a bit raw at the time being. Dolbywise, 4.5.0 was just to have the system working, but you couldn't stop/pause/skip within the last 5 seconds of any clip, or the IMB would crash. 4.5.1 is a "working in progress", apparently the crash has been addressed in an easy way: the 745 wouldn't accept any commands during the last 5 seconds of any clips (guess how long is my personal framing chart? 5 seconds, hence I cannot pause it!).
4.5.3 is the only one working with encrypted content.
Again, I'm with Steve here, it's pretty normal for a new piece of equipment, but I wouldn't define it as "flawless" though!