|
|
This topic comprises 4 pages: 1 2 3 4
|
|
Author
|
Topic: GDC SX2001 RAID HDD - Compatible drives?
|
|
|
|
|
|
|
|
|
Leo Enticknap
Film God
Posts: 7474
From: Loma Linda, CA
Registered: Jul 2000
|
posted 02-27-2018 01:55 PM
Would the power saving over the service life of the drives offset the higher cost of buying them in the first place? At the moment at least, the cost of larger SSDs is so much more than their HDD equivalents that I'd be surprised if it would (even with California electricity prices!).
It's like when I still had the '99 Civic, friends and relatives were urging me to replace it on the grounds that a 20-teens new one would save me $1-2k a year on gas. That it would have done, but the '99 was, by that time, costing us almost nothing in depreciation (with 200k miles on the clock, it had already depreciated to a book value of near zero), compared to around $3k a year on a brand new one.
Unless the price of SSD storage comes down a lot, they're never going to make sense for large-scale storage applications, unless very rapid random access is absolutely essential, as Scott notes. Smaller ones for OS drives make a lot more sense, because the speed benefit is actually useful.
For DCP server storage, I suppose they could give you an ingest speed advantage if the ingestion is through a super-fast fiber network, but from physical media you're going to be constrained by the limits of USB3 anyways.
| IP: Logged
|
|
|
|
Carsten Kurz
Film God
Posts: 4340
From: Cologne, NRW, Germany
Registered: Aug 2009
|
posted 02-27-2018 05:37 PM
DCP playback is not a long continuous read operation - the server has to read at least two fully separate streams of data, typically more like 5-6, at least that is what the RAID has to support worst case, with subs, on reel changes. And that is for 'simple' CPLs only.
I don't know why USL chose SSDs, but maybe, as they have no closed engineering links with projector manufacturers, they evaluated discs and SSDs, and found the energy/heat budget easier to manage for an integrated server. Maybe they also thought that potential buyers would instantly jump onto the assumed higher reliability factor of SSDs vs. spinning discs. Keep in mind, that the USL IMS allows to render MDA from local storage, so it needs a higher bandwidth as well. Plus, at the time the USL was designed, HFR seemed to have a happy future. So, you had to design at least 500MBit/s for the video part alone, then add audio, subtitle, and MDA audio demands both in bandwidth and seek times.
- Carsten
| IP: Logged
|
|
|
|
|
|
|
|
Marcel Birgelen
Film God
Posts: 3357
From: Maastricht, Limburg, Netherlands
Registered: Feb 2012
|
posted 03-01-2018 01:54 AM
Indeed, if you're running an x-plex (where x is a significant number and the average number of seats per auditorium is also a significant number ![[Smile]](smile.gif) ) and you decided to centralize everything to this single room, then this becomes very much a datacenter like operation.
I guess the only reliable way to serve e.g. 14 screens simultaneously from the same storage, would involve a lot of SSDs in at least a RAID10-like setup. Even then, a single snafu on the network or storage side will bring down all those screens at once.
quote: Carsten Kurz
DCP playback is not a long continuous read operation - the server has to read at least two fully separate streams of data, typically more like 5-6, at least that is what the RAID has to support worst case, with subs, on reel changes. And that is for 'simple' CPLs only.
That's where many indeed go wrong. While most media files use interleaved formats, where all or most assets are interleaved in one continuous stream, DCP does not. As such, a DCP is far more intensive to play than e.g. am MJPEG compressed AVI file.
Still, since the separate assets themselves are mostly sequential, an intelligent RAID controller combined with some semi-intelligent firmware on the disks can pre-fetch a lot of the data, reducing the amount of seeks needed.
quote: Mike Blakesley
The last time it happened we thought it was the drive box, but replacing that didn't cure the problem. The drives themselves were the issue. We've been digital for 8 years as of June and I don't think we've ever had a set of drives longer than 3 years.
I mentioned it before. It's a pity we're using "industry standard" RAID solutions for cinema playback servers. It's obviously a cost-saving measure, but those RAID solutions are built for data-integrity, mostly targeted at office applications.
So, instead of ejecting a lagging drive as erroneous, the RAID tries to recover the problem. This behavior is fine in a normal office environment, where e.g. somebody needs to wait a few seconds longer before he can open his Word document, but creates a problem in a real-time playback situation, where the data is now past-due.
| IP: Logged
|
|
|
|
|
|
|
|
Leslie Hartmier
Expert Film Handler
Posts: 100
From: Edmonton, Alberta, Canada
Registered: Jul 2012
|
posted 03-01-2018 07:13 PM
The only two drives that GDC has qualified are:
HGST 2TB HUS726020ALE610
HGST 1TB HUA722010CLA330
However, I use WD Gold (Datacenter) drives, as the HGSTs are hard to source in Canada except via online (and honestly, the price for datacenter drives for me are about the same as the NAS/RAID drives anyway, with a better MTBF, MTTF and URE rate).
As for SSD RAIDs, though old SSDs had firmware that did not permit use in RAIDs, all SSDs that I know of now do permit their use. Benefits include no spin-up/spin-down, faster reads/writes, etc, but there is one thing to note if you choose to use SSDs - SSD lifespans are rated in writes, and if one starts to go sideways on you, the others are neck and neck in doing so as well.
Most SSD RAIDs created are not done in a straight-forward way -
1) totally different brands from each other are used to create it so that the lifespan will more likely vary instead of all drives failing at about the same time,
2) each drive will be put in the RAID at significantly different times so that if one drive fails, there is a lower chance of more than one failing at the same time, or
3) all drives are simply replaced simultaneously at regular intervals approximating the lifespan of the drives based on expected number of writes.
Avoiding RAID rebuilding is best as rebuilding is a lot of writes, which reduces lifespan of SSDs.
| IP: Logged
|
|
|
|
All times are Central (GMT -6:00)
|
This topic comprises 4 pages: 1 2 3 4
|
Powered by Infopop Corporation
UBB.classicTM
6.3.1.2
The Film-Tech Forums are designed for various members related to the cinema industry to express their opinions, viewpoints and testimonials on various products, services and events based upon speculation, personal knowledge and factual information through use, therefore all views represented here allow no liability upon the publishers of this web site and the owners of said views assume no liability for any ill will resulting from these postings. The posts made here are for educational as well as entertainment purposes and as such anyone viewing this portion of the website must accept these views as statements of the author of that opinion
and agrees to release the authors from any and all liability.
|
Home
Printer-friendly view of this topic