enterprise grade
-
- zen recordist
- Posts: 8876
- Joined: Mon May 19, 2003 12:10 pm
- Location: NYC/Brooklyn
- Contact:
enterprise grade
So I just brought this up in another thread, and I thought it might benefit a few people.
I have my own server setup thee days for Studio G Brooklyn, and it is working quite well. I went through a bunch of hardware and software before I landed on an Xserve with enterprise grade drives in it.
I didnt believe in "enterprise grade" until being the tech/admin of my own 50 client file server, with streaming for iOS devices and a bunch of other features specific to our needs... like our very own dropbox, sort of.
Anyway...
Then I got hip to using these drives and investigated what made them slightly more pricey than the consumer grade drives of the same capacity.
It turns out that there are a LOT of differences in the tiolerances, fault recovery, heat dissipation, and overall performance. They are made to be ON, forever. Really. Like 24hrs a day, 365 days a year... for years of streaming and read write operations constantly.
Now I put one in my protools rig , which is used almost EVERY single day, and I am amazed at how solid things are. The startup drive is now mirrored, just like the server, in a hardware controlled RAID1 setup... this makes the rig as solid as an air traffic control setup... its really good.
Looking into RAID 1+0 or 0+1 situations for printing and mixing audio, I realized that just having a 1TB enterprise toshiba in my print path was perfect, with alll the manual and automated backup stuff my studio has these days... the server is backed up automatically every hour to a 5TB Drobo setup, which is a RAID 5 array.
very smart.
Anyway: if you have ever lost a bunch of stuff and had to explain it to a client, think about the fact that you are putting a lot of hours/days/dollars worth of work on a consumer grade drive. I did, and i have been very happy with the switch to a slightly pricier drive.... but even if you are getting paid next to nothing for your services, and you are put in the position of keeping someone elses data safe after tracking... take a look at OWC's deals on enterprise grade drives. The toshiba's play nice with Mac.... and they really are a small price to pay for solid performance day after day.
I have my own server setup thee days for Studio G Brooklyn, and it is working quite well. I went through a bunch of hardware and software before I landed on an Xserve with enterprise grade drives in it.
I didnt believe in "enterprise grade" until being the tech/admin of my own 50 client file server, with streaming for iOS devices and a bunch of other features specific to our needs... like our very own dropbox, sort of.
Anyway...
Then I got hip to using these drives and investigated what made them slightly more pricey than the consumer grade drives of the same capacity.
It turns out that there are a LOT of differences in the tiolerances, fault recovery, heat dissipation, and overall performance. They are made to be ON, forever. Really. Like 24hrs a day, 365 days a year... for years of streaming and read write operations constantly.
Now I put one in my protools rig , which is used almost EVERY single day, and I am amazed at how solid things are. The startup drive is now mirrored, just like the server, in a hardware controlled RAID1 setup... this makes the rig as solid as an air traffic control setup... its really good.
Looking into RAID 1+0 or 0+1 situations for printing and mixing audio, I realized that just having a 1TB enterprise toshiba in my print path was perfect, with alll the manual and automated backup stuff my studio has these days... the server is backed up automatically every hour to a 5TB Drobo setup, which is a RAID 5 array.
very smart.
Anyway: if you have ever lost a bunch of stuff and had to explain it to a client, think about the fact that you are putting a lot of hours/days/dollars worth of work on a consumer grade drive. I did, and i have been very happy with the switch to a slightly pricier drive.... but even if you are getting paid next to nothing for your services, and you are put in the position of keeping someone elses data safe after tracking... take a look at OWC's deals on enterprise grade drives. The toshiba's play nice with Mac.... and they really are a small price to pay for solid performance day after day.
- EarlSlick
- takin' a dinner break
- Posts: 153
- Joined: Thu Jun 17, 2004 7:10 pm
- Location: Chicago
- Contact:
The main difference in enterprise level drives, has little to do with quality, tolerances, or performance. The main difference is fault recovery.
On a desktop drive, if a sector cannot be accessed, the hardware controller on the drive will keep trying to access that data up to a timeout. Most desktop drives I've encountered set that timeout to 2 minutes, which is rather high. The drive is made to do everything it can to access that data, since in a desktop environment, there is only one copy of that data.
Until recently all desktop drives, had a manual override to shorten that timeout, but as of 6-8 months ago, this feature has almost completely disappeared.
Enterprise drives, which would expect to be run in a raid array, do not exhibit this behavior. If a sector cannot be read in short order, the raid controller knows to just try the alternate location for that data.
That being said I've run 100+ TB data centers on only desktop drvies, with no problems.
On a desktop drive, if a sector cannot be accessed, the hardware controller on the drive will keep trying to access that data up to a timeout. Most desktop drives I've encountered set that timeout to 2 minutes, which is rather high. The drive is made to do everything it can to access that data, since in a desktop environment, there is only one copy of that data.
Until recently all desktop drives, had a manual override to shorten that timeout, but as of 6-8 months ago, this feature has almost completely disappeared.
Enterprise drives, which would expect to be run in a raid array, do not exhibit this behavior. If a sector cannot be read in short order, the raid controller knows to just try the alternate location for that data.
That being said I've run 100+ TB data centers on only desktop drvies, with no problems.
- Randyman...
- takin' a dinner break
- Posts: 186
- Joined: Sun Mar 04, 2012 8:30 pm
As does Google IIRC. They had an interesting reliability report that encompassed their masses of consumer 3.5" HDD's. Basically, as long as they last beyond 4-6 months (infant mortality), they should be good for 2+ years - but I never bet on single drives or single copies of my data. While I didn't read the Google report (only a recap of it) - I am curious how they are dealing with the lack of TLER with these consumer drives in their data centers?EarlSlick wrote: That being said I've run 100+ TB data centers on only desktop drvies, with no problems.
I recently upgraded my DAW's Areca RAID-5 with 2TB Hitachi Enterprise Drives (2011 pre-WD versions) only because they were CHEAPER than the 2TB WD Reds I was looking at, and the Hitachi's perform WAY better while still offering the "RAID Friendly" TLER Earl mentions.
I would not put any consumer drive that does not support TLER in any important Hardware RAID Array - bar none (software RAID is likely more friendly with non-TLER consumer drives).
In any event - RAID IS NOT A BACKUP! Back up that data people. Copy it to as many physical drives as you can! Over here, Important Data is stored on two independent hardware RAID-5 arrays, a hardware RAID-6 array, a RAID-6 ReadyNAS, internal single drives, and two sets of "offline" single drive archives spread across 2 locations. If your data is important, treat it accordingly IMO...
Randy V.
Audio-Dude / Musician / PC Guru / Crazy Guy
Audio-Dude / Musician / PC Guru / Crazy Guy
- Scodiddly
- speech impediment
- Posts: 4013
- Joined: Wed Dec 10, 2003 6:38 am
- Location: Mundelein, IL, USA
- Contact:
"RAID" mean "Redundant Array of Inexpensive Drives", so really the drives don't have be that great to get the benefits. What you do need to do is have spare drives, know how to replace a bad drive (hardware *and* software), and especially be paying attention to know when the RAID controller starts complaining.
Offsite backups are also important - our "new"server (installed 2010, which replaced a similarly overspecced beast from Dell that did its job for a full decade) puts nightly full backups onto hard drive cartridges.
Offsite backups are also important - our "new"server (installed 2010, which replaced a similarly overspecced beast from Dell that did its job for a full decade) puts nightly full backups onto hard drive cartridges.
It's all about redundancy and the time to recovery
It's a given that one must have multiple copies of any data.
The problem comes down to:
1. How often the various copies are synchronised.
2. How quickly a copy can be retrieved.
The optimal -- and most expensive -- are:
1. Simultaneous writing.
2. 'Instant' failover.
A related issue is how secure one wants the data. The solution is having the data in multiple sites that won't be wiped out by the same catastrophic event.
Basically, the best measures are to use full redundancy at component, sub-system and system levels, across multiple sites.
With even significant financial enterprises skimping on such levels of redundancy, is it surprising that small DAW studios don't do it?
The problem comes down to:
1. How often the various copies are synchronised.
2. How quickly a copy can be retrieved.
The optimal -- and most expensive -- are:
1. Simultaneous writing.
2. 'Instant' failover.
A related issue is how secure one wants the data. The solution is having the data in multiple sites that won't be wiped out by the same catastrophic event.
Basically, the best measures are to use full redundancy at component, sub-system and system levels, across multiple sites.
With even significant financial enterprises skimping on such levels of redundancy, is it surprising that small DAW studios don't do it?
Comp: i7 920, GA-EX58-UD4P, 12GB, ATI2460x2, SSDx5, UAD-2Q, Dell S2340Tx2+3008x2, TV.
HW: U87 Ai, JA251x2, YRG, padKontrol+microKEY2, FF400+800, M-Patch-2, Tannoy RA+TS-12.
SW: W8P64, Cubase 7.0.5 64, samplers & plugins, Ozone5A, RX2A.
HW: U87 Ai, JA251x2, YRG, padKontrol+microKEY2, FF400+800, M-Patch-2, Tannoy RA+TS-12.
SW: W8P64, Cubase 7.0.5 64, samplers & plugins, Ozone5A, RX2A.
- Nick Sevilla
- on a wing and a prayer
- Posts: 5593
- Joined: Mon Mar 03, 2008 1:34 pm
- Location: Lake Arrowhead California USA
- Contact:
I long for the day Pro Tools can be told to write data on two separate hard drives at once.
Surely that cannot be impossible to implement...
When I am recording a Major artist, I have two multitrack recorders running at the same time.
Either Pro Tools plus a second ProTools rig,if less that 16 tracks at once, or a Tascam Mx2424 for up to 24 tracks, synched to ProTools so the timeline stays the same.
Sometimes I used to run. DAT tape off the mixbuss, and just have it recording most of the time, to get off takes, commentary, or the odd noodling which was useful.
Surely that cannot be impossible to implement...
When I am recording a Major artist, I have two multitrack recorders running at the same time.
Either Pro Tools plus a second ProTools rig,if less that 16 tracks at once, or a Tascam Mx2424 for up to 24 tracks, synched to ProTools so the timeline stays the same.
Sometimes I used to run. DAT tape off the mixbuss, and just have it recording most of the time, to get off takes, commentary, or the odd noodling which was useful.
Howling at the neighbors. Hoping they have more mic cables.
There are eSATA RAID1 external cases. Put a couple of SSDs in one and it should keep up. If your system does not have eSATA, just get a SATA to eSATA adaptor cable. Use one with hot-swap drives to be able to change a failed drive without downtime.Nick Sevilla wrote:I long for the day Pro Tools can be told to write data on two separate hard drives at once.
Comp: i7 920, GA-EX58-UD4P, 12GB, ATI2460x2, SSDx5, UAD-2Q, Dell S2340Tx2+3008x2, TV.
HW: U87 Ai, JA251x2, YRG, padKontrol+microKEY2, FF400+800, M-Patch-2, Tannoy RA+TS-12.
SW: W8P64, Cubase 7.0.5 64, samplers & plugins, Ozone5A, RX2A.
HW: U87 Ai, JA251x2, YRG, padKontrol+microKEY2, FF400+800, M-Patch-2, Tannoy RA+TS-12.
SW: W8P64, Cubase 7.0.5 64, samplers & plugins, Ozone5A, RX2A.
Who is online
Users browsing this forum: No registered users and 9 guests