Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Storage Strategy for 3 Tb Media Server (Read 2043 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Storage Strategy for 3 Tb Media Server

Hello forum.

It has been four year since I got my music server and it is time for an upgrade.

It is a linux machine with 4x 120Gb ATA in a software RAID5 configuration.
I also have a backup linux machine with 3x 120 GB ATA in software RAID0.

The RAID5 machine has been nothing short of a nightmare. One time I experienced a disk failure (not a complete disk crash). I replaced the disk and was quite happy.
However three times I experienced a controller flaw. A controller flaw takes two disks offline, thus breaks the array and all data is lost.
The first time I was lucky and had a very recent backup. After a reboot the controller worked fined again and I rebuilt the array and restored data.
The second time was a disaster. I lost a month worth of data, as I was too busy to make backups. I replaced the entire machine (short of the disks) with my workstation, a machine I knew was dead reliable.
The third time I did not loose any data, but my attitude towards linux software raid got somewhat negative and I panicked for some time. The problem was probably not be the controller (it was replaced), but the disks (or one of the disks) that did something that provoked a controller failure.

The backup server runs without any problems, but it only stays on a few hours every week. It does however take up a lot of space and makes a lot of noise. So a week ago I bought an external USB2.0 Seagate 750 Gb disc. Marvellous little thing. Stops spinning when idle, and thus saves power. Maybe it will last longer by doing this?

For my new media server I would like some 3 Tb space mainly for FLAC files and DVD rips. Speed is not all important, but data-loss is. There are several configurations I can think of:

Software RAID5 with 5x 750 Gb SATA disks.
Hardware RAID5 with 5x 750 Gb SATA disks.
Four seperate internal SATA 750 Gb disks.
Four seperate external USB2.0 SATA 750 Gb disks.
Six seperate external USB2.0 SATA 500 Gb disks.
Ten seperate external USB2.0 SATA 300 Gb disks.

Unfortunately I do not have enough money to buy a 3 Tb backup device. The original DVDs will have to do as a backup. But it takes a very long time to restore 750 Gb of DVD data.

RAID5 is an obvious solution. However my bad experiences with controller failures and complete raid failure draws in the other direction. But SATA is not ATA. Maybe a controller failure for one disc does not affect other discs? Loosing 3 Tb of data with only DVDs as backup would be a disaster.

Using seperate disks is way to minimise data-loss. If a disk fails without crashing completely or the controller dies, then only a little data is lost. That data can easily be retrieved from DVD.
But a disk can crash completely. I have tried this with the 60-90 Gb IBM Desk Stars. Loosing 750 Gb of data is still a lot. Using smaller disks will minimise this problem.

And finally: It is clever to allow the discs to spin down? If I use seperate discs, then the movie discs would be idle for most of the time. The power bill is large enough and if the spinning down extends the life of the disk?

Input needed.

Thanks for readin.

 

Storage Strategy for 3 Tb Media Server

Reply #1
Some thoughts:

Historically, I've found that add-on IDE controller cards (such as the promise ATA-100/ATA-133 PCI cards) tend not to handle disk errors very well.  This leads me to believe they should be avoided, *especially* for soft-RAID.  However, they also tend to be the ones that were most often used for soft-RAID.

[This is odd, because some on-board controllers are also by Promise.  So, who knows...]

I suppose you have two directions to go when dealing with storage arrays:  build it yourself, or have someone else build it.  Then you have question of whether or not you want it to be local storage or shared storage (e.g. a network file server).

I don't have any self-built Soft-RAID arrays.  I'm not sure I trust them, or rather more specifically, I'm not sure I trust myself to manage them well.  I prefer the array to appear to the OS as a single disk.

A couple years ago, I put together an old Areca 5xPATA<->PATA RAID-5 controller (with built-in disk trays) in an external SCSI-style box, with a USB/Firewire bridge board.  At the time, the ARC-5010 controller was available direct from Areca's US office for a significant discount. That's worked flawlessly.  So, I trust their hardware.  They make both standalone as well as PCI-card-based RAID controllers and the recent ones include both RAID-5 and RAID-6 (double redundancy for the latter).  Their 8-port and 12-port SATA RAID cards are pretty darn nice deals *and* have been well received by reviewers.

OTOH, I had a PCI Nextcell 5xRAID-3 PATA controller card.  That worked fine for about a year, then one day, a drive died and suddenly the entire array wasn't recognized, and all the data wasn't recoverable.  And of course, they went out of business the same month, which meant their forum was gone.

In terms of home network storage...personally, I've been very happy with the Infrant ReadyNAS devices.  Four bays, SATA.  Current max is 2.25TB with redundant RAID, or 3TB with RAID-0 configurations (I do not recommend RAID-0, however) - more when 1TB drives are available.  It's not meant to replace a high-end file server, but it's easy to manage and until recently was faster than all of the consumer-level NAS appliances.  Plus the new NV+ units are small, quiet and built like tanks (or at least, dense enough to feel built that way...).

Edit - addition: the old-school answer was "run continuously, disk spin-down/spin-up is stressful"; the new-school answer for current tech is "spin-down/spin-up a few times a day will never exceed the rated spin-down/spin-up cycles within the lifetime of the drive, assuming the other physical tolerances aren't exceeded (e.g. temperature/voltages) - however, the choice is up to you"

And finally:  RAID redundancy isn't meant for disaster recover, it's mean for uninterrupted uptime.  Even with RAID, backups are required.

-brendan

 
SimplePortal 1.0.0 RC1 © 2008-2021