Urgent information for Seagate drive users.

[smash]

Hey Guys,

For those who aren't aware, there is a firmware fault with certain Seagate HDD's - causing them to spontaneously disappear.

Just had a client lose an 8 TB RAID... and an 8TB backup...

Supposedly, the data is recoverable from a single drive, but in our case we are yet to see if the recovered Data will enable the RAID to be rebuilt. It was a RAID 5 system, there were multiple, near simultaneous, drive failures.

I would strongly recommend checking out whether your drives may be affected here:

https://apps1.seagate.com/rms_af_srl_chk/


Smash.

___________________
Simon "Smash" Ashby
I share fruit over Fibre

[Kurt]

Ouch!
I don't trust IDE raid setups, a mates computer had both drives in one fail. I replaced the drives and they failed again within a couple of months. At least with single drives there is a fair chance of recovering some data, much lower with raid I suspect.

Just had a client lose an 8 TB RAID... and an 8TB backup...

[musikwerks]

The whole purpose of a RAID system is to provide data safety. You have zero chance of getting your data back if a single drive fails. With a RAID 5 the data is written accross all 8 drives (in my case it's 8 drives) so that if one fails you can pop in a new one and the data that was on that failed drive can be reclaimed from the remaining 7. You run into trouble when you have more than one drive fail simultaneously...which is incredibly bad luck and not that common.

Everything I do is RAID 5, minimum. I take a speed hit to do that and more so for RAID 6 but it's well worth it when I have 4 days of work sitting on the drives that needs to be delivered on the 5th day.

[Kurt]

If it's an 8TB raid then it was probably raid 0 (for speed) One disk dies, it's all gone.

The whole purpose of a RAID system is to provide data safety. You have zero chance of getting your data back if a single drive fails. With a RAID 5 the data is written accross all 8 drives (in my case it's 8 drives) so that if one fails you can pop in a new one and the data that was on that failed drive can be reclaimed from the remaining 7. You run into trouble when you have more than one drive fail simultaneously...which is incredibly bad luck and not that common.

Everything I do is RAID 5, minimum. I take a speed hit to do that and more so for RAID 6 but it's well worth it when I have 4 days of work sitting on the drives that needs to be delivered on the 5th day.

[musikwerks]

I seriously doubt it. If they're doing any ind of video work then it would be RAID 5 or 6. No one in professional post production would dare risk a RAID 0 set up.

[JulienG]

...No one in professional post production would dare risk a RAID 0 set up.

Sure, but in production it's a little more common as RAID-0's for a given data rate are smaller then RAID-5's (or 6's).

Personally I love my NetApp filers, but they're insanely expensive (~$15/gig, although it goes down for 20+ TB) compared to some of the smaller stuff out there.

The Drobo's are nice, but I'm waiting for them to have a better network interface then iSCSI.

[musikwerks]

Sorry Julien.... those first two paragraphs made zero sense to me..... do I just need more coffee???

A RAID 0 will have a bigger disk allocation size than a RAID 5 or 6 because the failure protection built into a 5 or 6 takes extra space from the array to do it's protecting. As an example, my 8Tb arrays have 8Tb total drive space but in RAID 5 we can only use 6.37Tb .

As for cost, again, my 8Tb array, including case, RAID controller card, cables and drives cost $2 per gigabyte, and I'm getting speeds averaging 450MB/s write and 550MB/s read. No sweat for uncompressed HD video and not even blinking for multi-track HD audio.

[smash]

Allow me to temporarily put on the IT hat.

In smaller raid systems (say 4 disks) there is a definite speed boost in running in RAID 0. Obviously, the disadvantage is that if you lose a drive, you lose all your data.

In larger RAID systems, the bottleneck is actually the interface itself, not the RAID level. We do RAID 6 systems running at 1.4TB per sec, over Quad 4Gb Fibre. In these cases, the limit is the bandwidth of the Fibre, not the bandwidth of the drives.

As for being unlucky to see multiple drives fail, It happens. Also, the more drives you have, the greater the chance that drive will fail.

In a 16 bay, we would ideally configure it as 3 x 5 disk Raid 5 Arrays, and a hot spare. This minimizes the chance for multiple drive failures to kill a system.

Also, with any RAID system, remember that you are only protected against hardware failure, not human error. It's never fun explaining to a client that all there data is gone because their disgruntled employed did a secure erase of 3 drives.

If all your Data is in one place, or simultaneously online, than it isn't backed up.

_________________________
Simon "Smash" Ashby
My boss made me write this bio.

[JulienG]

...We do RAID 6 systems running at 1.4TB per sec, over Quad 4Gb Fibre. In these cases, the limit is the bandwidth of the Fibre, not the bandwidth of the drives....

You mean 1.4*GB*/sec, or 11.2Gb (Just less then three full 4Gb FC links, although protocol overhead bumps it up a little).

To do 1.4TB/sec would require *1400* 8Gb links, so not very practical at the moment.

[Sheer Noise]

my head hurts.

I lost a drive of photos last year and it cost me $1k to recover the wedding photos on it for the client. Some place in Nth Sydney where they open the drive in a air-tight room wearing doomsday bio-warfare suits and pull out the disc/s and reassemble in a new thingie or something... Ouch $$-wise.... but I got the photos back.

I don't have anything technical to add...

[smash]

You mean 1.4*GB*/sec, or 11.2Gb (Just less then three full 4Gb FC links, although protocol overhead bumps it up a little).

To do 1.4TB/sec would require *1400* 8Gb links, so not very practical at the moment.

Oops, correct, 1.4 GB per sec...







__________________
Simon "Smash" Ashby
Brain explosion halted