NAS Server - Ideas?

[Arthur]

As John says the Oracle box is a whole package. It makes use of the same backend software stack as Nexenta (ZFS, Solaris etc) and adds good hardware and a nice gui analytics package. The advantage is the way it makes use of RAM, SSDs and HDs to give the fastest performance possible.

I think my next NAS will probably run NexentaStor Community Edition or ZFS Guru. ZFS is fantastic.

 

For some truly massive NASs you should take a look at the following thread on the HardOCP forums. Some people have 100TB+ at home!!!

 

http://hardforum.com/showthread.php?t=1393939

[Soulfish]

I think my next NAS will probably run NexentaStor Community Edition or ZFS Guru. ZFS is fantastic.

 

For some truly massive NASs you should take a look at the following thread on the HardOCP forums. Some people have 100TB+ at home!!!

 

The [H]ard Forum Storage Showoff Thread - Post your 10TB+ systems - [H]ard|Forum

 

Ohhh I hadn't seen the HardOCP thread before. That 9U chenbro case is a monster

[dhicks]

QNAP's are great for what you want to do

 

We're buying a QNAP TS-809U-RP - rack-mounted, dual power supplies, and for somewhere under £1,500, which really isn't bad. Am I correct in thinking that QNAP's devices are all software RAID (mdadm?) based? If so, this QNAP device is sold as being just the thing to store your virtual machine disk images on, therefore the software RAID must be able to keep up with typical I/O speeds. Does anyone else use a QNAP device for VM disk images, can you confirm that performance is okay? If so, why does anyone bother with hardware RAID, then? Have QNAP got some cunning tuneing trick they perform, or are they using their own software RAID system, or do they simply find that they installed mdadm and it worked fine?

 

Does anyone make use of the QNAP's medi library feature? How well does it work in a school context?

 

--

David Hicks

[gshaw]

Following on with the QNAPs can anyone comment on the iSCSI reliability?

 

I keep getting warned that their implementation is somewhat wobbly, which is no good if I want to use it as a backup target. If not QNAP then are any of the other SMB-type NAS appliances any better?

 

I've got a DL180 G6 in my office at the moment that was the original plan for backup server (with Veeam installed locally) but HP decided it would be a good idea to run 4 fans at 10000rpm the moment you plug a RAID card in, no way that can stay in here Problem is haven't got any other places in the building away from the server room and something in me says having backups in there isn't a good idea (unless I can do some sort of realtime sync to... a QNAP?)

[dhicks]

Building a Backblaze pod does seem tempting. I've even found a UK group that have fabricated the cases and got all the hard to buy parts (UK Blazers and Custom Electronic Enclosures for Engineers and Designers) but I'm also being tempted by a 36-bay 4U supermicro case (Super Micro Computer, Inc. - Products | Chassis | 4U | SC847A-R1400UB).

 

Just to update this thread: The BackBlaze Pod 2.0 is now out!

[Arthur]

Just to update this thread: The BackBlaze Pod 2.0 is now out!

^ Wow!!!

 

Here are a couple of interesting quotes... Proof if you need it, that you shouldn't use Western Digital (or Seagate) HDDs in RAID arrays.

 

We are constantly looking at new hard drives, evaluating them for reliability and power consumption. The Hitachi 3TB drive (Hitachi Deskstar 5K3000 HDS5C3030ALA630) is our current favorite for both its low power demand and astounding reliability. The Western Digital and Seagate equivalents we tested saw much higher rates of popping out of RAID arrays and drive failure. Even the Western Digital Enterprise Hard Drives had the same high failure rates. The Hitachi drives, on the other hand, perform wonderfully.

 

We see fairly high infant mortality on the hard drives deployed in brand new pods, so we like to burn the pods in for a few days before storing any customer data. We have yet to see any drives die because of old age, which will be fascinating to monitor in the next few years. All told, Sean replaces approximately 10 drives per week, indicating a 5 percent per year drive failure rate across the entire fleet, which includes infant mortality and also the higher failure rates of previous drives. (We are currently seeing failures in less than 1 percent of the Hitachi Deskstar 5K3000 HDS5C3030ALA630 drives that we’re installing in pod 2.0.)

[dhicks]

Here are a couple of interesting quotes... Proof if you need it, that you shouldn't use Western Digital (or Seagate) HDDs in RAID arrays.

 

It's a good point, and the article made some interesting points - like how they yet to have a drive fail due to age. Those Hitachi drives are a bit pricy (£150 each, minimum) but if they aid reliability then they're probably worth it. This thread might give some clue as to the early failure rate of WD drives:

 

http://www.edugeek.net/forums/hardware/77295-disk-backup-device.html#post687637

Edited July 21, 2011 by dhicks

[zag]

As someone who has been fascinated by cheap storage for many years (Have 25tb+ at home) I love that backblaze company!

 

Its what I have seen as well, the Hitachi drives I have in 3tb versions are rock solid.

 

Western digital has a near 20% failure rate for me and I will never buy their drives again. Lesson Learned!!

[gshaw]

Have Hitachi finally exorcised the ghost of the Deathstar now in that case? Always used to be very suspicious of any drives with the IBM "heritage"

 

So we've got Seagate (old Maxtor) and now WD that have question marks over reliability. If Samsung do sell up as suggested they might who can you trust to make a decent quality HDD these days?! Seems like they're doing their level best to help out the SSD manufacturers at the moment

 

This does raise questions about disk backup though with many people using these grades of drive, with some of the failure rates being experienced is anything less than RAID6 a bit of a risky proposition bearing in mind the size of arrays needed for 14-30 days of disk backup retention?

[SHimmer45]

WD are in the are/have aquired hitachi, im a fan of WD drives as not had a major failure which has caused any data lose.

i think in the next few years there will just be WD and Seagate left in the consumer mechanical drive market

[Duke]

Just to update this thread: The BackBlaze Pod 2.0 is now out!

 

Must admit I'm very impressed with what these guys are doing, the costs they're achieving are incredibly good! I'm about to do one for us on a much smaller scale - SuperMicro chassis, RAID card, 8x 2TB disks.

[dhicks]

This does raise questions about disk backup though with many people using these grades of drive

 

To be fair, that article does seem to be saying that disk failures tend to happen early in a drive's lifetime - either the disk has an issue and conks out soon after you install it or it doesn't and runs for years. Also, drive manufacturers seem fine about replacing faulty disks still in garuntee, so there's no real cost involved in a faulty drive.

 

It should be emphasised to those looking for a backup solution that the hardware we're talking about here doesn't neccesarily provide much by way of fault tolerence by itself - Backblaze have their own software that would seem to provide redundancy accross multiple hardware units, and they are operating at a seriously large scale. They simply build systems to expect and handle drive failure.

 

Peter Hopton's comments in the thread I linked to above make sense - on today's large disks, reading the whole disk stands a good chance of hitting an error, so if you're relying on a single disk to restore your RAID-5 array you might be pushing your luck - RAID 6 would seem like the senisble option these days for any larger array.