May just have had an idea that'll save a few hundred quid

[BatchFile]

What do you lot think?

 

I'm putting a new file server in this summer as the old one is (a) full and (b) 7 years old going for a Fujitsu S7 like this Fujitsu PY TX150 S7 XEON X3430 2x2GB 1333 ECC UB/DVD-RW (VFY:T1507SX100GB) - dabs.com . I need a lot of storage, like 2TB or so... but the Fujitsu HDDs that size (1TB and use 4 for Raid 10) are really really expensive, getting on for £250 each. They're only SATA disks in any case.

 

My brainwave is as follows: buy 4 small fujitsu drives, say 160GB, which are much cheaper, throw the disks away (well keep them as spares for workstations), keep the hot swap caddies and put 4 large standard £30ish SATA disks in instead. They're raided (and hot swappable) so reliability's not really an issue, and I've succesfully swapped a normal HDD into a caddy for our Acer servers so in principal it should work out fine shouldn't it?

 

Unless of course it's possible to buy the caddies seperately, but I haven't seen any anywhere...

[teejay]

Been there, got the t-shirt, now buy the manufacturers official ones rather than 3rd party. Not all HDD's play well in server RAID systems.

[Arthur]

Not all HDD's play well in server RAID systems.

+1. Western Digital drives are particularly bad.

[BatchFile]

I was going to stick to whichever manufacturer was in the caddies when they arrive... that any better?

[FN-GM]

The problem with SATA drives is the speed. It may cause bottle necks.

[jamesfed]

How many users are you going to be supporting with this server?

 

There is a big differance between standard desktop drives and 'enterprise class' drives which you will find out about all too soon if you go down the wrong path....

[glennda]

If you do it that way make sure you buy enterprise grade HDD's not the normal desktop ones - I bought 20 2TB ones and they where about 130 each

[mavhc]

They're not that different if they're the same spec. It's mostly just the Time Limited Error Recovery firmware setting. Depending on your ability to alter raid settings, and whether a 2 min pause would be a problem. Also "Enhanced RAFF technology includes sophisticated electronics to monitor the drive and correct both linear and rotational vibration in real time. The result is a significant performance improvement in high vibration environments over the previous generation of drives.". And they're tested for running 24/7.

 

Error recovery control - Wikipedia, the free encyclopedia

 

If you're using zfs, advice seems to be to disable TLER

[CyberNerd]

If you're using zfs, advice seems to be to disable TLER

 

Would you want to ditch the fake Raid altogether if you have the luxury of zfs? I'm sure I read somewhere that zraid prefers direct attached disks rather than LUNS or raided systems. One of my mini projects is zfs for our fileservers (I reckon around 2x dedupe from tests) and am wanting to know the best way of doing it (on SAN)

[accura2000]

+1. Western Digital drives are particularly bad.

 

What do you base this on? All my main server data drives are Western Digital Black drives, running RAID 5 without any issues or failures. They have been in use for the last 2 years now without a hiccup, so from my point of view, i would never use any other manufacturer.

[BatchFile]

:confused:Hmmm... Just been doing a little digging on the Fujitsu TX150 S7 that we've already got (Backup DC so only 160 GB drives) and found that the drives in it (Fujitsu supplied Hot Swap) are Seagate Barracuda ST3160318AS 's - been doing a bit of reading on seagate.com (attached here 100529369b.pdf) and from what I can tell they're just standard drives - aren't they? the only occurance of "Enterprise" in the entire document is to say

2.11.1Annualized Failure Rate (AFR) and Mean Time Between Failures (MTBF) The product shall achieve an Annualized Failure Rate (AFR) of 0.32% (MTBF of 0.75 million hours) when operated

in an environment of ambient air temperatures of 25°C. Operation at temperatures outside the specifications

in Section 2.8 may increase the product AFR (decrease MTBF). AFR and MTBF are population statistics

that are not relevant to individual units.

AFR and MTBF specifications are based on the following assumptions for desktop personal computer environments:

•2400 power-on-hours per year.

•10,000 average motor start/stop cycles per year.

•Operations at nominal voltages.

•Temperatures outside the specifications in Section 2.8 may reduce the product reliability.

•Normal I/O duty cycle for desktop personal computers. Operation at excessive I/O duty cycle may degrade product reliability. The desktop personal computer environment of power-on-hours, temperature, and I/O duty cycle affect the

product AFR and MTBF. The AFR and MTBF will be degraded if used in an enterprise application

If that's the case is there really that much to gain by using the Fujitsu branded stuff at several times the cost?

[mavhc]

Would you want to ditch the fake Raid altogether if you have the luxury of zfs? I'm sure I read somewhere that zraid prefers direct attached disks rather than LUNS or raided systems. One of my mini projects is zfs for our fileservers (I reckon around 2x dedupe from tests) and am wanting to know the best way of doing it (on SAN)

 

I just use ZFS on my home NAS currently. It definitely wants to just use bare drives. There's a slight interesting point about whether to use the raw drive or a whole partition, if you were replacing the drive with one a few sectors smaller it wouldn't work. Also there's the 4k align stuff for Advanced Format 4k drives.

 

ZFS speed depends on CPU more than other filesystems, as it CRCs everything, when I switched from an ATOM 330 to a AMD E350 it sped up quite a bit.

[CyberNerd]

I just use ZFS on my home NAS currently. It definitely wants to just use bare drives. There's a slight interesting point about whether to use the raw drive or a whole partition, if you were replacing the drive with one a few sectors smaller it wouldn't work. Also there's the 4k align stuff for Advanced Format 4k drives.

.

 

I didn't think zraid cares what size the drive is because it just adds it to the pool, unlike traditional raid systems.

[mavhc]

If you're replacing drives it does for raid5 or 6 type pools. For raid 0 type it doesn't matter. Also you can add more raid 5 type collections to a pool.

 

Eg 5 2TB drives in raidz1 (raid 5), mounted as /tank, then add 5 3TB drives in raidz1, add that to /tank, and the size will increase, although the files wouldn't be balanced between the 2 unless you copied them all off and on again. But to replace a drive, you need one the same size or bigger.

 

If you replace all 5 drives, 1 at a time, at the end you get a raidz1 of the new size. Of course you just spent a week with no spare drive. Also you can't add a drive to make a 6 disc raidz1 array. That only works for raid 0 type pools.

[localzuk]

What do you lot think?

 

I'm putting a new file server in this summer as the old one is (a) full and (b) 7 years old going for a Fujitsu S7 like this Fujitsu PY TX150 S7 XEON X3430 2x2GB 1333 ECC UB/DVD-RW (VFY:T1507SX100GB) - dabs.com . I need a lot of storage, like 2TB or so... but the Fujitsu HDDs that size (1TB and use 4 for Raid 10) are really really expensive, getting on for £250 each. They're only SATA disks in any case.

 

My brainwave is as follows: buy 4 small fujitsu drives, say 160GB, which are much cheaper, throw the disks away (well keep them as spares for workstations), keep the hot swap caddies and put 4 large standard £30ish SATA disks in instead. They're raided (and hot swappable) so reliability's not really an issue, and I've succesfully swapped a normal HDD into a caddy for our Acer servers so in principal it should work out fine shouldn't it?

 

Unless of course it's possible to buy the caddies seperately, but I haven't seen any anywhere...

 

Ok, firstly standard SATA drives have a lower 'MTBF' than enterprise class drives. If you want to maintain 'hot swap' capability, then you'd need to buy hot swap capable drives. 1TB hotswap drives aren't cheap basically. Cheapest I can find is about £180.

 

If it were me, I'd be going for the Fujitsu ones for piece of mind. Although, if it were me I'd be running a mile and screaming before I bought Fujitsu...

[Arthur]

I didn't think zraid cares what size the drive is because it just adds it to the pool, unlike traditional raid systems.

Ideally you would want to keep all drives in a vdev the same, not just for optimal performance, but also because your vdev will be limited by the size of the smallest drive if you used disks with different capacities.

 

Unlike traditional file systems, which reside on single devices and thus require a volume manager to use more than one device, ZFS filesystems are built on top of virtual storage pools called zpools. A zpool is constructed of virtual devices (vdevs), which are themselves constructed of block devices: files, hard drive partitions, or entire drives, with the last being the recommended usage. Block devices within a vdev may be configured in different ways, depending on needs and space available: non-redundantly (similar to RAID 0), as a mirror (RAID 1) of two or more devices, as a RAID-Z (similar to RAID-5) group of three or more devices, or as a RAID-Z2 (similar to RAID-6) group of four or more devices. In July 2009, triple-parity RAID-Z3 was added to OpenSolaris.

 

Thus, a zpool (ZFS storage pool) is vaguely similar to a computer's RAM. The total RAM pool capacity depends on the number of RAM memory sticks and the size of each stick. Likewise, a zpool consists of one or more vdevs. Each vdev can be viewed as a group of hard disks (or partitions, or files, etc.). Each vdev should have redundancy because if a vdev is lost, then the whole zpool is lost. Thus, each vdev should be configured as RAID-Z1, RAID-Z2, mirror, etc. It is not possible to change the number of drives in an existing vdev (Block Pointer Rewrite will allow this, and also allow defragmentation), but it is always possible to increase storage capacity by adding a new vdev to a zpool. It is possible to swap a drive to a larger drive and resilver (repair) the zpool. If this procedure is repeated for every disk in a vdev, then the zpool will grow in capacity when the last drive is resilvered. A vdev will have the same capacity as the smallest drive in the group. For instance, a vdev consisting of three 500 GB and one 700 GB drive, will have a capacity of 4 x 500 GB.

 

In addition, pools can have hot spares to compensate for failing disks. When mirroring, block devices can be grouped according to physical chassis, so that the filesystem can continue in the case of the failure of an entire chassis. (Source)

 

Here's what a hypothetical 30TB zpool could look like (please correct me if I am wrong )...

 

+---------------------------------------------------------+
| +-------++-------++-------++-------++-------+           |
| |       ||       ||       ||       ||       |           |
| |       ||       ||       ||       ||       |   16TB    |
| | 4.0TB || 4.0TB || 4.0TB || 4.0TB || 4.0TB |  RAID-Z1  |
| |       ||       ||       ||       ||       |   vdev    |
| |       ||       ||       ||       ||       |           |   30TB
| +-------++-------++-------++-------++-------+           |   zpool
| +-------++-------++-------++-------++-------+           |
| |       ||       ||       ||       ||       |           |
| |       ||       ||       ||       ||       |   8.0TB   |
| | 2.0TB || 2.0TB || 2.0TB || 2.0TB || 2.0TB |  RAID-Z1  |
| |       ||       ||       ||       ||       |   vdev    |
| |       ||       ||       ||       ||       |           |
| +-------++-------++-------++-------++-------+           |
+---------------------------------------------------------+

[m25man]

Bearing in mind that in most cases the MTBF is purely another hypothetical statistic as clearly it isn't possible to actually test an HDD for these periods unless you happen to own a TARDIS.

Its a bit like the manufacturer claiming that the Triton Cell of my Smart will protect me in the event of a 70mph head on collision, the fact that my internal organs will probably emerge from most orifice's doesn't come into it, likewise you really won't know if that economy drive is going to run for 8 years until it does or will it simply expire after 13 months along with all of your data?

 

The enterprise disks run a lot cooler, vibrate less, suffer from fewer errors as a result, that in itself is why you should buy them and entrust your data to them.

I have used Enterprise quality SATA in lots of HP Caddies without any issues.

Also, checkout the Netgear ReadyNAS list of approved HDDs this lists a whole bunch of drives that meet their stringent warranty requirements and as such should be more than capable of the requirements of the OP's proposed use.

[iPrimary]

Ok, firstly standard SATA drives have a lower 'MTBF' than enterprise class drives. If you want to maintain 'hot swap' capability, then you'd need to buy hot swap capable drives. 1TB hotswap drives aren't cheap basically. Cheapest I can find is about £180.

 

If it were me, I'd be going for the Fujitsu ones for piece of mind. Although, if it were me I'd be running a mile and screaming before I bought Fujitsu...

 

Nothing wrong with recent Fujitsu equipment, Fuj Siemens are a bit more questionable. Recent Fujtisu equipment comes with a free 3 yr on-site next day warranty for all education customers, bought through an appropriate vendor that is.