Samsung 1TB 5400RPM or Seagate 1.5TB 7200RPM

[ilovejedd]

My unRAID server needs filling up (currently has 2x Samsung 1TB HD103UJ 7200RPM) and I'm in two minds with regards to the choice of hard drive. I haven't exactly tallied all the data I'm backing up to this thing but I reckon 8TB should last me a while.

 

Samsung EcoGreen 1TB 5400RPM HD103UI

Spin-Up Current (max) 2.0A

Seek 5.6W

Read/Write 6.2W

Idle 5.0W

Standby/Sleep 0.5/0.8W

Warranty 3 years

Price/GB 12 cents/GB

 

Seagate Barracuda 7200.11 1.5TB 7200RPM ST31500341AS

Spin-Up Current (max) 2.8A

Seek 10.10W

Read/Write 10.80W

Idle 8.3W

Standby/Sleep 0.9W

Warranty 5 years

Price/GB 12.67 cents/GB

 

As a point of comparison:

Samsung Spinpoint F 1TB 7200RPM HD103UJ

Spin-up Current 2.4A

Seek 7.6W

Read/Write 8.4W

Idle 6.7W

Standby/Sleep 0.7/1W

Warranty 3 years

Price/GB 13.5 cents/GB

 

If anyone can fill out specs for the Western Digital WD10EACS, that would be much appreciated.

 

Samsung Pros:

lower power consumption

potentially quieter

potentially cooler running

 

Seagate Pros:

longer warranty

higher storage density

 

Any other things you think I can add to the pros and cons list?

[tstolze]

Western Digital 10EACS

 

Current Requirements

Power Dissipation

Read/Write 5.4 Watts

Idle 2.8 Watts

Standby 0.40 Watts

Sleep 0.40 Watts

[Joe L.]

5 of the 1.5 T drives would give you 6 T of space...(one of them used as parity) combine that with your existing 2 drives and you get 8T total.

 

I don't think the price per gig is going to change too much over then next few years, but who knows.  12 cents per gig is great, but then I used to think 50 cents per gig was great not too many years ago.  I just can't see the drives dropping to half their current cost, or doubling in size.

 

I would go with the drives with the longer warranty. I expect them to fail... and trust me, they will, it is just a matter of time.  If you don't need all the space now, get two 1.5 T drives.  Use one for parity, the other for data.  when combined with your two existing 1T drives you would then go from the current 1T of data storage to 3.5T of data storage space.  Then, a few years from now, when you can get 10 cents per gig, purchase more disks as needed.  If nothing else, their warranty will start a few years from now and you will not be paying for electricity to spin them until they are actually needed.

 

Joe L.

[SSD]

In order to add a 1.5T drive to your array, you'd have to upgrade parity to 1.5T.

 

Having a 1.5T parity drive means that your parity check will have to be performed on 1.5T instead of 1T.  This is going to significantly elongate parity check times.  Not a huge deal, but you need to be aware of this.

 

[ilovejedd]

Good point specially since I only have a Celeron 430. Is parity check CPU or hdd bound? I anticipate replacing the hard drives in a couple of years so I reckon I'll be fine even with a 3 year warranty.

 

[WeeboTech]

If possible, I would go with a faster spindle speed and larger cache at least for parity.

 

This way when you have multiple drives writing, they can write as fast as your parity drive can handle.

 

My recent tests today revealed that the WD Green Drives have a read test of 75MB/s and my 7200RPM/32MB Cache Seagates are getting around 105MB/s.

 

If your system is mostly idle with a few reads of media files here and there, the WD green 5400's will do fine (as it has been for me).

 

Since I've been doing a huge number of Torrents on these drives, I've noticed a slight bottleneck.

This in turn is leading me to move 2 of my 8 data disks to the 7200RPM/32MB cache drives for the added speed.

 

Take this in stride when weighing your average usage pattern vs power/heat.

 

As Joe mentioned, buy as many drives as you need, but not more then one additional as the price drops over time.

I always have at least 1 spare to grow into or replace unless I see a stupid price (Which I have a few times and pulled the trigger).

 

As far as the lengthened parity check, this point is true, however, this point also goes for the amount of drives that are the same size as parity.

 

If you bought one 1.5TB parity drive and the rest are 1TB, then after 1TB the parity sync/check will speed along as fast as that drive can read.

I.E. you gain speed as drives reach their end point.

 

I'm not saying to buy a 1tb drive for parity unless you need a data drive with contiguous space up to 1.5tb today.

Just shedding a lil more light into the process.

 

For me, I'll probably go to a 1.5tb parity and data drive soon as one of my directories is getting very large and I cannot access the directory via usershare.

 

[ilovejedd]

I don't anticipate using this for anything other than media. I reckon it'll even be delegated to back-up duties if and when I get around to building my "fake" RAID server (using Ciprico RAIDCore).

 

As of the moment, my storage needs are relatively small. However, I expect it to grow exponentially as I begin to build my Blu-ray collection.

 

I think I'll go with the EcoGreen 1TB 5400RPM 32MB Cache for the moment. The Samsung Spinpoint F will remain the parity drive. I'll just start replacing them when 2TB HDDs come out.

[WeeboTech]

"fake" RAID server (using Ciprico RAIDCore).

 

Fake? That controller looks like a full fledge raid controller. 

[ilovejedd]

"fake" RAID server (using Ciprico RAIDCore).

 

Fake? That controller looks like a full fledge raid controller. 

 

It uses software RAID unlike Areca, Adaptec, etc controller cards. I originally wanted something similar to unRAID sitting on top of a regular Windows set-up. Unfortunately, there doesn't seem to be an equivalent and from the research I've done, with JBOD on Windows, if you lose one drive, then you've effectively lost all your data.

[ilovejedd]

The storage gods have spoken. I'm getting some 1.5TB 7200.11 Barracuda's as the Samsung EcoGreen's are out of stock.

 

On a side note, has anyone noticed that the ABIT AB9 Pro is now $100 instead of $85 on Newegg? I was planning on getting one for a mirrored backup of my unRAID server while I'm still in the research stage of the performance rackmount server.

[ilovejedd]

Went with 3x 1.5TB and now, I've just added another 1.5TB drive to my unRAID array yesterday and was extremely surprised by initial parity check speed (100+MB/s during the first 10 minutes). Is that actually possible or are mine eyes deceiving me? Went to sleep so no idea if speed dropped or something.

 

Drives:

Parity

Seagate Barracuda 7200.11 1.5TB

 

Storage

Seagate Barracuda 7200.11 1.5TB x3

Samsung Spinpoint F 1TB

 

Does the amount of free space on the drives affect parity check speed? I wouldn't think so since the CPU is more than capable of keeping up with XOR calculations but there may be other factors I'm unaware of...

[Joe L.]

Went with 3x 1.5TB and now, I've just added another 1.5TB drive to my unRAID array yesterday and was extremely surprised by initial parity check speed (100+MB/s during the first 10 minutes). Is that actually possible or are mine eyes deceiving me? Went to sleep so no idea if speed dropped or something.

 

Drives:

Parity

Seagate Barracuda 7200.11 1.5TB

 

Storage

Seagate Barracuda 7200.11 1.5TB x3

Samsung Spinpoint F 1TB

 

Does the amount of free space on the drives affect parity check speed? I wouldn't think so since the CPU is more than capable of keeping up with XOR calculations but there may be other factors I'm unaware of...

Free space has absolutely nothing to do with parity check.  Parity check has no concept of files, it just looks at each physical block on the disk in turn.

 

That is by far the fastest parity check speed I've seen reported. 

What motherboard / disk controllers are you using?  (sounds like a real winner)

 

Joe L.

[SSD]

Hope its true but skeptical. Are you sure all the drives were assigned to the array, and not JUST the parity disk?

[ilovejedd]

Free space has absolutely nothing to do with parity check.  Parity check has no concept of files, it just looks at each physical block on the disk in turn.

 

That is by far the fastest parity check speed I've seen reported. 

What motherboard / disk controllers are you using?  (sounds like a real winner)

 

Joe L.

 

Thanks. If the above speed isn't a fluke, I think it can be attributed more to the drives and less the motherboard / disk controller. The Seagate 1.5TB drives contain 375GB platters and the Samsung Spinpoint F 1TB is one of the faster desktop-class hard drives I've seen benchmarked.

 

Specifications:

MB: Abit AB9 Pro

CPU: Intel Celeron 430 @ 1.8GHz

RAM: Kingston 2x1GB DDR2 667

GPU: ATI RageXL 8MB

 

As of the moment, all hard drives are connected to the Intel ICH8R controller.

 

Hope its true but skeptical. Are you sure all the drives were assigned to the array, and not JUST the parity disk?

 

Yeah, I'm kinda skeptical, too, that's why I'm asking here if it was even possible. All drives are assigned to the array, however, two of them are pretty much still zeroed out. Are parity check speeds logged in the syslog? Parity check was already done by the time I woke up so I'm curious if it actually did finish it in the 4 hours it said it would. I rather think it might have. I slept 12-ish and checked the unRAID server around 5:30AM. All drives were already spun down with the exception of the Samsung (that one seems to spin down on its own schedule).

[RobJ]

Look in the syslog for a line with 'sync done'.  An example:

md: sync done. time=9557sec rate=51102K/sec

 

The rate above is the average parity check or sync rate, for the entire run.  Above is about 51MB/s.  The message seems to be the same for both parity syncs and parity checks.  You have to look earlier in the syslog to tell if a check or sync was started.

[SSD]

The web gui also gives the date and time when the last parity check finished.

[ilovejedd]

Thanks. I'm away from home right now so can't telnet into the unRAID server (does the iPhone have a telnet app?), but I did check the web GUI.

 

Last checked on 10/16/2008 3:15:26AM, finding 0 errors.

 

Forgotten what time exactly the parity check started. Will check when I go home, before I add yet another 1.5TB drive to the server.

[ilovejedd]

Not exactly 100MB/s but at 83MB/s, it's still pretty impressive:

 

Oct 15 22:28:27 unRAID kernel: mdcmd (61): check
Oct 15 22:28:27 unRAID kernel: md: recovery thread got woken up ...
Oct 15 22:28:27 unRAID kernel: md: recovery thread checking parity...
Oct 15 22:28:27 unRAID kernel: md: using 1152k window, over a total of 1465138552 blocks.
Oct 16 03:15:26 unRAID kernel: md: sync done. time=17218sec rate=85093K/sec
Oct 16 03:15:26 unRAID kernel: md: recovery thread sync completion status: 0

 

Lol, I'm gonna take a guess that most people use Green Power drives. I would've taken that route, too, but they don't make 1.5TB versions of those, yet.

 

That or old recycled hard drives. Ironically, this was what I build my unRAID server for, but I eventually came to the conclusion that it wouldn't give me the storage density I require.

[WeeboTech]

Not exactly 100MB/s but at 83MB/s, it's still pretty impressive:

 

That is quite respectable.

I don't get those numbers, then again, I use the WD green drives for all except my parity, torrent share and frequently accessed data share.

I prefer cool and low power.

[Romir]

For the record here's what the Seagate 1.5tb drives are capable of in a non-mixed array. I have four of them on a ICH8 controller.

 

Parity creation: Oct 24 06:56:37 unRAID kernel: md: sync done. time=25987sec rate=56379K/sec

Parity check:    Oct 24 11:19:28 unRAID kernel: md: sync done. time=15743sec rate=93066K/sec

 

They're replacing some Hitachi 400gb drives whose GB per dollar was 7.5x more expensive. At least none of them went bad, they were the first 5 platter drives since the infamous 75GXP "Deathstars".