Brian

Seagate Barracuda Green 2TB Review

34 posts in this topic

Right before the end of the year, Seagate launched an update to their eco-friendly line of hard drives. Formally known as Barracuda LP (Low Power), Seagate has changed to a more mainstream Barracuda Green branding. The Barracuda Green isn't just about a marketing message though, the 3.5" hard drives feature capacities up to 2TB, SATA 6Gb/s interface, 64MB cache and 5,900 RPM spin speed. This combination of specs leads Seagate to claim the Barracuda Green to be the best performing green drives on the market.

Full Review

Share this post


Link to post
Share on other sites

What about Seagate SmartAllign? How the partitions were alligned? Did you tested different allignments? Is it possible to analyse this seagate technology?

Share this post


Link to post
Share on other sites

What about Seagate SmartAllign? How the partitions were alligned? Did you tested different allignments? Is it possible to analyse this seagate technology?

I second this. Seagate says their SmartAlign technology is meant to be 'transparent' to the end user i.e. that their implementation automagically aligns partitions etc Their marketing states "SmartAlign technology maintains consistent hard drive performance, even when it encounters hard drive partition misalignment conditions". Could you test it with the QNAP TS-459 Pro+ Turbo which had problems with the 4k WD 2TB? Thanks for any response.

Share this post


Link to post
Share on other sites

I second this. Seagate says their SmartAlign technology is meant to be 'transparent' to the end user i.e. that their implementation automagically aligns partitions etc Their marketing states "SmartAlign technology maintains consistent hard drive performance, even when it encounters hard drive partition misalignment conditions". Could you test it with the QNAP TS-459 Pro+ Turbo which had problems with the 4k WD 2TB? Thanks for any response.

In the case of the QNAP, we would have loved to test it as much as the Caviar Green, but we only have one drive, so we couldnt compare RAID build times with it.

Share this post


Link to post
Share on other sites

Just imagine when they take these platters and spin them at 7,200rpm. If they can do it the sequential throughput should put many 15,000rpm drives to shame. The access time would be lacking but the capacity. . .

Share this post


Link to post
Share on other sites

I wonder how this drive compares to the ST2000DL001 that I pulled out of a GoFlex enclosure? The 001 reports SATAII and no cache (seems unlikely) according to HDTune. It also reports 512b alignment.

Edited by cbrworm

Share this post


Link to post
Share on other sites

That's the prior generation LP I think. And yes, I'm sure it has a cache ;) If you look at our results, you should see a pretty steady performance boost out of the new Green, as we saw in comparison to the LP.

Share this post


Link to post
Share on other sites

Once again disappointed that there is no SAS (+raid/vibration tolerant) version of these LP large capacity drives. For nearline apps where you need several hundred TB's of storage and with COW file systems these would dramatically cut down on power use. You don't need fast spindle speeds when the workload is light and you have hundreds of them.

Heck even here at home if I replaced my arrays it would save over 600W of power; granted I'd probably buy more drives so won't see it in my bill. ;)

Share this post


Link to post
Share on other sites

That's the prior generation LP I think. And yes, I'm sure it has a cache ;) If you look at our results, you should see a pretty steady performance boost out of the new Green, as we saw in comparison to the LP.

What about the Seagate Smartalign and partition alignment? What physical and logical sector size drive reports to the OS?

Share this post


Link to post
Share on other sites

It has SmartAlign, we spent a little more time on what that means in the 750GB Momentus review. It's honestly still more than a bit ambiguous though.

What physical and logical sector size drive reports to the OS?

Let us double check on this and report back to you.

Share this post


Link to post
Share on other sites

Let us double check on this and report back to you.

Hello Brian

That Information about 4kn vs. 512e is the most important in any HDD review for me right now. I don't want to buy any drives with emulation, especially if they report the wrong physical sector size. So thanks in advance for digging that up for us.

You can find out with newer Linux kernels...

In Linux, these parameters are exported via the following sysfs nodes.

physical sector size : /sys/block/sdX/queue/physical_block_size

logical sector size : /sys/block/sdX/queue/logical_block_size

alignment offset : /sys/block/sdX/alignment_offset

(from https://ata.wiki.kernel.org/index.php/ATA_4_KiB_sector_issues)

Share this post


Link to post
Share on other sites

Would an hdparm -I printout on each of our drives help you out with the sector emulation information? Unless you might know otherwise... would just using the latest version of Ubuntu work fine in this case? I have a spare system we could setup in the lab just to connect drives to as part of the test to pull off any information needed that might be hidden/hard to get inside Windows.

Share this post


Link to post
Share on other sites

Would an hdparm -I printout on each of our drives help you out with the sector emulation information? Unless you might know otherwise... would just using the latest version of Ubuntu work fine in this case? I have a spare system we could setup in the lab just to connect drives to as part of the test to pull off any information needed that might be hidden/hard to get inside Windows.

Yes, if you can:

hdparm -I {/dev/device}

cat /sys/block/{device}/queue/physical_block_size

cat /sys/block/{device}/queue/logical_block_size

cat /sys/block/{device}/alignment_offset

smartctl -a {/dev/device}

all of which can be run from a linux live cd like ubuntu desktop. You will need to install hdparm/smartmontools (sudo apt-get install hdparm smartmontools) for the commands.

Share this post


Link to post
Share on other sites

I will have to verify that one of my spare systems will report the correct information for all drives (those 3TB models throw a wrench in the works unless through the addon card) and make sure what gets reported is consistent. If everything is a go I will setup a little script and just output that data to a text file for each drive and link it in reviews. For the average reader it is a bit "out there" in terms of what it means, but we will soon have a spot for those things as we list our full test results.

Sound good?

Share this post


Link to post
Share on other sites

Yes, very good and I was going to suggest having them linked to the drive reviews/model database as this would be very good information to collect. Granted the number of people who can make use of it is probably limited but since it's rather easy to acquire the net gain is still positive.

One suggestion in procedure with drive reviews I would suggest a smartcl -a </device> before the review is started and another one after all the tests are done. This (for drives that support it) would also show items like error count increase over the test window which even though not linear may show some data that drives even though performing well would have higher correctable errors than one would like.

Share this post


Link to post
Share on other sites

I suppose since there is really no added workload making linux dump one more item into a text file, I will add that too. Can't promise when it will start, since I need to prep the system, but it should be soon. To make things easier on our end, does someone want to make a script that will automatically run in the background looking for newly attached devices, automatically create a file named the drives device name, and then dump the information listed above inside it? I am a bit rusty on my linux scripting, so if someone had that working for us off the bat I could probably get everything in our lab going by this weekend. Only problem is on some drives, you would be seeing the after effects of our tests in terms of smart data, since they are 'used' at this point.

And yea going forward as we have more man hours to do more fun/interesting stuff, we plan on linking to a much larger source of information in our reviews. For the longest time providing the entire iometer output stats have been something we have been thinking about, just no place to really stick them on the site.

In other news we are testing an SSD that costs more than some new cars right now. :)

Share this post


Link to post
Share on other sites

Well, I haven't really looked into doing active scan to notify of newly added devices (since w/ recent kernels the hot add/replace scsi stuff has been automatic and once you know a device you reference it by WWN or similar as the device names (/dev/sdX) are dynmaic). Anyway, if you know the device itself which from booting from CD you should be able to discern by looking at /proc/partitions you can run the following:

---

#!/bin/bash

# Set static variables
CURDATE=`date +%Y%m%d%H%M%S`
CURYEAR=`date +%Y`
OUTDIR="/tmp"
DEVICE=$1

MANUFACT=`smartctl -i ${DEVICE} | grep "Device:" | awk '{print $2}'`
MODEL=`smartctl -i ${DEVICE} | grep "Device:" | awk '{print $3}'`
SERNUM=`smartctl -i ${DEVICE} | grep "Serial number:"| awk '{print $3}'`

smartctl -a ${DEVICE} > ${OUTDIR}/${CURDATE}-${MANUFACT}-${MODEL}-${SERNUM}-smartctl.txt
hdparm -I   ${DEVICE} > ${OUTDIR}/${CURDATE}-${MANUFACT}-${MODEL}-${SERNUM}-hdparm.txt

which when given the /dev/sdX device name will pull out what it can and store it in /tmp.

I would be interested if anyone /does/ have a script to watch and report newly hot-added devices though, would be interesting to look at. My guess would be to look at udev and build an array and do compares every x seconds.

Share this post


Link to post
Share on other sites

Well we don't want to play around with a live-cd, so this will be a system idling along 24/7 waiting for a drive to be plugged in and have its info parsed :). It should only be one drive at a time, so maybe it wont be too bad to find out what was "just plugged in".

Share this post


Link to post
Share on other sites

Well if it's going to be a static build and I'm assuming a hot-plug carrier or something that you will be dropping devices in? If so then once it's built take a look at your /dev/disk/by-path/* files and find out what your carrier is. Use that as your device id.

i.e. /dev/disk/by-path/pci-0000:84:00.0-scsi-0:0:0:1

this should remain static as it's a pci bus/address designation. It may point to different /dev/sdX numbers as those are dynamic but you can use the path for a match in the script (as hdparm/smartctl don't understand by-* syntax) so something like

DEVICE="/dev/"`ls -l /dev/disk/by-path/ | grep "pci-0000:84:00.0-scsi-0:0:0:1" | cut -d\/ -f3`

opposed to the argument pass of $1 above.

Share this post


Link to post
Share on other sites

Ok, some of the reading in what I listed above seem off as to what we expected. The part about the offset being 0 on all of them and the both generation of greens being 512/512 is really strange. Any ideas which linux kernels fully supported proper recognition of newer 4K drives?

Linux version info:

Linux version 2.6.32-25-server (buildd@allspice) (gcc version 4.4.3 (Ubuntu 4.4.3-4ubuntu5) ) #45-Ubuntu SMP Sat Oct 16 20:06:58 UTC 2010

Share this post


Link to post
Share on other sites

You mean logical sector-0 offset? It's not explained as to what that is for in the source code (looking at 5.40 smartmontools package) but I /think/ that was there for virtualized devices (i.e. luns on a san or iscsi volumes (zvols) where the virtual block device is not aligned to a boundary). I have never seen this presented however and if true would pretty much always be zero on a discrete device.

The greens (assuming you mean the WD30EZRS-11J99B0) shows that it has a physical sector size of 4096bytes which is what we would expect but a logical size of 512 (i.e. the drive is doing translation). I would have hoped there was a jumper or something to allow the drive to show it's logical size to be equal to the physical size of 4096. At least it's showing up correctly as a physical (native) size.

As for the linux kernels, > 512b sectors should be supported (if presented) on all recent kernels > 2.6.29 I think. Martin Peterson submitted patches a while ago (mainly for T10 DIF functions (i.e. 520; 528byte and 4160byte sectors) which at this point I think only the LSI HBA's support.

Actually from what you've posted is what I was expecting to see, that even the large drives w/ 4K sectors physically are still presenting 512b to the host. The only one I think did not was the external seagate goflex 3TB drive if you cracked it out of the case and direct attached it. Do you have one of those we can look at?

Share this post


Link to post
Share on other sites

If you look at the WD20EARS (both versions) they report as 512 logical/physical... not 4K. Only that 3TB Caviar Green was 512/4K.

Here is a 3TB Barracuda XT

ST33000651AS.txt

It too is showing 512 logical/physical which is just perplexing.

Share this post


Link to post
Share on other sites

And these drives are direct SATA attached (i.e. not going through USB or any other type of interposer board or expander chips?)

That is interesting that you're not seeing it if you're direct attached, that means that the drive is masking it itself probably for compatibility reasons but that also throws a wrench into the works for systems that really /want/ the 4k sectors.

EDIT: another thing you can try is sudo fdisk -l <device> for the drives in question, it's possible that smartctl/hdparm could be messed up.

Edited by stevecs

Share this post


Link to post
Share on other sites

I will give it another go tomorrow. Stashed the computer to the side to get catch up on some TV shows :)

All drives were connected through an Intel SATA chipset (whatever revision came with the Intel 965 desktop boards). The 3TB models I connected through a HighPoint HBA and Intel, and attached both reading but as you can see both listed same results.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now