Sign in with
Sign up | Sign in
Your question

My apple software RAID vs. esata port multiplier apple software raid

Last response: in Storage
Share
March 9, 2010 5:15:01 AM

Hi all,

I have 2x1tb samsung spinpoints internal connected sata. Apple software raid striped for speed.

these two disks achieved these scores in xbench:

Sequential 424.12
Uncached Write 753.86 462.86 MB/sec [4K blocks]
Uncached Write 674.14 381.43 MB/sec [256K blocks]
Uncached Read 228.30 66.81 MB/sec [4K blocks]
Uncached Read 446.21 224.26 MB/sec [256K blocks]
Random 333.12
Uncached Write 467.82 49.52 MB/sec [4K blocks]
Uncached Write 975.53 312.30 MB/sec [256K blocks]
Uncached Read 194.83 1.38 MB/sec [4K blocks]
Uncached Read 269.36 49.98 MB/sec [256K blocks]


Alternatively I have 4x500gb samsungs in a rosewill external port multiplier connected esata to the enclosures included silicon image PCIe card. Again, striped Apple software RAID.

those disks achieve this xbench score:

Sequential 135.88
Uncached Write 197.33 121.16 MB/sec [4K blocks]
Uncached Write 149.88 84.80 MB/sec [256K blocks]
Uncached Read 80.09 23.44 MB/sec [4K blocks]
Uncached Read 191.91 96.45 MB/sec [256K blocks]
Random 169.83
Uncached Write 115.24 12.20 MB/sec [4K blocks]
Uncached Write 297.65 95.29 MB/sec [256K blocks]
Uncached Read 163.68 1.16 MB/sec [4K blocks]
Uncached Read 184.95 34.32 MB/sec [256K blocks]



My question is this: Is it common for externals using esata in a port multiplier to score lower benchmarks than drives connected straight sata?

Is it because the internals each have their own sata, and all four drives run off of a single esata?

I guess I thought the speed would be a little bit better since there are four drives capable of writing and reading at the same time vs. two. But honestly, it's not that big of a deal as the external is just for a backup, so speed isn't really that important. I'm just curious. thanks in advance.


a c 127 G Storage
March 9, 2010 12:32:07 PM

Your benchmarks are false. Please use proper benchmarks and post again. With 2 drives, a sequential speed of 462MB/s is clearly bogus.
m
0
l
March 9, 2010 1:16:44 PM

hey pal, I just pasted what xbench told me, perhaps yu could offer a more proactive response next time you decide to answer someones question. Perhaps suggest a specific software for benching my speeds, or even average speeds for a two hard drive set up. THANKS!
m
0
l
Related resources
a c 127 G Storage
March 9, 2010 3:32:29 PM

Since you're on Mac i cannot offer information beyond what you can find with google. You can, however, use a low level benchmark with dd, using the command line terminal.

First you would need to know where your RAID volume is mounted, for example /raid. You should find this out by using the "mount" and "df -h" commands. If you got your RAID mountpoint, you can start testing:

dd if=/dev/zero of=/raid/zerofile.000 bs=1m count=20000

This command would create a file called zerofile.000 on the /raid volume, and write 20GB of zeroes to it. It may take a long time to complete, and it wouldn't give you any output until its complete - so just be patient. After this test, try again by reading it:

dd if=/raid/zerofile.000 of=/dev/null bs=1m

Warning: using the dd command may be dangerous if you make typing errors or the commands are erroneous or malicious. Always make sure you have a good backup before experimenting with your live data. That said, the commands above should be harmless; after the test you can remove the testfile with this command:

rm /raid/zerofile.000

The output of both dd commands include the throughput speed, these should be accurate with this benchmark.
m
0
l
March 16, 2010 12:04:59 AM

sub mesa said:
Since you're on Mac i cannot offer information beyond what you can find with google. You can, however, use a low level benchmark with dd, using the command line terminal.

First you would need to know where your RAID volume is mounted, for example /raid. You should find this out by using the "mount" and "df -h" commands. If you got your RAID mountpoint, you can start testing:

dd if=/dev/zero of=/raid/zerofile.000 bs=1m count=20000

This command would create a file called zerofile.000 on the /raid volume, and write 20GB of zeroes to it. It may take a long time to complete, and it wouldn't give you any output until its complete - so just be patient. After this test, try again by reading it:

dd if=/raid/zerofile.000 of=/dev/null bs=1m

Warning: using the dd command may be dangerous if you make typing errors or the commands are erroneous or malicious. Always make sure you have a good backup before experimenting with your live data. That said, the commands above should be harmless; after the test you can remove the testfile with this command:

rm /raid/zerofile.000

The output of both dd commands include the throughput speed, these should be accurate with this benchmark.




thanks, that's exactly what I'll try. It has to be a driver issue. I believe it was a false reading.. That will prove it.
m
0
l
March 18, 2010 3:57:34 AM

ok, rad thanks for the terminal speed check.. I only did 10Gb instead of 20 though..

here are my results:

internal sata 2x1tb samsung Apple Software RAID 0:

write:

10000+0 records in
10000+0 records out
10485760000 bytes transferred in 47.720346 secs (219733528 bytes/sec)

read:

10000+0 records in
10000+0 records out
10485760000 bytes transferred in 44.917022 secs (233447355 bytes/sec)


Crucial SSD boot drive ( I only did 4 gb ):

write:

4000+0 records in
4000+0 records out
4194304000 bytes transferred in 28.086539 secs (149335025 bytes/sec)

read:

4000+0 records in
4000+0 records out
4194304000 bytes transferred in 0.543718 secs (7714118479 bytes/sec)






external esata 1.5Tb RAID 5:

write:
10000+0 records in
10000+0 records out
10485760000 bytes transferred in 57.396308 secs (182690497 bytes/sec)

read:
10000+0 records in
10000+0 records out
10485760000 bytes transferred in 78.999019 secs (132732787 bytes/sec)




? do the speeds seem a little off?



m
0
l
!