|
Performance:
Next, we will put the NAS through the ringer and
test all three RAID modes for speed. In order to test the
speed, I used a 1080p .mkv file that measures 10,621,635 kilobytes
in size. A friend of mine was kind enough to write up a Perl
script that initiates the file transfer, and spits out both the time
of transfer and the average transfer rate. Gigabit Ethernet
connections were used throughout; from the NAS itself, to the
DGL-4500 router, to the Gigabit connection on the DFI P35-T2R
motherboard.

|
RAID 0 |
JBOD |
RAID 1 |
| Write
Speed |
13.911
MB/s |
17.502
MB/s |
17.33
MB/s |
| Read
Speed |
7.488
MB/s |
6.423
MB/s |
6.378
MB/s |
| Write
Time |
745.66
s |
592.656 s |
598.51
s |
| Read
Time |
1385.323 s |
1614.944 s |
1626.247 s |
The first test was in RAID 0. For those
unfamiliar, RAID 0 promises to increase performance and partition
size by splitting the writing duties between both drives. A
RAID 0 partition will be equal to the size of all the physical
drives combined, and parts of each file will be written on each
drive. This RAID mode is a misnomer, as there is no
redundancy; if any disk in the array dies, all data is lost.
As we can see here, the write speed was actually the slowest of the
bunch, though the read speed was the fastest. Writing the file
took about 12.5 minutes, and reading the file took about 23 minutes.
The second test was in JBOD mode. JBOD is
technically not a RAID mode, as it also does not provide any
redundancy; it just takes the disks in the array and joins them
together. So, like RAID 0, the available size is equal to the
cumulative size of the disks in the array. Unlike RAID 0,
however, the files that are written are only written on one of the
drives, so the overall risk of data loss is reduced. The JBOD
speeds were very comparable to RAID 1 in this test, as writing to
the array took 9.8 minutes, while the read time took nearly 27
minutes.
The final test was in RAID 1. RAID 1 offers
full redundancy, which means all data is fully copied to both
drives. This means that if a drive dies, no data is lost.
The drawback is, you only have the capacity of a single drive
available to you. The speed test for RAID 1 was on par with
JBOD, with the write time taking 9.8 minutes and the read taking
27.1 minutes.
All tests exceeded the theoretical limits of a
100Mb Ethernet connection, which is around 12.5MB/s, so we can see
the value of the Gigabit connection here. Surprisingly, RAID 0
offered the worst write performance, and the best read performance,
but nothing that makes it sufficiently better over JBOD. The
oddest thing, however is that the read speeds across the board were
much slower than the write speeds; typically it is the other way
around. However, retesting shows these results to be
consistent.
The next test I performed was to simulate a drive
failure in a RAID 1 configuration. I copied some files to the
RAID 1 array, and while it was still powered on, I pulled one of the
drives out. I took that drive and formatted it in a different
PC and plugged it back in to see how it would react. It
immediately began repairing the array, and the status could be
monitored under the "Disk" page on the Configuration URL.
Overall, rebuilding the array took about 3 hours.
Conclusion:
So there we have it for the Muse X-Duo RAID NAS
box. In general, I feel the device has a lot of potential,
however there are still improvements to be made. The firmware
support for Vista is still a bit immature, in my opinion, and I hope
Thermaltake continues the trend of releasing updates in a
timely manner to improve support. Until this happens, however,
I have to knock the NAS device for stability.
The software package is just how I like it: lean.
The web configuration is feature packed yet still simple, and all
the features there worked like expected, though the non-Java
discovery tool still needs to be fixed. Since this is a
feature added with the new firmware, I won't knock it too badly for
that.
As performance goes, the write speeds in all
modes are fairly impressive, but the read speeds are apparently lacking.
I say this because most reviews use a 700MB file as a transfer
medium to test speeds, and I chose a 10GB file to represent the
potential for HD movies or game images. It appears that file size
plays a huge role in transfer speeds, as both a 150MB and 700MB file
will both write and read at about 25-30 MB/s in RAID 1, while both
write and read speeds suffered with the 10GB file. Since these
large file sizes are going to be a more accurate representation of
usage in the future, I am hoping other review sites will start using
them as well. Because of this, I gave performance a pretty
high score.
Note: Club Overclocker is
now using a new rating system based on a score of 1 to 5.
Please go to our rating system page for more information.
|
 |
| Performance: |
4 out of 5 | |
Innovation: |
4 out of 5 | |
Quality: |
4 out of 5 | |
Stability: |
2 out of 5 |
| Aesthetics: |
3 out of 5 | |
Software/Drivers Pack: |
4 out of 5 |
|
Overclocking: |
N/A |
|
Value: |
3 out of 5 |
|
 |
 |
|
 | Project Skill Level (5 being most difficult) | 3
out of 5 |

|