7. Performance

In order to gauge performance, we ran our standard NAS test procedure using both 100 and 1000 Mbps connections. We couldn't run the gigabit tests with Jumbo frames, because the SSII doesn't support them, but we did run tests in tests in both Spanning and Mirroring modes. All tests were run using the version 3.1.2 firmware that came installed on the SSII.

Figures 16 and 17 show 100 Mbps write and read performance using a 64 kByte record size and file sizes from 32 MB to 1 GB, plotted along with comparable previously reviewed products.

Figure 16: 100 Mbps Write test (click to enlarge)

100 Mbps write results show that both Span and Mirror (RAID 1) modes perform better with smaller file sizes with a fall-off in speed that flattens out at 128 MB and larger files. Read performance is flat across file sizes with no difference in span and mirror mode reads. The Synology DS-106e has essentially the same read performance profile.

Figure 17: 100 Mbps Read test (click to enlarge)

Write performance with a gigabit LAN connection (Figure 18) shows the same performance falloff from small to large file sizes. But this time, read performance (Figure 19) shows a falloff between 32 and 64 MB files. The Synology is again the closest in performance for write, but is outpaced for read.

Figure 18: 1000 Mbps Write test (click to enlarge)

Figure 19: 1000 Mbps Read test (click to enlarge)

You can get a comparison of the SSII against our entire database of NAS products by using our NAS Charts.

To get a different perspective on performance, an interactive file-copy test was run in order to see how the SSII might perform in real-world use. These tests were done with a different computer than the iozone tests. I used Windows XP SP2 running natively on my MacBook Pro, a 2 GHz Intel Core Duo with 1.5 GB of RAM.

The results in Table 1 below are averages of multiple runs of moving a directory tree containing 4100 files using up just over a gigabyte. It's interesting to note that the speed difference between read and write is negligible because normally I see read from a NAS performing much faster that the writes. This may be due to the fact that the NTFS partition I'm using on my MacBook Pro is fairly full. The test may have put some stress on my local filesystem when writing the 4100 files.

Table 1: File copy test results