.
OKay, so a preliminary evaluation of the Rosewill enclosure. Received this afternoon and was able to play with it for a couple of hours tonight. Everything applies to a Win XP Pro system with all current patches.
First off: JBOD. This mode, does indeed, mount the two hard drives as completely separate entities.
You can place a single drive in this enclosure and it will work. Today I used drives that had already been formatted and filled with data. Drives slip right into place beautifully, right onto their SATA data & power connectors without any undue stress. In fact, I was concerned they might not be tight enough connections, they go on and off so smoothly.
I used two drives: one a Samsung 2 TB as described above, the other a Western Digital Caviar Black 750 GB as described above ... except I got the model slightly off. Mine do not use the SATA 6.0 GB/sec specification .. they use the 3.0 spec (SATA II or SATA 300). The Samsung is 5400 rpm, the WD is 7200 rpm, both have 32 MB cache.
My system is a few years old. It uses the first SATA I specification (SATA 150). Actually, it primarily relies on PATA-100 and "just happens" to have two SATA 150 connections as an option. My PATA drives have been failing and giving me other fits & serious issues, thus one major reason for this project: to get away from the PATA drives completely.
So, the Samsung drive auto-detects. It is a SATA II spec drive, but when connected to a SATA I controller it auto-adjusts its data rate. The WD drives need to be jumpered to limit them to SATA I data throughput. It's especially important because I use them as boot drives and hard-forcing the specification reduces confusion by the BIOS & operating system.
Important why ..? This evaluation allows me to figure out how to transfer ~2 TB of data by the best method, close to 800 minutes of time saved per TB, or well over 20 hours. Damn, but I sure do wish I was using a USB 3.0 system .... :'(
Here's the speed tests. The WD drive has a jumper in place when installed as an internal drive, but the jumper is removed when it is installed in the enclosure. The enclosure auto-detects between the SATA 3.0 & 6.0 GB/sec specifications, and I could not find any reference to it handling the SATA I spec ... so, figured I didn't want the WD drive limited for at least that reason.
I used a torrent directory:
The Twelve Kingdoms because of its nice size (15.31 GB/ 16,441,679,049 bytes) and it's broken down into 45 files approx. 352-367 MB each, ~355-356 being most common. I also used four photo directories (15.4 GB/ 16.623,070,358 bytes) that consist of 3,953 files approx. 4,700-5,900 KB each, ~5,000-5,300 being most common.
Data transfers are more efficient/faster when in fewer large files than when in many smaller files. These show the trend. It is casual: I did not perform duplicates in true scientific/statistical fashion, and my timing could be off by ~3 sec since it was all manual. The "enclosure" is when the drive is mounted in the external hard drive case, "internal" is inside my PC using the SATA controller on the motherboard.
from enclosure (Samsung 2TB) > internal drive (WD 750 GB)
~10:15 min (12 Kingdoms)
~12:26 min (pictures)
from enclosure (WD 750 GB) > internal drive (Samsung 2TB)
~9:47 min (12 Kingdoms)
~12:53 min (pictures)
from internal drive (WD 750 GB) > enclosure (Samsung 2TB)
~11:42 min (12 Kingdoms)
~14:30 min (pictures)
from internal drive (Samsung 2TB) > enclosure (WD 750 GB)
~11:56 min (12 Kingdoms)
~12:29 min (pictures)
from internal drive (Samsung 2TB) > internal (WD 750 GB)
~5:23 min (12 Kingdoms)
~5:49 min (pictures)
from internal drive (WD 750 GB) > internal (Samsung 2TB)
~4:34 min (12 Kingdoms)
~7:05 min (pictures)
from enclosure (Samsung 2TB) > enclosure (WD 750 GB)
~20:01 min (12 Kingdoms)
[N/A] (pictures)
from enclosure (WD 750 GB) > enclosure (Samsung 2TB)
~20:06 min (12 Kingdoms)
[N/A] (pictures)
All in all, no real surprises. It's all in line with USB 2.0 transfer speeds. Weird aberration: the 4:34 & 7:05 times when transferring from the WD to the Samsung when both drives were mounted internally. I think there was some sort of data bottleneck to cause the 7 min time; I think it should have been about 6 or slightly under. I was watching the CPU usage and noticed a couple big spikes that coincided with delays in file transfer. Actually, fairly normal activity that can be observed when watching USB 2.0 transfers.
Interesting in that it seems a trend: when transferring several large files from the WD to the Samsung, I get my best times ... but I pay for it when transferring many smaller files from the WD to the Samsung; the Samsung to WD transfers seem more consistent. Must be an interesting caching/hardware configuration going on between them. It'll be interesting when I can try this between two Samsung drives.
Transfers between the enclosure and internally mounted drives used the most processing power. I have a 3.0 GHz P4 duel core, which settled in about 30%-35% load across the USB connection. When transferring across the internal SATA controller, the load dropped by about 5%-10%, bouncing around 20%-30%. When transferring data between drives when they were both mounted in the enclosure, CPU load dropped about another 5% ... poor compensation for such a miserable data transfer rate.
Generally speaking, my system was tweaked to run at fairly minimal load. No major apps, not even web surfing or WinAmp. I was running utorrent, but CPU load was steady at 3%-5% before any data transfers started. Memory use was nominal at ~320 MB.
The greatest disappointment was: I had hoped that when two drives were mounted inside the enclosure in JBOD mode that they might transfer data between themselves at the SATA II specification. Instead, I got the worst of all worlds: a miserable half-speed of the USB 2.0 specification. Clearly the external case routes the data through the motherboard, then use the USB to route it back to the case again, thus doubling transfer time. That's also why I didn't bother transferring the picture directories on the last set; it probably would have taken 30 or 40 minutes for each.
I'm really pleased with the enclosure. It is solid, brushed black anodized aluminum. It has substantial mass; I was surprised & happy since it will provide nice heat sink capacity. The enclosure uses a small fan, which is reasonably quiet, but not silent. I don't expect any heat problems; I don't have a utility handy that captures the drive temp over the USB connection, but the case remains cold to the touch after several hours of continuous data transfer.
The greatest annoyance with the case is the way it mounts the drives to the system. While it is powerful -- it treats the drives as essentially mounted internally, so formatting, partitioning, etc. are easy -- it does NOT allow use of the Remove Hardware utility with Windows. The drives remain "In Use" and the only ways to 'unmount' the case are to
1) power the computer off to a cold boot or
2) turn the power switch off. Neither attractive options. I suppose
3) jerk the USB cable out of the socket ... but that runs the highest risk of unpleasant side effects.
Watching movies/anime, listening to music and torrenting (all at the same time, even!) over the USB connection to directly stream data off the drives works perfectly, as expected.
So, I got what I wanted: large, slow-ish storage that seems to be well-protected and reasonably versatile. When the other drives arrive, I'll have ~3.63 TB formatted capacity to archive things, in duplicate on physically separate drives for the unique, real important stuff (like my photos and live recordings). Not much more to the story anymore ... other than, I'll report back or respond to questions on how well it all holds up under use over time.
