ICYCube MB561U3S-4S 4 Bay 2.5″/3.5″ USB 3.0/eSATA External Enclosure
Starting at the rear we see a standard power connector, Kensington Lock port, three position fan speed control switch, eSATA, and USB 3.0 ports.
The fan can be easily removed for cleaning and maintenance. Our fan had a bit of bearing noise (regardless of speed) that cleared up a little after letting the unit run for ~30 minutes. Keeping the fan removed is not recommended with 3.5" HDDs installed as their power draw will eventually cause things to run very hot without active airflow.
The drive trays come with mounting holes for both 3.5" and 2.5" drives, as well as two bags of necessary screws to mount four of either type.
Here's what the ICYCube looks like while installing a drive.
Here's a look at the indicator lights, which light up white during drive activity (off for no activity).
Performance
As this is a straight JBOD enclosure with no hardware RAID capability, all drives appear individually. If you wish to RAID them, you will have to do so with your OS. Here are the results of a single WD Red 6TB HDD:
We chose to test under USB 3.0 only as eSATA ports have been dropped from most shipping laptops and desktop motherboards. Windows simple Dynamic Disk RAID does not support USB connected devices. We could have gotten around this if the ICYCube supported UASP (UASP devices mount as SCSI connected), but since it does not, our only option was to test using Microsoft Storage Spaces. Here were the results for a simple two drive mirror:
…and here is a two drive striped array:
As we can see, Storage Spaces can only do so well when connected to a non-UASP USB 3.0 device, essentially limiting itself to single drive performance. We would have seen better results here using Windows Dynamic Disks and eSATA, but again, that is a rare and dated connection method at this point.
Finally we repeated our test of a single Samsung 850 Pro:
Good speeds seen from a single SATA channel of the ICYCube here, especially over USB 3.0 and without UASP.
UASP is interesting. How do
UASP is interesting. How do they increase throughput? By lowering overhead? 5 Gb/s translates to 625 MB/s. Even with the fastest 2.5″ drive in the world, it managed to cap at 440 MB/s at a not practical high Q depth.
I remember PCIe Gen 1 and 2 had a 20% overhead due to poor encoding. They decreased the overhead to roughly 1.5% with better encoding algorithms. That, of course, increases power consumption.
Can’t they do something similar to that with USB? I understand it has to jump through more hoops to get to the processor since it’s not a direct bridge like PCIe 3.0 (All Intel CPUs, some AMD CPUS) is but surely they can do better than that.
Same story with SATA 3. 750 MB/s theoretical throughput but in practice it hardly ever goes over ~550 MB/s. That’s 35% overhead.
Future SoCs from Intel will integrate PCH into the die, eliminating the need for DMI. Maybe then we’ll see more efficient data transfer rates.
I looked at these today,
I looked at these today, before I knew you had done a review Allyn,
Bit too pricy for me though,so I poped these into my whish list instead
http://www.amazon.co.uk/dp/B00JX7XTOK/
http://www.amazon.co.uk/dp/B005OFID5A/
Sorry it’s UK, and one is unavailable, but it’s around half the price of the MB662U3-2S
Have you had any experience of ORICO?
EDIT:
Actually reading around, they seem a little low quality, however I just managed to snag “ORICO 6619US3” a single docking station for £6.01 inc del from amazon, so will try that out.