
We connect 5 hard drives via sata multiplier with one sata cable

In this article I will talk a little about the sata multipliers CFI-B53PM based on the SiI3726 chip, which I bought for a large burial ground at a lot of terabytes. How and where to order them I wrote in this article (at the end of the article I attached the tracking package). I will also talk about linux speed in the 6th raid.
Benefits
1) Requires less sata wires.
2) Requires less power wires.
3) The low cost of the multiplier and controllers for them.
4) Fast read speed.
disadvantages
1) Multipliers are not sold in Russia, although there are a lot of controllers on every corner ( sil3132 ).
2) Solid capacitors break off very easily (caution and straight arms are required).
3) Relatively low write speed.
According to silicon image , it works something like this:

To connect to a computer, I used a stlab a-410 controller worth 600 rubles.
That's how ubuntu is going to raid6 out of 5 drives stuck in the multiplier.
cat / proc / mdstat
Personalities: [raid6] [raid5] [raid4]
md0: active raid6 sdh1 [2] sdj1 [4] sdg1 [1] sdi1 [3] sdf1 [0]
8790795264 blocks super 1.2 level 6, 512k chunk, algorithm 2 [5/5] [UUUUU]
[==============> ......] resync = 73.0% (2140472708/2930265088) finish = 504.8min speed = 26074K / sec
Drives wd green 3TB , special alignment done in parted for 4k blocks (if not, terrible brakes will be observed in random operations).
The speed of writing to the 6th raid with the xfs file system starts at 100 megabytes and settles at 40 MB. Read speed is stable 320 MB.
The low write speed, I think, is associated with the command-based switching method of multiplication, in which at one point in time the controller can send a command to only one hard drive.
Well, now a bunch of pictures.
The front

controller The rear

controller The controller in the interior of the hard drive

We stick 5 wd 3tb green drives into the controller Controller


chip

We start the test operation

The stlab a-410 controller Parcel

tracking scan
