Supermicro: 6047r-e1r72l 72 drives in 4U

I think that's a typo, because it says 72 3.5" drives but then only counts 24 in the front and 12 in the rear and details space for 2x 2.5" drives on the inside.
 
Not a typo

72x 3.5" SAS2/SATA3 HDDs in 36x
(24 front + 12 rear) Hot-swap Drive
Bays (2x HDD per bay w/ LEDs for
each drive)
 
awesome
now lets just wait for 90x jbod version

and pricing on both of em
 
If you mouse over the picture from the link above, you can see where two drives fit in the tray. Another thing to note is the server is 35" in depth, where your typical Dell PowerEdge 1950 is 31.5" in depth, HP DL380 Gen7 is 27.25" in depth, IBM X3550 is 29" in depth.

That chassis is very deep, but is definitely very dense (and probably hot) storage. That chassis with the motherboard is around $2300 on ebay.
 
I actually have a slight [H]ard on seeing that much hardware and hard drives in such a form factor.
 
Im just wondering if its possible to add 3 24 ports arecas and directly connect to them or that it comes with backplanes. Cant find that on their website
 
Im just wondering if its possible to add 3 24 ports arecas and directly connect to them or that it comes with backplanes. Cant find that on their website

It uses backplanes (all of the SM servers like this do afaik) -- if you click "parts list" at the bottom of the page it shows you the model and information about the backplanes.
 
You can't really hotswap drives then, because to replace one you have to disconnect two.
 
It uses backplanes (all of the SM servers like this do afaik) -- if you click "parts list" at the bottom of the page it shows you the model and information about the backplanes.

Yea i know but what i mean is if it comes only for Sas with an expander or that it comes for sata disks too like the 847A that you can connect your sff cable directly. Sas drives are expensive :p
 
SSG-6047R-E1R72L_1.jpg

SSG-6047R-E1R72L_2.jpg
 
That is my biggest concern with this.

Is it possible that each bay (holding 2 drives) is treated as 1 drive on the backplane? Therefore, you would still only be disconnecting 1 drive that is presented to the array?
 
two thoughts...

Pulling two disks to replace one failed is not ideal, but with this density wouldnt you be running enough redundancy to have the ability to pull two disks without upsetting the raid

less likely, do the trays come all the way out like a normal hotswap? or do they have some clever engineering which allows the tray to be extended with the disks still connected, you can then pop the failed drive without upsetting the other
 
Pulling two disks to replace one failed is not ideal, but with this density wouldnt you be running enough redundancy to have the ability to pull two disks without upsetting the raid

When I put the admin hat on I try to avoid degrading any array if possible even though I have 1 or more tape backups. Although a semi recent addition to linux software raid would allow me to tell the raid array that I want both drives replaced and the raid would automatically mirror the disks I want to remove to spares provided I have enough spares. HW raid can also have this feature.
 
Last edited:
Is it possible that each bay (holding 2 drives) is treated as 1 drive on the backplane? Therefore, you would still only be disconnecting 1 drive that is presented to the array?

I believe it is possible with additional hw raid chips.
 
Sure there are ways to mitigate this, clearly for me (home use) it would just be shutting down the thing and cold swap. But if you use redundancy for this then you have to double your redundancy, not practical. A better way would be to make several arrays (or vdevs), putting each tray on two different ones, but still, it would have to be at least RAIDZ3 for comfort (imagine 2 drives failures in 2 different trays), and not sure you can do it with hardware RAID, depends on how the backplanes are made.
 
Back
Top