• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Battery/Supercap backed DRAM cache SSD

S

s0rce

Limp Gawd
Joined
Jan 17, 2011
Messages
495
It seems that there is some demand for battery backed SSDs that can be safely used as ZIL/SLOG in ZFS pools without the risk of loosing data/the pool if power is lost during writing and the DRAM cache isn't written to the non-volatile memory.

From commercial sources these drives cost a significant price premium over standard (non battery backed SSDs) and put them out of reach for many.

I was wondering if it might be possible to design a small battery backup circuit hat sits in-line with the drive between the drive and the SATA power-connector, effectively a mini-UPS with a small battery and charging circuit and voltage regulator to provide 5 & 12V in the case of power failure. Would this actually work? Would the data be written automatically from the cache to the NAND or does the drive need different firmware to know what to do without any commands from the computer (since it just lost power).

It seems that if a simple solution was possible it could be built and sold for probably under $50 and might be an interesting solution for Home/SOHO NAS systems where the price premium of ready-made battery backed SSD or battery backed DRAM drives is too much.

Does anyone know if this is possible, is there interest? Thanks!

TL;DR: can you power an SSD with a simple ups-like battery backup if the computer power fails and have it save the data in the volatile cache?
 
I think you're main hurdle will be notifying the controller that the SSD just lost power and it needs to flush its cache.. Otherwise, the SSD won't know any different and still keep everything in volatile memory util the next flush cycle.

I guess, if you knew what the time was between flush cycles you'd just have to keep the SSD up and running for that amount of time.. Assuming it didn't freak out by the loss of the PC on the SATA port.

Riley
 
FYI, the Intel 320 series SSDs have capacitors. You might want to look into those.
 
s0rce said:
from this thread: http://hardforum.com/showthread.php?t=1577141

it seems that the intel 320 might not be fast enough to gain performance when used as a ZIL.
Click to expand...

I think you have misread the situation. If the larger Intel 320 SSDs are not fast enough, then probably no SSDs are fast enough. But I don't think that is the case. It may be true that the 40GB Intel 320 is too slow, though -- its sequential write is very low.
 
I think I've pretty much given up on an SSD for ZIL - at least in regards to the current generation of SSDs. To get the performance you need an SSD has to be ~240GB or higher (which is 232GB too large in my case) and even then you suffer from greatly decreased IOPS after it gets used for a while (still better than a traditional hard drive though). Maybe once TRIM is introduced into ZFS for the ZIL then it will be better.

At this point, I think battery-backed RAM is a better solution.

Riley
 
You must log in or register to reply here.
Back
Top