Power failures and SSDs - Help Requested

[Patrick]

Prior to trying to contact some of the SSD vendors, I was wondering if anyone knows exactly the behavior of the super capacitors that are found on enterprise SSDs.

The question is basically, are they there to provide a few seconds of power to allow the controller to finish writes directly to NAND from the DRAM buffer (transitory data from the host to the NAND), or is there a logic which is flushing other non-transitory data to NAND?

 

[geezlouise14]

From what I've read the SSDs with supercaps are supposed to have enough juice to complete any queued writes. I think Intel had serial supercaps to mitigate risk of one failing.

 

[john4200]

The Intel 320 series do not have super capacitors. Just six 470uF capacitors, which comes to a total of just under 3mF. In contrast, the Sandforce enterpise SSDs have a single 90mF super-capacitor. I like the Intel design better, since conventional capacitors are less expensive, and having six in parallel eliminates a single point of failure (single supercapacitor), since the Intel SSDs power-loss protection could still function with five or maybe even four intact capacitors.

http://newsroom.intel.com/servlet/J...eries_Enhance_Power_Loss_Technology_Brief.pdf

During an unsafe shutdown, firmware routines in the Intel SSD 320 Series respond to power loss interrupt and make sure both user data and system data in the temporary buffers are transferred to the NAND media.