[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Swap space choice on a SSD <- Current best practice on?



On 2/13/19 1:23 PM, Dan Ritter wrote:
David Christensen wrote:
On 2/13/19 6:11 AM, Dan Ritter wrote:
If you want maximum SSD longevity, increase the amount of space that
the SSD can use for remapping by never writing to some amount of
space. Easiest is to not fill the disk with partitions -- leave 5-10%
empty.

AFAIK over-provisioning has no effect on longevity -- longevity is
proportional to total number of cells times rated erase/ write cycles
per cell divided by write throughput.


But, over-provisioning can improve write performance:

https://en.wikipedia.org/wiki/Write_amplification#Over-provisioning

Effective longevity depends on the number of times a block is
erased and written to. You can reduce that over the course of an
SSD's lifetime by limiting the space you think you have
available.

From the article you cited:

"Over-provisioning often takes away from user capacity, either
temporarily or permanently, but it gives back reduced write
amplification, increased endurance, and increased performance."

Increased endurance is increased longevity.

SSD manufacturers taking away part of the capacity and then telling us the drive has increased endurance measured against the reduced size is marketing speak. (Similar rant for kB, MB, GB, etc..)


I use the term "over-provisioning" in the context of this mailing list -- e.g. what we can do as Debian users (and system administrators):

"Furthermore, if any SSD is set up with an overall partitioning layout smaller than 100% of the available space, that unpartitioned space will be automatically used by the SSD as over-provisioning as well."


Whether you partition the first 90% of an SSD and write X blocks at random intervals over some time period T, or partition 100% and do the same, the number of Program/ Erase (P/E) cycles will be the same. (But the timing/ performance of clustered writes may differ.) When X gets large enough, the drive will eventually fail. I would call that X the endurance of the drive. (Intel converts it to a MTBF of 1,200,000 hours for my Series 520 SSD's.)


So, over-provisioning *by Debian users* has no effect on longevity.


David


Reply to: