Percent Lifetime Used vs. Total Host Writes

Kilobyte Kid

Percent Lifetime Used vs. Total Host Writes

I have two M500 240GB SSD's set up as a RAID array. using CrystalDiskInfo, one shows:

- 92 "Percent Lifetime Used" (which measures % life remaing)

- 9.9TB Total Host Writes

 

The other shows:

- 59 "Percent Lifetime Used"

- 8.7TB Total Host Writes

 

The M500 spec is 72TB total writes.

How can the second have 59% life remaining with fewer host writes, and how can 8.7TB writes of 72TB limit equal 59% life remaining?

 

Thanks in advance.

5 Replies
Bit Baby

Re: Percent Lifetime Used vs. Total Host Writes

What is host write? Is the PC (presumably, the host) telling SSD to write? In that case maybe it matters where and how you write. For example, write the entire SSD from start to end, would be 240GB host write, each block written once. Write one and same block 240GB times, that would still be 240GB host write, but SSD would relocate blocks internally. So while the host wrote 240GB the SSD internally might have written much more due to relocations, amplifications, overheads, and maybe that would not show up in the host writes but still show up in as rate percent life used.

 

I do not know if this is the case. Just a guess.

JEDEC Jedi

Re: Percent Lifetime Used vs. Total Host Writes

Yeah, it's very difficult to quantify drive life in terms of quantity of data written.  For all the reasons zapper mentions!

 

Unfortuantely, the drive wear percentage that you quote doesn't really bear any relation to a quantity of data written.  Wherever such a quantity is quoted in a spec, odds are there'll be a disclaimer about it representing a typical usage pattern or something.

 

The drive wear percentage actually relates to the the average number of erases per block as a percentage fo the rateed number of erases for the drives type of NAND.  It's erasing that wears out drive NAND and erasing has to occur before NAND is re-written. So there is a correllation between writes and drive wear but not so much with regards to the quantity of data written.

 

For example, the erase block size on the MX300 is 24MB ( I can't remember what it is on an M500).  So to write *anything* to the drive requires 1 x 24MB block to be erased.  So worst case, a 1 bit write could potentially cause as much wear as a 24MB write.  Obvbiously, it's not going last 72TB 1 bit at a time!  That would be a pretty extreme usage example though (although database servers will kill them pretty quick with many small writes which is why consumer ssd's shouldn't be used as server drives)

 

The odd thing about your scenario is you would expect that in a RAID the wear would be mroe or less equal.  Are they the same firmware version?

______________________________________

Did a user help you? Say thanks by giving Kudos!
How do I know what memory to buy?
Still need help? Contact Crucial Customer Service
Remember to regularly backup your important data!

JEDEC Jedi

Re: Percent Lifetime Used vs. Total Host Writes

[ Edited ]

targetbsp wrote:

 

 

The odd thing about your scenario is you would expect that in a RAID the wear would be mroe or less equal.  Are they the same firmware version?


Actually the thought occurs that if we're talking stripped rather than mirrored, it would depend on your stripe size.  Small files, the ones we already mentioned are worse for SSD wear*, would only hit the first drive I guess.

 

 

*though ironically the thing SSD's are much better at performance wise than HDD's.

______________________________________

Did a user help you? Say thanks by giving Kudos!
How do I know what memory to buy?
Still need help? Contact Crucial Customer Service
Remember to regularly backup your important data!

Kilobyte Kid

Re: Percent Lifetime Used vs. Total Host Writes



targetbsp wrote:

The odd thing about your scenario is you would expect that in a RAID the wear would be mroe or less equal.  Are they the same firmware version?


Actually the thought occurs that if we're talking stripped rather than mirrored, it would depend on your stripe size.  Small files, the ones we already mentioned are worse for SSD wear*, would only hit the first drive I guess.


They're mirrored drives, not striped, and they are the same latest firmware version, MU05, so the unequal wear doesn't make sense.

Support suggested checking SMART values using Micron Storage Exec. Software after removing the drive(s) from RAID, but the values were the same as Crystal DiskInfo's, 59% life remaining for the most impacted drive.

Kilobyte Kid

Re: Percent Lifetime Used vs. Total Host Writes

The Micron Storage Exec Software values were the same as Crystal DiskInfo's, so support suggested I RMA the SSD.