• 0

S.M.A.R.T end of lifetime (SSDs)


Go to solution Solved by flyride,

Question

I used two used Samsung EVO 256 GB SSDs as SSD Cache for about one year and a half. Now DSM notified me about the SSDs estimated lifetime reaching end. When looking at the values of my SMART tests I am unsure what to believe. As far as I know DSM predicts the lifetime based on the Wear_Leveling_Count. However when comparing the SMART values those two SSDs makes it unclearer for me.

 

DSM reports this SSD is OK, although the "raw value" is way higher than the limitthreshold

1485232379_Bildschirmfoto2021-03-19um21_18_50.thumb.png.9d1bdd964868089aba8559bc37114fae.png1975565806_Bildschirmfoto2021-03-19um21_18_10.thumb.png.60d03aadc2625d2cf12875839a5699fa.png

 

 

 

whereas this SSD should be replaced

439377240_Bildschirmfoto2021-03-19um21_19_04.thumb.png.3a3cd7e0da97aed433a6d1a260ffde46.png

852375350_Bildschirmfoto2021-03-19um21_19_32.thumb.png.cb794dc6b1200d852a78b72f78dce3b5.png

 

So is there a possibility that DSM or the SMART test isn't reading the values properly or am I misinformed?

Link to post
Share on other sites

7 answers to this question

Recommended Posts

  • 1
  • Solution
Posted (edited)

Raw value does not correspond to anything, it's unique to the manufacturer.

 

The value you are concerned with is the "Worst" column.  Your disk 5 is at 71% life (29% used).  Your disk 6 is at 4% life (96% used).  The disks have not failed SMART so they are healthy, but DSM is rightly warning you that one of your SSD's has the potential of expiring very soon.

 

SSD cache is very hard on consumer drives that have very low TBW ratings.

Edited by flyride
  • Thanks 1
Link to post
Share on other sites
  • 1

On AmigaOS there was something called "buffers" that was basically ram allocated space that user could set for disk.
In those days hard drives had very limited cache and adding buffers did speed up HDD access.
 

To be honest I don't know how it's organised in modern OSes since HDDs has their own cache now.
But here we have situation where actually we need kind of additional cache, so maybe it would be good idea to go back to that old solution?
You could create  ram disk and maybe you could allocate it as SSD cache?
Actually Synology DSM may not permit to use something that's not typical SSD, so maybe it would let use something like DDRive which is PCIe SSD based on DDR memory.
This kind of solution would have no r/w cycles limit. Of course the size can be rather limited.
I don't know if there is any newer/other RAM based SSD solution, but I hope this idea won't be forgotten.

Edited by amikot
  • Like 1
Link to post
Share on other sites
  • 0

Thanks, now I get it. I know my SSD choice wasn't the best but it was at the beginning for testing purpose only. What du you think, is it better to replace the drive with a synology  SAT5200 or would it be enough to buy a samsung pro ssd?

Link to post
Share on other sites
  • 0

I'm not a big fan of SSD cache myself, and definitely not r/w cache which has resulted in many corruptions.  I think you get most of the bang for your buck from adding lots of RAM instead.

 

But if you really want to do it, I'd suggest any enterprise-class drive with higher write ratings. Write ratings are represented by DWPD (drive writes per day) or TBW (total bytes written) ratings.  TBW can be converted into DWPD and vice-versa if you know the warranty period and the drive size.

 

The higher, the better.  Assuming your 256GB Samsung EVO is the 860 model, it has 150 TBW and a 5 year warranty.

The enterprise equivalent from Samsung is the 240GB PM883, which has a 3 year warranty and 1.3 DWPD which translates out to 342 TBW.

 

Micron 5300 PRO offers 438TBW in the 240GB size.

  • Thanks 1
Link to post
Share on other sites
  • 0
Posted (edited)
4 hours ago, amikot said:

On AmigaOS there was something called "buffers" that was basically ram allocated space that user could set for disk.
In those days hard drives had very limited cache and adding buffers did speed up HDD access.
 

To be honest I don't know how it's organised in modern OSes since HDDs has their own cache now.
But here we have situation where actually we need kind of additional cache, so maybe it would be good idea to go back to that old solution?
You could create  ram disk and maybe you could allocate it as SSD cache?
Actually Synology DSM may not permit to use something that's not typical SSD, so maybe it would let use something like DDRive which is PCIe SSD based on DDR memory.
This kind of solution would have no r/w cycles limit. Of course the size can be rather limited.
I don't know if there is any newer/other RAM based SSD solution, but I hope this idea won't be forgotten.

RAM disks aren't new to me, but I didn't know that DSM supports it. I already found a guide on how to implement it. I guess for read-only it would be fine but the SSD adviser reported that I have 25GB of "hot" and 1TB of "warm" storage. Unfortunately the recommendation for a SSD cache is 170GB and my CPU only supports 64GB RAM total.

 

I will buy Samsung PM883 SSDs and will let them run in RAID1 (with checked "skipping sequential I/O" box).

Edited by smileyworld
Link to post
Share on other sites
  • 0
21 minutes ago, smileyworld said:

 I already found a guide on how to implement it. I


Really ? could you share the link, please? I'm googling, duckduckgoing and binging and nothing found ;)

Link to post
Share on other sites
  • 0
The question(s) in this topic have been answered and/or the topic author has resolved their issue. This topic is now closed. If you have other questions, please open a new topic.
Link to post
Share on other sites
Guest
This topic is now closed to further replies.