Wednesday, August 25, 2010

When will a solid state drive die?


Solid State Disk (SSD) drives are all the rage for dramatically increasing read performance by eliminating seek time and by enabling much higher throughput.  For example, an SSD drive can read at a throughput of over 200MB/sec compared to just 10MB for a regular disk drive.  One question that I don't hear many people talking about though is how do you tell when the SSD drive is nearing the end of it's life span?

The standard method that most IT professionals use to determine the state of a disk drive is to run a tool that reads the S.M.A.R.T. data from the disk drive.  In the case of a SSD drive, the key metric to watch is the "Reallocated Bad Sectors".  For example, I use the S.M.A.R.T utility for Mac/OSX systems.  After roughly a month of normal use, my OWC Mercury SSD drive reported that over 5,500 sectors (a.k.a. bad blocks or BB) had been relocated to the bad block relocation table.  a.k.a. the over-provisioned portion of reserved capacity of an SSD drive.  For a normal disk this would indicate that the drive is nearly dead.  In fact, my S.M.A.R.T. utility said that the drive was in a state of "FAILING".  But this is not necessarily the case for an SSD drive.

See update below for updated life span estimate.

In theory, the way that you tell if an SSD is near the end of its life span is if it has exhausted its BB reserve.  The logical next question is, how many sectors are in the reserved BB capacity of a given SSD drive?  In my case, my 240GB SSD drive has 16GB (e.g. (2^34)/(2^9) = 33,554,432 sectors).  This means that at the current rate of BB sector consumption of 5,500/month it will theoretically take over 500 years before the SSD drive reaches the end of its life span.  Somehow I don't think that the drive will last quite that long. ;-)

Hopefully as SSD drives become more prevalent in the market place S.M.A.R.T. data and corresponding utilities will be extended to show data that will assist in estimating the life span of an SSD drive.  For example, the following metrics would be very useful:
  • that the drive is an SSD rather than a normal disk,
  • the reserve capacity sector count,
  • the current BB sectors consumed, 
  • the rate of consumption, 
  • and most importantly of all an accurate diagnosis 
If you want to avoid BB sector math, the simplest way to tell when your SSD drive is reaching the end of its life span is when it starts slowing down dramatically.  This usually means that the BB reserve capacity is exhausted.

That is it for this blog post.

Have a great Day!

Brad

Important Update (11/1/2010)!!!
While doing some SSD LifeSpan analysis related to the rumored Third Generation Intel SSD devices, I learned that SSD life span is governed primarily by the volume of data randomly written to a device over time.  For example, the Intel SSD X25-M G3 life span is estimated to support between 30TB to 60TB of 4K random writes.

Further, some SSD controllers such as the SanForce 1500 controller used in my OWC SSD use SMART field 231 (see also CrystalInfo Health Status information) to report the estimated life left in the drive and uses SMART attribute 241 to reveal the total number of Logical Block Addresses (LBAs) written to the device since the last boot.

Given this new information, lets re-visit how long that my SSD drive has to live.

Here is the a recent SMART report for my drive.

Notice that the Reallocated Sector Count (Bad block relocation) value of 5504 is the same as it was when I first tested the drive.  However, SMART attribute 231 says that my OWC SSD drive has a value of 77.  I suspect that this is not the percentage of life left because it has been the same number for several months.   Further, SMART attribute 241 says that the drive has had written 1791 LBAs.  

Unfortunately, until the solid state vendors unify on a common set of SMART disk properties, it is going to be hard to tell exactly how much life is left in a solid state disk drive.

Have a great day!


Brad
If this information has helped you, please consider helping me through investing in your health and in the health of those you love through purchasing Mannatech wellness products. Ambrotose is the key ingredient of all Mannatech Wellness products.  Place your order at my Mannatech Web Store today.

3 comments:

Ross Harvey said...

Thank you for this post - I just ran the utility with the same 240G OWC SSD (a week old), and saw FAILING, due to the Reallocated Bad Sectors value 7552. I can relax again now :¬)

frankg said...

The math is simpler if you take the manufacturer warranty and add 1 day.

Dj Key$trike said...

I believe have a CORRECTION.

The article says "However, SMART attribute 231 says that my OWC SSD drive has a value of 77."

If you look at the picture. Attribute 231 (the leftmost column) is actually called "Temperature Celsius!! I'm not sure if this is a mistake just bringing to ur attention!