Does persistence kills USB sticks?

Some claims (here and here) on the internet states that persistence reduces lifespan of the USB stick.

Is this true?

If Yes, Why is this true?

How reduced is the lifespan?


Some claim on the internet that persistence reduces lifespan of the USB stick

Is this true?

  • Yes it is true, particularly compared to a cloned live-only boot drive with a read-only iso9660 file system, which is never written until a new Ubuntu system is flashed into it. I think it is also true, compared to a standard storage device, where you store files manually.

If Yes, Why is this true?

  • The flash memory cells of USB sticks have a limited number of write cycles to the expected end of life, and this number is lower than those of an SSD and the magnetic memory of a hard disk drive.

  • Persistence in an installed system as well as in a persistent live drive makes it possible to write to the memory.

    • A standard installed system writes a lot, but the writing can be reduced.

    • A persistent live system writes too, but it is designed so that it writes less to the USB stick than an installed system. It works with a file system, that resides in RAM.

  • Anyway, it is important to backup the persistent data regularly, if you want to avoid loss of data.


How reduced is the lifespan?

  • It is difficult to say how much the lifespan is reduced. It depends on the number of write cycles, and the time can differ a lot between use cases. I have persistent live drives that have been working for years, but I use them carefully. Only one of my USB pendrives has failed, and very early (not because of too many write cycles).

    But I know of several cases, where SD memory cards have failed, I think because of excessive wear, when they have been used as system drives in Raspberry Pi computers. These are normally installed systems, tweaked to reduce the wear. SD cards have the same kind of memory cells as USB sticks.

  • Also, the quality of the memory cells and the wear leveling methods have improved over the years, so warnings that are several years old would be exaggerated now.


Links,

  • Pendrive lifetime

  • Installed system - Final system tweaks

  • Boot Ubuntu from external drive

  • Can't format my usb drive. I have already tried with mkdosfs and gparted - Analysis of the problem


Using a flash drive with a persistence probably does reduce life span but does it matter?

A good flash drive is good for between 10000 and 1000000 writes. The time for one write to a 64GB flash drive is about ~ 64GB/50MB/s = 1200sec or 20min.

The time for 10000 to 1000000 writes is 200000 to 20000000 minutes or between 416 and 41600 8 hour days. (actually only a small portion of the time is generally used for writing).

At $25 for the flash drive that works out to a maximum of 6 cents a day.

Update 2018

NAND flash is coming out in different configurations that trade endurance for volume.

Following are some life expediencies for a 128GB flash drive based on 100 GB/day normal maximum use:

SLC NAND 100 000 Writes 128 GB/write x 100 000 writes -> 12 800 000 GB / 100 GB/day = 128 000 days ~ 350 years

MLC NAND 5000 - 10 000 Writes 128 GB/write x 5000 writes -> 640 000 GB / 100 GB/day = 6400 days ~ 17 - 34 years

TLC NAND 1000 Writes 128 GB/write x 1000 writes -> 128 000 GB / 100 GB/day = 1280 days ~ 3.5 years

QLC NAND 200 Writes 128 GB/write x 200 writes -> 25 600 GB / 100 GB/day = 256 days

Trying to find out what type each vendor uses is not easy, Sandisc claims to only use MLC, Intel seems to be rushing to QLC.

Based on 1TB drive with 5 year warranty. 0.1 DWPD ~ 100 GB/day normal maximum use.