...
A Wikipedia article
is hardly authoritative.
had this as the first citation. I've copied the whole thing,
> because I know how you like details.
I like accuracy and conciseness, Paul, I am usually bored by your verbosity and unwillingness to trim irrelevant text. Yours is like top posting, trying to check the advice you are giving is laborious, but there is nothing technically wrong with your posts and I appreciate your attempts to help me.
formatting link
Unfortunately, the referenced MTBF FAQ link to IBM's web site is dead.
In particular, look at the first sentence in the third
> paragraph.
I think you are trying to point somewhere else, Paul.
In terms of the "bathtub curve"
The term "bathtub curve" is common and easy to understand, Paul. I think it usually refers to solid-state devices. Your unquoted terms "wearout phase" and "wearout mechanism" below are not nearly as common.
MTBF is computed based on the flat section in the middle of the
> "bathtub", and is not based on the wearout phase (end of the
> bathtub).
Unfortunately, Paul, there is no such terminology "wearout mechanism" in the document you referenced above.
The MTBF doesn't take the wearout mechanism into account.
Unfortunately, Paul, there is no such terminology "wearout mechanism" in the document you referenced above.
I would be surprised to learn that solid-state device testing is stuck in the Stone Age, Paul, as apparently you are suggesting.
We know if you do enough writes (like leave a write benchmark
> running by accident), it will wear out and die on you, and in an
> interval much shorter than the MTBF quoted.
I see, Paul... Leaving a write benchmark running is the "wearout mechanism".
Again... To be clear, I am not saying that SDD MTBFs are accurate. I am saying that the current supportive argument for the allegation that SDD drives fail prematurely sounds like bullshit. Every Dick on the planet has a sensational claim...