| 1 |
On Monday, May 4, 2020 3:19 AM, antlists <antlists@××××××××××××.uk> wrote: |
| 2 |
|
| 3 |
> On 03/05/2020 22:46, Caveman Al Toraboran wrote: |
| 4 |
> |
| 5 |
> > On Sunday, May 3, 2020 6:27 PM, Jack ostroffjh@×××××××××××××××××.net wrote: |
| 6 |
> > curious. how do people look at --layout=n2 in the |
| 7 |
> > storage industry? e.g. do they ignore the |
| 8 |
> > optimistic case where 2 disk failures can be |
| 9 |
> > recovered, and only assume that it protects for 1 |
| 10 |
> > disk failure? |
| 11 |
> |
| 12 |
> You CANNOT afford to be optimistic ... Murphy's law says you will lose |
| 13 |
> the wrong second disk. |
| 14 |
|
| 15 |
so i guess your answer is: "yes, the industry |
| 16 |
ignores the existence of optimistic cases". |
| 17 |
|
| 18 |
if that's true, then the industry is wrong, must |
| 19 |
learn the following: |
| 20 |
|
| 21 |
1. don't bet that your data's survival is |
| 22 |
lingering on luck (you agree with this i know). |
| 23 |
|
| 24 |
2. don't ignore statistics that reveal the fact |
| 25 |
that lucky cases exist. |
| 26 |
|
| 27 |
(1) and (2) are not mutually exclusive, and |
| 28 |
murfphy's law would suggest to not ignore (2). |
| 29 |
|
| 30 |
becuase, if you ignore (2), you'll end up adopting |
| 31 |
a 5-disk RAID10 instead of the superior 6-disk |
| 32 |
RAID10 and end up being less lucky in practice. |
| 33 |
|
| 34 |
don't rely on lucks, but why deny good luck to |
| 35 |
come to you when it might? --- two different |
| 36 |
things. |
| 37 |
|
| 38 |
|
| 39 |
> > i see why gambling is not worth it here, but at |
| 40 |
> > the same time, i see no reason to ignore reality |
| 41 |
> > (that a 2 disk failure can be saved). |
| 42 |
> |
| 43 |
> Don't ignore that some 2-disk failures CAN'T be saved ... |
| 44 |
|
| 45 |
yeah, i'm not. i'm just not ignoring that 2-disk |
| 46 |
failure might get saved. |
| 47 |
|
| 48 |
you know... it's better to have a lil window where |
| 49 |
some good luck may chime in than banning good |
| 50 |
luck. |
| 51 |
|
| 52 |
|
| 53 |
> Don't forget, if you have a spare disk, the repair window is the length |
| 54 |
> of time it takes to fail-over ... |
| 55 |
|
| 56 |
yup. just trying to not rely on good luck that a |
| 57 |
spare is available. e.g. considering for the case |
| 58 |
that no space is there. |
| 59 |
|
| 60 |
> > this site [2] says that 76% of seagate disks fail |
| 61 |
> > per year (:D). and since disks fail independent |
| 62 |
> > of each other mostly, then, the probabilty of |
| 63 |
> > having 2 disks fail in a year is: |
| 64 |
> |
| 65 |
> 76% seems incredibly high. And no, disks do not fail independently of |
| 66 |
> each other. If you buy a bunch of identical disks, at the same time, and |
| 67 |
> stick them all in the same raid array, the chances of them all wearing |
| 68 |
> out at the same time are rather higher than random chance would suggest. |
| 69 |
|
| 70 |
i know. i had this as a note, but then removed |
| 71 |
it. anyway, some nitpics: |
| 72 |
|
| 73 |
1. dependence != correlation. you mean |
| 74 |
correlation, not dependence. disk failure is |
| 75 |
correlated if they are baught together, but |
| 76 |
other disks don't cause the failure (unless |
| 77 |
from things like heat from other disks, or |
| 78 |
repair stress because of other disk failing). |
| 79 |
|
| 80 |
2. i followed the extreme case where a person got |
| 81 |
his disks purchased at a random time, so that |
| 82 |
he was maximally lucky in that his disks didn't |
| 83 |
synchronize. why? |
| 84 |
|
| 85 |
(i) offers a better pessimistic result. |
| 86 |
now we know that this probability is actually |
| 87 |
lower than reality, which means that we know |
| 88 |
that the 3.5k bucks is actually even lower. |
| 89 |
this should scare us more (hence us relying on |
| 90 |
less luck). |
| 91 |
|
| 92 |
(ii) makes calculation easier. |
Archives