| 1 |
On Monday, May 4, 2020 2:50 AM, hitachi303 <gentoo-user@××××××××××××××××.de> wrote: |
| 2 |
|
| 3 |
> Am 03.05.2020 um 23:46 schrieb Caveman Al Toraboran: |
| 4 |
> |
| 5 |
> > so, in summary: |
| 6 |
> > /------------------------------------------------\ |
| 7 |
> > | a 5-disk RAID10 is better than a 6-disk RAID10 | |
| 8 |
> > | ONLY IF your data is WORTH LESS than 3,524.3 | |
| 9 |
> > | bucks. | |
| 10 |
> > \------------------------------------------------/ |
| 11 |
> > any thoughts? i'm a newbie. i wonder how |
| 12 |
> > industry people think? |
| 13 |
> |
| 14 |
> Don't forget that having more drives increases the odds of a failing |
| 15 |
> drive. If you have infinite drives at any given moment infinite drives |
| 16 |
> will fail. Anyway I wouldn't know how to calculate this. |
| 17 |
|
| 18 |
by drive, you mean a spinning hard disk? |
| 19 |
|
| 20 |
i'm not sure how "infinite" helps here even |
| 21 |
theoretically. e.g. say that every year, 76% of |
| 22 |
disks fail. in the limit as the number of disks |
| 23 |
approaches infinity, then 76% of infinity is |
| 24 |
infinity. but, how is this useful? |
| 25 |
|
| 26 |
> Most people are limited by money and space. Even if this isn't your |
| 27 |
> problem you will always need an additional backup strategy. The hole |
| 28 |
> system can fail. |
| 29 |
> I run a system with 8 drives where two can fail and they can be hot |
| 30 |
> swoped. This is a closed source SAS which I really like except the part |
| 31 |
> being closed source. I don't even know what kind of raid is used. |
| 32 |
> |
| 33 |
> The only person I know who is running a really huge raid ( I guess 2000+ |
| 34 |
> drives) is comfortable with some spare drives. His raid did fail an can |
| 35 |
> fail. Data will be lost. Everything important has to be stored at a |
| 36 |
> secondary location. But they are using the raid to store data for some |
| 37 |
> days or weeks when a server is calculating stuff. If the raid fails they |
| 38 |
> have to restart the program for the calculation. |
| 39 |
|
| 40 |
thanks a lot. highly appreciate these tips about |
| 41 |
how others run their storage. |
| 42 |
|
| 43 |
however, i am not sure what is the takeaway from |
| 44 |
this. e.g. your closed-source NAS vs. a large |
| 45 |
RAID. they don't seem to be mutually exclusive to |
| 46 |
me (both might be on RAID). |
| 47 |
|
| 48 |
to me, a NAS is just a computer with RAID. no? |
| 49 |
|
| 50 |
|
| 51 |
> Facebook used to store data which is sometimes accessed on raids. Since |
| 52 |
> they use energy they stored data which is nearly never accessed on blue |
| 53 |
> ray disks. I don't know if they still do. Reading is very slow if a |
| 54 |
> mechanical arm first needs to fetch a specific blue ray out of hundreds |
| 55 |
> and put in a disk reader but it is very energy efficient. |
| 56 |
|
| 57 |
interesting. |
Archives