[Wylug-discuss] Help needed with 'failed' Linux software RAID 10

Paul Brook paul at codesourcery.com
Mon Jun 28 11:58:09 UTC 2010


> 1. Buy/build 3 dedicated systems
> 
>    1.1. A Desktop PC with a single 4-6 core CPU
>    1.2. A 'modest' 4-5 TB File server
>    1.3. A near-identical backup mirror of the file server

>...

> I'd appreciate further comments and won't be in the least bit upset if
> someone makes a constructive suggestion alongside any observation to the
> effect that my ideas are 'barking mad'.

My only question is do you really need this level of redundancy on the file 
server?

From your description so far I'm assuming that the archive is for practical 
purposes readonly (exception one-off re-encoding and addition of new 
material), and has a small number of users.  It also sounds like this doesn't 
need to be a highly available - if you loose access to a chunk of it for a day 
while restoring from backups it's not the end of the world.

If you do need a reasonably high level of availablility then stop reading now 
;-)

It seems to me that the simplest solution may be the best here. i.e. a single 
machine with a pile of disks, and one partition per disk. Maybe use LVM to 
carve things up to allow things to be rearranged later.

Then have some sort of offline backup system. When a disk dies, replace it and 
restore that section of the archive from the backup.  Maybe have the system 
volume on a raid array so that a disk failure doesn't bring the whole machine 
down.

I admit I'm not sure what state-of-the-art is for backups nowadays. I'd guess 
your options are optical media (blue-ray disks look comparable to hdd $/Gb), 
tape drives, cloud storage (too slow for large backups in a reasonable time), 
or an offline disk array (aka. a pile of esata/usb enclosures).

Paul




More information about the Wylug-discuss mailing list