Re: Documentation about allowing >2 disk failures?

[Roberto Fastec <roberto.fastec@xxxxxxxxx>]

Hello Tim

LVM is a sort of layer and it can reside on any substrate

So go for raid 60 and avoid using too many drives for each RAID 6

Ideally I would use maximum  6x drives for each RAID 6 to next be stack on RAID 10 with the others

bear in  mind that there is a calculation about guaranteed failure of given RAID rebuild process when you exceed a certain amount of terabytes capacity to rebuild

That is why enterprise storages with lot of terabytes are always subdivided in the needed amount of RAID 5 or RAID 6 blocks to be next stack on RAID 50 or RAID 60

At recuperodatiraidfastec.it we have seen so many cases of failed rebuilds due to leak of calculation before implementation

Kind regards

Roberto Gini

CTO @ https://www

RecuperoDatiRAIDFAsTec.it

Ottieni BlueMail per Android

Il giorno 10 lug 2023, alle ore 08:47, Tim 'mithro' Ansell <me@xxxxxxx> ha scritto:

Hello!

RAID5 survives 1 disk failure and RAID6 survives 2 disk failures. What about going beyond 2 disk failures? Does anyone know where I can find information on what good options might be?

I was looking at RAID10, but every extra mirror added only increases failure tolerance by 1 disk. I am also not seeing any inbuilt support in LVM for RAID50 or RAID60 but it seems like if I manually build the sub-volumes, things should still be fine?

My system has 25 disks and I'd like to tolerate up to ~1/3rd of the disks failing so I was thinking that a RAID1 with 3 mirrors built out of RAID6 with 8 disks each (and 1 hot spare) would be a potentially good option?

Thank you for your help!

Tim 'mithro' Ansell

linux-lvm mailing list
linux-lvm@xxxxxxxxxx
https://listman.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

_______________________________________________
linux-lvm mailing list
linux-lvm@xxxxxxxxxx
https://listman.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/