Re: [tlug] SSD filesystem

[Edward Middleton <emiddleton@example.com>]

Francois Cartegnie wrote:
[snip]
> There's an opinion following linus recent remark about SSDs:
> http://ldn.linuxfoundation.org/blog-entry/should-filesystems-be-optimized-ssd%E2%80%99s
>
> I see more and more MLC devices appearing on the market with, or with more, 
> RAM memory to improve performances (256..512MB...). That way you keep your 
> frequent writes in ram (fs logs) and increase the reliability of those cheap 
> dumb ssd devices for a low additional cost.
>   
[snip]
> And there's still a need for geeks (?) doing SSD disks with regular flash 
> devices (compact flash, SDHC: you can even find multiple raid 0/1 controller 
> for those ones). Most current embedded devices uses a read only FS for wear 
> levelling reasons.
>
> The kernel might need to end up with uniformization for SSD access, and SSD 
> makers firmware to provide a way to tune parameters, the way you can do with 
> scsi disks (or in a limited way with sata).
>   

When you talk about SSD, usb flash, CF, etc. you aren't talking about
direct flash access.  There is no guarantee that any operations on the
device will actually translate into any specific operation on the
flash.  Trying to second guess the controller on an immature device like
SSD doesn't make sense.

If you are directly accessing the flash then it makes sense to consider
wear-leveling etc.  You can do this though the MTD[1] interface using a
filesystem like jffs2[2] now.

Edward

1. http://www.linux-mtd.infradead.org/
2. http://en.wikipedia.org/wiki/JFFS2