ssd mode on rotational media

ssd mode on rotational media

[Kyle Gates]

What issues would arise if ssd mode is activated because of a block layer setting the rotational flag to zero? This happens for me running btrfs on bcache. Would it be beneficial to pass the no_ssd flag?
Thanks,
Kyle
 		 	   		  

Re: ssd mode on rotational media

[Austin S Hemmelgarn]

[-- Attachment #1: Type: text/plain, Size: 928 bytes --]

On 2015-01-07 13:55, Kyle Gates wrote:
> What issues would arise if ssd mode is activated because of a block layer setting the rotational flag to zero? This happens for me running btrfs on bcache. Would it be beneficial to pass the no_ssd flag?
> Thanks,
> Kyle
In theory, it would result in a greater degree of high-level 
fragmentation.  You should (in theory) get marginally better performance 
by using no_ssd (or using a udev rule to set rotational to 1, as the 
flag is writable).

Ideally, bcache shouldn't set the rotational flag to 0 unless the 
backing store (the big device that the cache is reading from) also has 
it set.

Also, ideally, we should be looking at the type of block device, not 
just whether the rotational flag is 0 or not, as there are a number of 
cases where rotational is 0 that enabling ssd will make things preform 
worse (Xen/Virtio block devices come immediately to mind).


[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 2455 bytes --]