Asynchronous writing to zvols (ZFS)

[Peter Schuller]

[zvol write performance less than file-on-zfs]
> Not sure why's that, I spent no time on optimizing ZVOL yet, sorry.

Absolutely, I was not complaining! I just felt it was worth mentioning. It was 
not meant to be negative criticism.

> With the patch above we synchoronize in-memory transactions every 5
> seconds or when queue is full or when we receive BIO_FLUSH. 

That was my understanding. Sorry, I probably wasn't being clear. I thought 
your original comment ("The problem is that we don't between async and sync 
I/O request on GEOM level") implied that there was some reason one could not 
trust bio_cmd to be preserved correctly (somehow being downgraded from 
BIO_FLUSH before it reaches the zvol class).  Because if this was the case it 
would make perfect sense under the circumstances to treat all writes as 
flushes, in order to achieve correct semantics for *actual* flushes.

So my question was meant to confirm that this was not the case, because if it 
were, the patch would mean that *actual* flushes would not get treated as 
such.

Thanks,

-- 
/ Peter Schuller

PGP userID: 0xE9758B7D or 'Peter Schuller <peter.schuller at infidyne.com>'
Key retrieval: Send an E-Mail to getpgpkey at scode.org
E-Mail: peter.schuller at infidyne.com Web: http://www.scode.org

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 195 bytes
Desc: This is a digitally signed message part.
Url : http://lists.freebsd.org/pipermail/freebsd-fs/attachments/20080808/3f0d4148/attachment.pgp