On Fri, Feb 11, 2005 at 01:22:58PM -0500, Matt Grab wrote:
> On Friday 11 February 2005 08:05 am, Tomasz Wegrzanowski wrote:
> > I don't remember which one is the default, but it's certainly not the only
> > choice in the installer, and it's a very bad choice.
> 
> It is actually a recommended practice on the guest for 2 performance 
> scenarios.
> 1, if the host uses ext3, then the guest and the host both would be doing 
> journalling.  Performance suffers tremendously here.
> 2, if the host uses RAID, (or RAID and ext3), then again, doing double 
> journalling hinders performance.

You are not quite correct, in fact it's all the other way:
* ext3 does metadata-only journalling, so the host's journalling doesn't get
  involved in guest's filesystem operations (which are just reads and writes
  of already allocated parts of the file, from host's point of view) *in any 
way*.
* UML can and does crash independently from the host. That's kinda the point of 
it.
* Journalling (of metadata) is very cheap, lost data and wasted administration 
time
  is much more expensive. Especially since UML is not used for performance.
* I fail to see in what way RAID is supposed to help here - the data would be 
lost
  after UML crash because it was still in memory. No matter how smart is the 
host's
  disk and filesystem technology, it can't guess what guest was going to write 
before
  it crashed.

Journalless filesystems should just die.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
User-mode-linux-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user

Reply via email to