Indeed, I was having a think about it myself on the train back home last night 
and i concluded that any useful caching would already be done on the LDom 
itself.

Maybe we should point out this nice side-effect when using mkfile for creating 
the virtual disk files...

Anyhow, as I store my virtial disk files on separate ZFS filesystems (to allow 
snapshotting), I'm now wondering if there are any quick and easy optimisations 
available there (bigger block sizes etc).  I already set noatime on creation as 
it seemed a fairly obvious win.

-Steve
---

Probably a bad idea, as the images files would normally be larger than the 
amount of RAM allocated to the control/service domain.

*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
Octave J. Orgeron
Solaris Systems Engineer
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

----- Original Message ----
To: ldoms-discuss at opensolaris.org
Sent: Thursday, April 10, 2008 1:39:10 PM
Subject: [ldoms-discuss] should we allow the service domain to cache ldom 
virtual disk files?

When we create ldoms, we use mkfile to create the files for ldom virtual disks. 
 The mkfile command automatically sets the sticky bit which turns off file 
caching.  Would we experience better performance if we removed the sticky bit 
and allowed the service domain to cache the file to memory or is it too big to 
usefully cache anyway?

-Steve


_______________________________________________
ldoms-discuss mailing list
ldoms-discuss at opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss




__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Reply via email to