Para-virtualization - the next level ?

[Technologov]

Para-virtualization can improve performance. This is a known thing.

See a theoretical paper:
http://forgeftp.novell.com/lfl/.html/vi ... ation.html

Since VirtualBox came so very far, now supporting VirtIO network with v3.1, does it makes sense it go a step deeper?

See VirtualBox 3.1 BETA - VirtIO:
http://forums.virtualbox.org/viewtopic.php?f=15&t=24477

Maybe look at VMware VMI (Virtual Machine Interface) specification ?
BTW: 32bit Linux kernel support VMI since kernel 2.6.21 (?).

Maybe look at other directions ? What is next ? Integrating Containers ?

-Technologov

[vbox4me2]

How about spending the next 3 months on chasing and solving bugs and nothing else

[Technologov]

This idea is long-term one, not short-term one.

I quite agree that we need to resolve bugs, as I did answer about 100 (!) bugs yesterday, and put my recommendations. Not only we have 2000+ bugs outstanding, but many of them are unanswered for over 6 months !

According to my recommendations Sander already closed 30.

-Technologov

[vbox4me2]

Always good to kick up the attention span but in any healthy dev.team there is nothing wrong with dedicating a bug-window once a year to clear up the bugbase and go thru the code, since vbox doesn't have a post build patch regime you have no choice if you want a stable product. Stability goes alot further then new features.

[unggnu]

Support for the Virtio Storage Devices would be great. This seems to be the main reason for me that KVM is much faster but maybe it is just a bug in Virtualbox - for example Vista is so much faster (starting, using ...).