Outside of hackers, no one really replaces their kernels. 1. Home users use whatever their distro have (Ubuntu, openSUSE, Fedora, ...) 2. Enterprise users stick to their Red Hat Enterprise Linux (or CentOS or OEL) and use it's kernels only.
But the question is: How hard would it be to achieve a stable API/ABI for host-side vboxdrv, like KVM does ? (KVM kernelspace can be matched with any recent Qemu user-space, so it is compatible) IMO Today vboxdrv does too much guest-host communication (such as Vbox execute on the host), which changes from one version to the next. But if it did only virtualization (VMX/SVM) it would be stable. Is it worth the effort ? (I dunno, but if it won't add new community developers, probably not.) -Technologov On Wed, May 8, 2013 at 3:08 PM, Stephan von Krawczynski <[email protected]> wrote: > On Wed, 08 May 2013 15:40:04 +0200 > Klaus Espenlaub <[email protected]> wrote: > >> On 08.05.2013 06:09, quickbooks office wrote: >> > Please Merge VirtualBox Kernel Modules into the Kernel. >> >> Sorry, this will not happen in the foreseeable future. >> [...] > > Would it be valid to interpret your email that vbox is likely more > concentrating on virtualisation of Desktop-driven guests and less likely on > server-type guests? > Because talking of users not updating kernels can only mean classical > distribution users that simply don't replace with stock kernels. > On the other hand we experience very bad network performance on multi-cpu > guests mostly being servers, whereas sinle cpu guests are a lot better (a lot > means at least factor 5 in bandwidth). > > -- > Regards, > Stephan > > _______________________________________________ > vbox-dev mailing list > [email protected] > https://www.virtualbox.org/mailman/listinfo/vbox-dev -- -Alexey Eromenko "Technologov" _______________________________________________ vbox-dev mailing list [email protected] https://www.virtualbox.org/mailman/listinfo/vbox-dev
