Date: Fri, 19 Oct 2007 10:21:53 -0600 From: James Gritton <jamie@gritton.org> To: freebsd-arch@freebsd.org Cc: Julian Elischer <julian@elischer.org> Subject: Re: jail/vimage level virtualisation requirements. Message-ID: <4718D9A1.8040200@gritton.org> In-Reply-To: <4717F983.5060707@elischer.org> References: <470E5BFB.4050903@elischer.org> <ff5vdh$jol$1@ger.gmane.org> <200710172216.l9HMGhbd067251@apollo.backplane.com> <200710181359.46720.qpadla@gmail.com> <4717F983.5060707@elischer.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Julian Elischer wrote:
> I don't want to discuss virtualisation that duplicates the entire kernel,
> other than the single question "Should we drop support for jails and
> vimage
> style virtualisation in favour of "Userland linux/dragonfly/freeBSD"
> or Xen or {your favourite virtualmachine}.
That's an easy one - a big "no". Broader virtualization technologies
are fine,
but they serve a different need and shouldn't take over what jails
(especially
with certain extensions) offer. Namely: ...
> IF we decide to keep teh jail/super-chroot/vimage support, then
> what do we want to see out of it?
Light weight. The main thing I want to see is being able to partition
the jail
into a separate virtual environment without bloating the kernel, or
adding new
kernels. To be able to offer a complete FreeBSD userspace environment
largely
indistinguishable from the unjailed, with the benefit of keeping different
jails from interfering with other as much as possible. I'm not interested
(for the purposes of this topic) in acting like another OS, or acting like
you're on some kind of different hardware.
Basically, I'm looking for everything Matt Dillon said we shouldn't bother
to do at this level.
- Jamie
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4718D9A1.8040200>