On 3/9/2014 2:33 PM, Alia Atlas wrote:
In the last few years, there seems to be a drive towards overlays and
additional
packet encapsulations.
If by "few" you mean over 15, agreed: http://www.isi.edu/xbone
E.g., it was nearly 10 years ago that I gave a tutorial at Infocom on
virtual and overlays.
> What problems do you see these as solving?
These have been addressed in a multitude of IETF WGs and RGs over the
past years - currently L2VPN, L3VPN, NVO3, TRILL, LISP, and the past
VNRG RG.
The goals are already understood as virtualizing resources (routers,
subnets, links, etc., for the same benefit an OS adds to computation:
- abstraction
- resource sharing
- shared resource isolation
- software/service reuse
Is
there a
more focused way to consider the drivers and downsides?
Although network virtualization has been around for a long time, even
actively being developed in the IETF, stepping back to look at the long
view is starting to gain more traction. I agree that the wheel is being
reinvented, esp. with a lot of recent "X over UDP" proposals.
FWIW, INTAREA has a tunnels document that Mark Townsley and I started a
few years ago on which work is resuming. A revision will be out soon.
That document focuses on IP over encapsulation X, esp where X is also
(or ultimately) IP, but a lot of the principles may be of more general
applicability.
Joe