> Tunnels are used for more than just this. And, it's not the tool that it

It might help to enumerate these, or, perhaps, to define which of the uses
in Pekka's original note correspond to virtualization and which to some
other use.

> flawed per se, Pekka is suggesting that perhaps there is something else
> that is flawed hence we need to apply the tool of Tunneling. We should
> look for that "something else."
>

Well, to explore Joe's analogy with virtual memory a bit deeper, virtual
memory is needed because the architecture of von Neuman machines consists of
a central processor and a memory store organized as a linear address space
with fixed addresses. This architecture only allows a single process to run
at one time. Virtual memory allows multiple process images to be mapped
sequentially and in parallel into the fixed linear address space, and thus
to have multiple tasks running.

One could argue that the nature of the basic IP network architecture, fixed
addresses located within a static subnet topology, only supports static
hosts and networks at fixed locations. Virtualization via tunnels allows
mobility of networks and hosts, among other things.

Now, I suppose one could argue that something is broken in the basic IP
nework architecture, but, from an engineering standpoint, I'd observe that
changing that might be as hard as changing the von Neuman machine
architecture. Possible, but likely to take a long time.

BTW, in my previous email, I mentioned only address translation as an
alternative tool for virtualization. I think one could argue (and perhaps
Pekka would agree?) that HIP is also such a tool. As it currently exists,
HITs are identified to hosts, but one could expand the concept to better
support address space virtualization so that they identifiy larger
topological aggregates. But, at that point, one has to ask what is the
difference between a 128 bit HIT and a 128 bit IPv6 address.

                      jak



_______________________________________________
Int-area mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/int-area

Reply via email to