[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: compression and encryption
On Wednesday, November 13, 1996 11:42 AM, Phil Karn[SMTP:karn@qualcomm.com] wrote:
<snip>
@
@ It may well be that it will never be possible to build one integrated
@ network that can handle both realtime and nonrealtime traffic with an
@ acceptable degree of efficiency and delay performance for both.
@
Phil,
In my opinion, I think that depends on how you define "integrated".
If everyone assumes that the flat addressing and equal peer model
of the Internet is the only way to scale the system, then I would
agree with you.
Imagine if everyone on the "toll" telephone network was allowed
to plug everything from a phone, to a PBX, to a 5ESS into the
network, this model would have problems scaling and being
properly "integrated".
In my opinion, the existing IPv4 network needs to be migrated
to become a hardened, centralized, highly-reliable, "core" that
primarily provides two things:
1. Low-cost bit transport to many regions of the world.
2. A distributed DNS system for mapping symbolic names
to binary quantities (a.k.a. DNS).
Once this "core" is stable it can be made more secure by
encouraging the migration of users to better integrated layers
which can designed around the edges of the core. The layers
can be engineered based on "application" requirements as
opposed to transport requirements.
After it becomes more clear what the application requirements
are (based on customer demands), the core can be evolved
to more efficiently handle the outer layers. This type of integration
assumes a separation of the transport core from the application
layers. In my opinion, this can be viewed as "one" network,
although I can see how some people might view it as more than
one.
--
Jim Fleming
UNETY Systems, Inc.
Naperville, IL
e-mail:
JimFleming@unety.net
JimFleming@unety.net.s0.g0 (EDNS/IPv8)
Follow-Ups: