►
From YouTube: Move beyond the ossified Forwarding Plane
Description
HotRFC for IETF108
A
Welcome
my
name
is
thomas
eckhart
and
I
would
like
to
talk
how
to
move
beyond
the
ossified
forwarding
plane.
50
years
ago
we
did
ipv4
out
of
which
the
internet
evolved.
25
years
later,
the
ipng
next
generation
ip
process
resulted
in
the
creation
of
ipv6,
which
gave
us
a
bug,
fixed
version
that
replaced
short
addresses
with
long
addresses
and
in
the
25
years,
following
with
then
we've
seen
an
ever
increasing
ossification
of
the
network
layer.
A
Evolution
for
many
reasons
think
that
now
really
is
a
good
opportunity
to
catch
on
the
ongoing
revolution
in
forwarding
plain
hardware,
we're
seeing
a
lot
of
softwareization
data,
centers
virtualization
of
hardware,
user,
programmable
forwarding,
plane
engines
even
programmable
traffic
managers.
Custom
built
forwarding
asics
become
as
easy
to
build
as
ever,
and
the
performance
evolution
also
has
given
multiple
orders
of
magnitudes
over
what
we
had
25
years
ago.
So
would
really
be
good
to
have
an
ability
to
also
look
into
research
and
architecture
for
the
next
revision
of
our
network
layer.
A
The
the
problem
really
is
a
little
bit
that
we
have
no
good
place
to
go
amongst
all
the
different
working
groups.
We
have.
We've
got
working
groups
primarily
for
encoding
and
core
features
like
six-man,
spring
mpls
and
so
on,
and
for
each
of
the
individual
features
and
services
within
the
encodings.
We
have
specific
working
groups
and
research
groups,
there
are
additional
working
groups
and
maybe
even
research
groups
to
do
the
securing.
A
A
So
what
we're
really
looking
for
is
to
find
solutions
on
that
organizationally
and
this
slide
deck
is
suggesting
that
we
could
have
a
forwarding
cane
discussion
and
dispatch
forum
and
that
we
might
want
to
have
a
future
ip
evolution,
research
group,
so
the
forwarding
plane,
discuss
and
dispatch.
We
had
a
session
on
forwarding
plane,
realities
in
2019.
A
It
covers
you
know,
from
the
transport
layer
down
to
the
subnets
network
layer,
layer,
2.5,
all
those
aspects
that
relate
on
how
to
bring
our
protocols
into
actual
forwarding
hardware,
including
cpu
nfv,
a6,
low-end
iot,
with
all
the
good.
You
know,
add-on
problems
like
user
probability,
virtualization
slicing.
A
What's
the
forum
we
could
build,
I
think
it
would
be
great
if
we
could
start
with
a
mailing
list
to
get
organized,
and
then
I
think,
a
forum
like
that,
with
expertise
on
these
forwarding
plane
issues
could
help
becoming
a
directorate.
We
could
see
if
iab
would
want
to
pick
up
aspects
of
this
or
the
operations
area
and
there's
certainly
a
lot
of
good
things
that
can
be
done
with
such
a
forum
with
respect
to
the
actual
protocol
evolution
itself.
A
I
think
that,
obviously
efficient
and
high
speed
packet,
forwarding
and
encoding
mechanisms
are
required
for
all
functional
areas.
The
functional
areas
that
we
feel
have
the
most
need
for
improvements
are
the
addressing
semantic
to
become
really
more
varied
than
we
have
it
with
just
a
single
128-bit
encoding
in
ipv6,
but
more
flexible,
also
future
proof
to
avoid
future
experiences
like
the
25
net
versions,
we
had
between
ipv4
and
v6.
A
The
second
main
bullet
point
is
to
really
evolve
beyond
the
8-bit
qs
that
we
currently
have
in
ip
for
more
high
precision,
latency
throughput
loss
and
congestion
management
and
then
last
but
not
least,
really
looking
at
the
network
layer,
forwarding,
plane,
interaction
with
security
like
application
to
network
trust,
the
conflicts
between
privacy,
protection
and
the
network,
service
and
accountability,
and
so
on.
A
So
what's
what's
our
plan
at
this
itf
friday,
1500
utc,
we're
inviting
you
for
a
webex
site
meeting
to
discuss
these
issues.
Please
come
and
feel
free
to
contact
us
in
before.
We've
also
listed
the
references
here,
some
of
the
documents
that
we've
written
in
preparation
of
getting
started
with
that
work.
Thank
you
very.