►
From YouTube: CNCF Telecom User Group Meeting - 2019-10-07
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
A
B
B
I
guess.
I
would
emphasize
our
interest
in
having
telecom
operators
as
co-authors
of
the
talk.
We're
also
very
interested
in
feedback
from
vendors
about
the
content
and
to
essentially
ensure
that
we
don't
get
anything
wrong.
But
we
want
to
think
probably
start
with
just
some
high
levels
about
what
CNS
are
and
how
people
are
going
about
them
and
then
in
areas
where
there's
less
agreement
to
try
and
list
the
different
ways
of
doing
things,
as
opposed
to
being
particularly
prescriptive,
especially
up
front
and
so
we're
very
we
would
love
to.
B
In
fact,
we
are
planning
to
highlight
a
few
telecom
operators
and
the
approaches
that
they're
taking
and
would
like
those
folks
if
they
can
get
permission
to
be
co-authors
in
the
paper.
We're
then
also
very
interested
from
hearing
from
vendors
and
other
members
of
the
community
of
oh.
You
should
also
consider
this
alternative
or
this
option,
or
you
know,
keep
track
of
some
of
the
positives
and
negatives
of
different
approaches.
So
alex
is
going
to
be
reaching
out
to
you
but
I,
wouldn't,
if
particularly
the
operators
on
the
call
to
please
pay
me.
B
Our
hope
is
that
it's
going
to
be
Rakuten,
verizon
and
Vodafone,
but
if
we're
also
chatting
with
charter
and
Bell
Canada
a
couple
folks
other
folks
in
there.
So
with
that
I
think
I
would
hand
it
off
to
Mussina
and
Taylor
to
give
a
little
bit
of
an
update
and
maybe
present
this
slides
about
exactly
how
the
CNF
test
bed
has
changed
over
the
last
three
months
and
then
what
the
the
plans
are
for
the
next
three
months.
B
C
Good
news,
Thanks
I,
can
I
can
say
somewhere
OSI
entity
and
Matt
yeah.
Please
go
ahead
yeah.
So
what
was
happened
before
last
week
is
the
second
meeting
of
the
see
entity
which
was
faced
face
and
public
to
both
their
Cove
indoors
and
and
operators,
and
we
agreed
on
to
start
working
on
the
on
the
kubernetes
based
reference
architecture,
which
means
that
we
will
describe
an
infrastructure
which
is
based
on
kubernetes
and
which
can
support
one
or
more
profiles
from
the
reference
model
previously
defined
in
in
any
entity.
C
The
initial
content
has
been
already
pushed
as
a
poor
request.
I
think
last
week
and
now
heavily
discussed
the
there
are
like
different
views
on
how
much
should
we
consider
kubernetes
as
a
full-blown
and
FEI
slash
infrastructure
service
and
which
parts
of
kubernetes
covers
from
the
heavy
mineral
oil
texture
to
overcome
these
B's,
like,
let's
see
mostly
philosophical
debates,
we
agreed
on
the
meeting
that
we
will
focus
on
the
interface.
C
C
C
B
B
The
docks
yeah
all
the
docks
that
we
produce
will
be
open
for
public
review,
and
so
we
would
love
to
get
I
mean
I.
It's
a
little
bit
of
a
metaphysical
question
on
whether
the
review
is
from
you
as
a
telecoms
engineer
or
you
as
an
official
representative
of
CNT
tea,
but
we're
very
interested
in
your
review.
C
C
Yes,
thank
you,
I
might
so
my
beliefs
that
I
will
you
know
I
will
send
from
time
to
time.
This
all
white
paper
to
the
synthetic
mailing
lists
and
I
will
send
from
time
to
time
the
content
for
the
CNT
degenerates
to
or
slack
or
mailing
lists.
So
you
know
just
to
initiate
some
kind
of
cross
communication
between
the
two
groups.
Yeah.
D
Now
this
is
a
Watson
like
to
add
ban
in
the
CNT
te
multi-tenancy
was
brought
up
multiple
times.
It
seems
like
that
would
be
a
good
if
we
get
created
user
story
or
some
requirements
on
how
kubernetes
can
address
multi-tenancy
issues
for
CNS
for
the
telcos
it
seems
like
that
would
be
a
good
place
to
try
to
collaborate.
B
Might
just
respond
about
that
by
saying
that
that
probably
is
its
own
whole
chapter
of
the
white
paper
or
whatever
we
wind
up
calling
different
sections,
but
it
definitely
could
both
specify
some
requirements
and
then
say,
what's
feasible
today
with
things
like
our
back
and
then
what
things
are
coming
down.
The
pike
in
our
alpha
level
features
or
we're
not
even
shipping.
Yet.
D
F
Think
I
think
in
general,
it's
gdpr.
It's
any
sort
of
regulation,
also,
operators
typically
classify
their
applications
into
security
groups,
and
and
they
would
like
to
have
separation
between
those.
So
there
are
vnfs
which
sort
of
live
in
a
so-called
demilitarized
zone,
which
of
course
means
much
higher
security
than
others.
F
E
F
A
Excuse
me:
I
got
a
cost
at
cube
con
with
the
two
to
our
blocks
that
were
trying
to
get
the
rooms,
for
it's
appreciate
your
help
on
that
down
and
there's
a
link
to
the
reference
model,
thanks
for
adding
that
I.
That's
here
in
the
notes
for
anyone
wants
to
look
at
the
the
current
discussion
on
the
same
DD
si
I.
H
A
A
H
G
A
A
A
lot
of
the
use
cases
that
we're
in
there
that
were
primarily
based
on
ansible
and
other
custom
scripts
have
been
re-implemented
with
either
helm
or
EML
that
you
can
use
with
you,
keep
it
apply
and,
along
with
that,
we've
added
several
and
network
service
mesh
use
cases,
including
one
that
uses
a
physical,
a
physical
gateway,
a
physical
note
and
network
card
gateway.
So
we
actually
have
the
capability
to
acceleration
and
the
in
that
network
function
and
connect
other
service
chains.
We.
A
Use
that
during
ons
as
part
of
that
tutorial,
which
there's
links
on
the
Telkom
user
group,
a
document
as
well
as
they
this
road
map.
Now
all
of
these
are
clickable
links
to
go
through,
and
so
a
lot
of
it
has
been
getting
those
building
blocks
in
place
with
the
new
structure
forever,
to
make
it
easier
to
add
new
ones
and
get
contributions
and
and
then
testing
that
out.
A
With
this,
is
our
roadmap
going
forward
and
there's
probably
some
stuff-
that's
going
to
be
shifting,
but
this
is
covering
quite
a
bit
of
it.
So
some
of
these
are
built
on
some
existing.
We
do
have
an
IPSec
use
case,
but
what
we
don't
have
is
a
multi
node
setup
with
network
service
mesh
you're.
Looking
at
putting
those
in
place,
we've
had
a
existing
benchmarking
test
case
that
uses
IP
route,
simple
IP
routers,
either
ipv4
our
v6
and.
A
H
A
Able
to
benchmark
that
what
I
don't
have
on
here,
but
listening
to
a
lot
of
feedback,
was
understanding
the
results
when
we
are
doing
benchmark
examples.
Some
of
these
are
functional
tests
and
you're.
Looking
into
n
test
results,
some
of
them
are
benchmarking,
but
the
end
feedback
was
what
are
the
results
were
looking
at
and
what
do
we
expect
whenever
we
run
the
test,
so
we'll
be
updating
the
documentation
and
kind
of
structure
to
make
sure
when
you
go
in
to
a
specific
use
case,
and
you.
H
A
A
In
addition,
that
we're
working
to
add
Dan,
M
and
some
other
different
technology
like
Malta's
and
other
pieces
that
we'll
be
adding
in
there
and
hopefully
at
a
GSM
5g
use
case
like
using
the
packet
facilities
that
are
connected
to
Sprint's
Network
and
towards
the
end
of
the
year.
We
have
a
call
for
help
on
the
open
sex
side.
We
do
support
the
pulling
open
stack
right
now.
A
H
Can
you
please
switch
back
to
the
roadmap
that
would
help
me
Karen.
Thank
you,
so,
yes,
the
last
month
and
a
half
so
something
we
have
been
adding
this
very
initial
use
cases
with
Tennyson.
We
started
with
a
very,
very
simple
one,
and
then
we
expanded
to
external
connectivity.
As
a
result
of
this,
we
had
some
like
feedback
that
is
very
small
for
our
community.
H
H
So
we
have
one
basic,
probably
more,
examples
that
are
in
the
queue
as
a
shows,
maybe
for
the
next
months,
and
at
least
from
my
point
of
view,
this
could
be
shared,
actually
opponet
working
on
it,
the
tutorial
and
during
the
session
that
we
did
about
any
segment
or
rather
link.
You
see
that
this
bed
we
had
actually
the
most
complex
NSF
deployment
that
I
have
personally
seen,
like
probably
I.
Don't
know
20
or
15
20
number
of
end
points
connected
to
changes
have
always
again
thanks
to
Michael.
Thank
you.
H
Michael
for
putting
countries
together
and
I
have
reached
out
to
a
number
of
people
already
I'm
really
inclined
to
see
this
GSM
gateway
5g,
and
we
have
to
look
at
what
what
is
available.
There
is
an
open
source
software
that
can
be
used
a
tool
to
implement
this,
so
I
would
say
that,
at
least
for
for
me,
like
personally,
and
also
from
the
same
point
of
view,
I
think
that
not
a
problem.
This
is
the
next
big
thing
that
I
will
be
looking
forward.
A
I
I
think,
if
anything,
you
can,
you
can
share
the
DC
and,
if
test
bits,
github,
potentially
the
pull
requests
that
we
have
in
there
right
now,
but
just
to
try
and
start
out
where
I
Nicolai
left
off.
We
we
did
the
the
ons
demo,
where
we're
showing
this
external
gateway
and
as
it'll
also
be
B,
be
visible
from
the
documentation
we
added
to
that
one.
I
It
would
be
nice
to
have
something
where
we
have
a
bit
more
control
over,
what's
visible
inside
the
containers
and
and
and
similarly
I've
been
been
working
on,
adding
this
SR
IV
Network
device
plug-in
as
an
example
to
the
testbed
as
well,
and
this
one
is
kind
of
similar
in
terms
of
external
interfaces
in
VPP
running
inside
a
container.
I
So
so
what
this
plugin
allows
us
to
do
is
it
allows
us
to
assign
a
limited
or
a
subset
of
DT
physical
resources
or
interfaces
available
to
a
container,
but
since
we're
running
the
container
and
privileged
mode
and
from
inside
the
container,
you
can
still
see
all
the
interfaces
and
you
still
have
the
option
of
using
them.
So
for
now
we're
using
a
somewhat
of
a
workaround
where
we
use
the
plug-in
to
assign
resources
which
are
or
lists
are
listed
inside
the
container
and
the
environment
variables.
I
And
then
we
generate
a
configuration
that
matches
the
key
requests
so
from
a
user
perspective,
you'll
still
see
that
you
have
the
resources
assigned
and
those
resources
will
be
available
inside
the
container.
But
if
you
get
a
CNF
in
this
case
status,
that
is
a
somewhat
on
that
stone.
It
has
the
ability
to
take
over
additional
interfaces.
A
I
C
A
I'd
probably
say
that's
in
a
general
case,
unless
you
made
exceptions
and
there
are
ways
that
you
could
isolate.
If,
if
we
said
no
Definity
or
something-
and
he
said
these
specific
nodes
are
going
to
not
have
any
other
Pods
running
on
them,
you
could
get
around.
It
agreed
that
that
should
be
the
general
goal
and
then,
if,
if
you
were
looking
at
exceptions
on
that
rule,
how
would
we
do
that.
A
Think
one
of
the
specific
things
I
didn't
mention,
which
Dan
mentioned
in
the
ONS
keynote,
is
we're
moving
towards
having
the
CNF
test
bed
as
a
full
dev
environment.
Anyone
could
use
for
and
working
on
use
cases
or
writing
CNS
and
then
eventually
being
able
to
run
any
type
of
some
test
for
conformance
to
in
collaboration
with
other
groups.
So
that's
kind
of
the
goal
and
if
you
have
any
feedback
on
usability
or
anything
else,
please
let
us
know
I.
A
Absolutely
we're
gonna
be
working
with
and
trying
to
collaborate
on
with
the
VP
certification
and
that's
going
to
be
tied
in
with
Santa
T.
So
there's
going
to
be
a
lot
of
cross
collaboration
between
all
the
groups
and
what
we're
looking
at
is
being
able
to
run
the
test.
I'll,
not
say
right
now.
What
does
that
mean
as
far
as
everyone
else?
But
we
would
like
to
anything
that
we
implement
and
the
CNF
test,
but
we'd
like
to
be
able
to
say
whether
it
passes
or
not
at
the
certification
conformance.
A
K
A
There
will
be
collaboration,
and
it's
definitely
if
our
kids
asked
and
I've
had
some
conversations
about
working
was
the
first
OTP
certification,
and
none
of
nothing
is
moved
forward
on
that.
So
this
is
real
early
kind
of
announcement
that
we're
going
to
be
doing
that,
but
we'll
definitely
be
talking
and
working
with
all
the
groups
to
make
sure
what
we're
doing
is
going
to
be
able
to
support
those
plans
and
that
we
can
run
them
and
implement
them
and
influence
them.
If
we
have
any
feedback.
B
A
A
A
C
F
Practice
practically
we
have
this
month
to
work
with
and
and
I
don't
know
how
do
we
want
to
handle
the
the
review
by
sea
entity
whether
that
can
be
done
before
cube
corner
after
cube
cone?
That's
an
interesting
question,
but
I
I
fully
agree.
I
think
we
have
a
very
ambitious
scope
today
defined
for
the
white
paper
and
and
if
we
could
agree
on
a
on
a
scope
that
we
prioritize
for
the
first
release,
then
we
have
a
better
chance
of
finishing
anything.
L
So
these,
from
my
perspective,
having
observed
the
progress
and
long
progress
of
the
white
paper
for
some
time,
it
seems
that
one
of
the
reasons
that
it's
not
making
progress
is
that
it
implicitly
has
a
scope
creep.
My
understanding
was
that
initially,
the
goal
was
to
define
how
do
I
mean
that
the
gold
standard
for
cloud
native
network
functions
look
like
and
what
is
necessary
from
the
platform,
so,
basically
limiting
the
scope
to
say
what
I'm
talking
about
the
hypothesis
actually
telco
functions
are
not
special
right
standard
queuing.
L
This
platforms
that
are
supporting
sophisticated
work,
loss
from
the
ITM
strings
should
be
sufficient
and
then
discuss
where
the
hypothesis
breaks
down
and
potentially
things
need
to
be
edit.
But
what
I
see
is
that
actually,
the
most
progress
is
being
made
on
adding
a
discussion
of
Etsy
mono,
adding
armed
support
etc,
which
are
not
which
I
don't
CSS.
Being
the
core
of
the
message
here,.
C
L
F
I
tend
to
agree,
I
think
there
is
there's
one
more
thing
that
perhaps
is
missing
before
defining
this
the
good
CNF,
and
that
is
what
is
the
thing
that
we
want
to
achieve
with
this
transformation
or
we've
been
having
C
and
I've
seen
the
terrico
contacts,
because
I
think
I
think
the
benefits
of
cloud
made
their
context
I
think
that's
a
bit
more
I
think
it's
important
to
match
up
the
architecture,
transformation
that
we
look
at
for
the
CNF
sorweful
from
the
vnfs
going
to
the
CNF.
The
device
needs
to
be
understood.
F
Why
is
it
what's
the
benefit
to
the
operator
of
having
CNF
Sun,
not
VN,
apps,
so
I
think
that's
that's
the
reason
why
I
suggest
that,
at
least
in
the
first
white
paper
we
should
try
to
define
the
wanted
position
or
or
the
goals
of
this
transformation
and
I
think
that
is
what
area
where
we
could
use
the
help
of
you
know
operators,
because
the
growers
shouldn't
be
certified.
Landers.
L
L
L
L
Also
like
that
text,
it's
maybe
overshooting
a
little
bit
on
the
Turkish
side,
because
at
least
things
where
the
text
claims
telcos
function
are
special
I,
don't
I,
don't
think
that's
necessarily
true,
but
let's
look
into
that.
I
think
what
we
need
is
first,
a
high
level
description
of.
Let's
start
from
the
hypothesis.
There
are
cloud
native
network
functions
that
we
want
to
operate
and
design
as
closely
as
possible
to
what
the
rest
of
the
IT
industry
does
with
their
special
functions
not
compared
to
web,
but
to
other
high-performance
function.
F
Iii
really
like
what
you
say,
but
I
think
I
think
the
purpose
of
discussion.
Why
I
have
this
this
item
on
the
agenda
is,
is,
of
course
we
could
argue
on
what
we
should
write
in
the
text,
but
I
would
like
to
see
a
proposal
emerging
in
terms
of
a
prioritized
table
of
contents
for
the
white
paper,
so
we
could
get
to
work,
and
you
see
my
proposal
on
on
the
screen.
F
I'm
I'm,
basically
I
took
all
the
parts
which
seemed
like
relevant
and
are
in
good
shape
and
can
be
related
to
the
topic
of
of
the
test
formation
of
the
CNF
themselves.
So
I
took
the
definition
of
the
CNF,
the
similarities
and
differences
between
enterprise
and
telecom,
then
I
sort
of
suggested.
We
need
to
talk
about
the
goals.
F
F
I
think
I,
like
the
text,
but
then
I
don't
feel
like
it
fits
into
this
white
paper
that
is
concentrating
on
the
CNS,
because
I
think
the
the
core
value
and
the
core
notion
of
being
cloud
native
is
to
be
agnostic
to
the
infrastructure
and
and
and
there
I
think
the
most
we
can
say
is
that
the
application
should
work
on
whatever
processor
architecture
where
I
can
run
kubernetes.
Yes,.
K
So
one
comment
because
I
had
chapter
myself
so
I
would
say
saying
F
and
there
would
be
you
know,
agnostic
to
the
architecture,
to
the
infrastructure
architecture,
to
the
processor
architecture.
So
for
the
thing
that
I
think
is
should
be
able
to
run
on
the
ARM
architecture
as
well
without
knowing
what's
unrealistic.
K
L
Totally
agree
on
the
goal
for
the
arm
support.
The
question
is:
do
we
need
to
give
so
much
space
to
something
that
is
at
least
as
obvious
I
mean
should
be
obvious
that
that
other
architectures
are
supported,
and
we
can
mention
that?
Do
we
need
to
create
a
whole
section
for
it?
That's
that's
a
different
question.
I
think.
K
E
B
That
up,
but
I'd
also
just
point
out
this
concept
of
chapters
that
just
putting
in
a
paragraph
and
saying
hey,
we're
gonna
circle
back
to
this
in
the
future.
Taylor
and
Denver
Lucena
are
actually
now
some
of
the
at
running
multi
architecture
containers
through
their
work
on
CNC
FBI,
and
so
they
would
be
happy
when
the
time
is
available
to
come
back
and
talk
about
some
of
the
lessons
learned
there
and
his
goals
and
such.
E
K
K
F
Like
to
go
with
the
suggested
approach
of
of
emphasizing
that
the
Hardware
agnostic
City
and
then
you
know
saying
that
that
there
is
gonna,
be
more
details
on
this,
because
then
I
think,
if,
if
we
do
something
around
hardware,
that
that
should
be
a
renewed
feature,
discovery
or
maybe
use
of
smart
Nix.
And
then
you
know,
different
processor
architectures,
then
can
get
their
own
chapter.
But
perhaps
not
in
the
very
first
white
paper.
K
I
understand
that
the
transformation
from
the
x86
or
to
arm
or
from
arm
to
x86
are
quite
time-consuming
work
in
there's
a
lot
of
stuff
involved,
so
I
think
we
should
mention
it
somehow
to
to
remind
people
to
understand.
You
need
to
take
that
into
consideration
before.
Were
they
really,
you
know,
implement
some
kind
of
application,
so
that
would
maybe
you
know
to
to
save
a
lot
of
time
and
the
money
and
the
human
resource.
K
E
At
the
very
first
I
think,
the
first
five,
the
intro
up
to
the
ideal,
CNF
I
think,
would
be
good.
I.
Think
the
importance
of
Etsy
model
management
support
is
going
to
be
a
very
interesting
topic
and
I
think
there's
a
lot
more
there
that
we're
gonna
have
to
talk
about
them
in
terms
of
like
how
does
it
fit
in
with
with
CNF
sand?
Do
we
do
we
need
to
drive
existing
it's
the
infrastructure
into
into
CNF
sand?
E
My
my
reaction
towards
us,
or
at
this
point
is
we
probably
don't
need
the
majority
of
what
Etsy
has,
but
we
need
to
make
sure
that
we
can
integrate
with
it
and
I
know.
Others
here
also
share
differing
opinions
on
that
as
well,
so
I
think
the
first
everything
above
that
I
think
we
can.
We
can
get
some
clarity
and
like
we
we
have
to.
We
have
seven
I
suspect
we
have
some
path
the
weekend
that
we
can
drive
and
deliver
on
something
I
would.
L
C
L
F
Have
problems
leaving
my
purpose
with
including
that
was
that,
no
matter
what
we
do
to
the
CNS,
we
somehow
need
to
show
continuity
in
the
telco
space
about
management.
I
think
God
would
be
in
a
large
amount
of
money
spent
towards
Etsy
on
a
fee
in
the
cost
and
and
I
think.
If
we
introduced
a
drastic
change
to
the
CNF
itself,
we
need
to
do
that
without
a
drastic
and
very
costly
change
to
the
operational
model
of
the
operator
and
so
start.
That's
why
I
thought
it
makes
sense
to
have
it
here.
M
N
So
I'm
an
operator,
so
I
don't
shoot
me
now,
but
I
totally
agree
with
the
fact
we
should
not
include
it.
For
now
we
have
tried
to
interplay
to
implement
this.
Everybody
note
the
challenge
we
have
right
now
with
those
kind
of
architectures
and
cyc
entities
still
there.
So
the
first
five
bullets
are
the
key
that
are
going
to
justify
the
glue
of
how
we
make
that
cloud
native
transformation.
How
to
make
our
network
workloads
look
more
like
IT,
knowing
that
the
major
differences
might
be
their
service
impact
afterwards.
N
Like
me,
a
telco
application
moving
into
the
cloud
native,
you
still
have
it
critic
critical
application.
There
ders
they're,
supporting
a
critical
function
in
our
network.
That
might
be
more
around
that
angle.
We
should
we
should
angle
after
that,
more
lesser
technology,
but
their
day
to
service
that
needs
to
be
relying
on
this
so
and
Manno
afterwards
will
be,
should
be
coming
afterwards
same
for
the
yard,
where
architecture
like
arm.
While
we
talk
about
something
like
there's
or
some
new
technologies
being
invented.
N
A
The
way
that
I
see
it
is
the
first
five
bullets
or
the
descriptive
way
of
saying
here's.
What
cloud
native
is
here
all
the
benefits
that
you're
gonna
get
from
it
and
then,
if
someone
says
I
like
those
I
would
like
to
use
it,
then
you
have
this
follow
up,
and
you
say
how
do
I
integrate
it?
Well,
we
need
we
use.
We
follow
a
sixth
and
ERDs.
We
need
to
integrate
it
there
we
have
arm
in
different
parts
of
our
network.
We
want
to
integrate
it
there
with
the
first
white
paper
would
be
here.
A
F
All
right,
I
said
everybody
disagrees,
I
will
not
be
insist.
While
we
stop
insisting
good,
so
I
think
that's
the
content
for
now
and
and
obviously
time
is
running
very
short
both
for
this
meeting
and
both
on
their
cute
corn.
So
I
like
to
see
if
we
could
gather
a
very
operative
team
that
could
start
laying
down
some
text
now,
then
you
have
mentioned
that
you.
You
will
then
need
someone
who
will
do
coordination
on
this
work
right.
B
J
E
We
have
some
pros
written
to
sit
down
and
talk
through
these
through
these
things
to
make
sure
that
we
get
efficient
because,
like
we
have,
we
have
about
a
month
left
and
we
only
have
two
meetings,
one
of
which
is
that
a
convenient
time
zone
for
for
many
of
the
United
States
authors.
So
those
would
be
the
two
requests.