►
From YouTube: CNCF CNF Testbed Meeting - 2019-04-15
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
Sure
so
this
is
Dan
Kahn
from
CN
CF
I
wanted
I'm
glad
to
see
more
attendance
on
this
call
and
I
wanted
to
give
the
good
news
that
we've
decided
to
create
a
telecom
user
group
that
is
explicitly
modeled
on
the
CN
CF
and
user
community.
But
the
difference
is
that
our
group
is
going
to
not
just
be
for
telcos,
but
also
for
their
vendors,
and
that's
just
based
on
telcos
have
never
been
included
as
part
of
the
end-user
community,
which
has
been
a
rule
for
sorry.
B
Telcos
have
never
been
included
in
an
user
community
because
it's
a
rule
about
whether
you
include
cloud
native,
whether
you
offer
cloud
native
technologies
to
your
customers
or
not,
and
so
this
is
a
essentially
a
new
separate
group
that
is
both
telcos
and
their
vendors
together.
And
it's
not
a
group
that
I'm
expecting
to
produce
code
any
code
that
they
want
would
be
done
in
the
upstream
projects.
B
So
anyway,
that
work
is
meant
to
overlap
pretty
directly
with
the
CNF
test
bed
to
the
degree
that
people
are
trying
new
things
like
network
service
match
for
configuring,
EPP
or
using
Sto
for
policy
or
other
kinds
of
stuff.
We
would
love
to
have
that
not
just
be
a
white
paper,
but
actually
have
it
be
demoed
in
the
in
the
test
bed.
If
people
find
that
useful,
and
so
our
plan
then,
is
to
essentially
transform
this
meeting
into
a
telecom
user
group
meeting
and
have
the
testbed
be
one
of
the
topic.
B
About
this
group,
the
last
piece
of
it
is
that
I've
asked
your
hyung
who
couldn't
be
here
this
morning,
unfortunately,
but
who
leads
our
and
user
community
to
share
this
group
as
well,
it's
possible
after
three
or
six
months
that
she
might
step
aside
and
have
a
member
charity
involved
at
first
and
then,
which
she'll
be
the
actual
chair
and
I'll
stop
there.
Thank
you.
A
B
B
B
A
B
B
A
Great,
so
there's
been
a
lot
of
collaboration
with
different
groups
already
from
the
CNF
testbed
initiative,
and
one
of
those
would
be
network
service
mesh,
so
I
just
wanted
to
mention
that
it
was
accepted
as
a
sandbox
project
last
week.
A
There's
a
lot
of
feedback
at
ons
from
different
vendors
and
projects
that
are
actively
working
on
this.
So
it's
good
to
see
the
collaboration
increase
and
there's
been
requests
also
for
working
on
white
papers
from
the
Linux
Foundation
Network
inside
and
I.
Don't
know
if
someone
on
the
call
like
Fred.
Maybe
you
have
some
more
details.
If
you
want
to
talk
about
any
of
those
things
or
network
service
mesh
specifically
I
know
it'd,
be
most
of
these
things
are,
would
be
of
interest
to
anyone
on
the
call.
C
A
Awesome
ons
we
did.
There
was
a
lot
of
different
talks,
of
course,
one
of
on
the
scene
of
tests,
but
specifically
we
had
a
talk
where
it
was
a
mix
of
the
intro
for
the
testbed
and
then
went
through
a
tutorial
QA
and
that
went
I.
Think
really
well.
We've
been
working
hard
to
ensure
that
everything
is
repeatable
from
scratch.
So
for
this
folks
don't
know
on
the
call
which
I
think
may
not
be
anyone.
A
So
the
next
pieces
are
going
to
be
really
showing
off
I'm
stuff
like
orchestration
and
everything
which
were
in
a
semi
in
so
I
think
we'll
start
seeing
more
complex
and
interesting
use
cases.
So
one
of
the
main
areas
that
we
one
is
to
get
feedback
on
what
use
cases
are
most
interesting.
I'm
for
telcos
and
feedback
from
vendors
and
what
they're
trying
to
implement
and
tom
and
herbert
is
on
the
call
red
hat
he's
looking
at
helping
with
Cintas
support,
so
that
a
lot
of
the
telcos
would
be
using
rel
and
Cintas.
C
A
That
we'll
have
the
capability
of
using
Ubuntu
or
sent
us
on
packet,
and
one
of
the
things
right
now
that
that
gives
on
packet
is
more
access
to
their
the
machines
that
they've
built
out
for
this
type
of
network
configuration.
So
these
are
the
Intel
network
card
based
machines.
They're
publicly
released
right
now
they
are
actually
all
on
Cintas
side
because
they
targeted
some
telcos
that
were
wanting
these.
This
configuration
is
based
on
feedback
from
the
work
with
CNF,
testbed
and
Michael
was
a
part
of
that
and
Peter
and
my
check
from
the
cset
projects.
A
All
of
our
feedback
is
actually
led
to
the
configuration
it's
into
extra-large
and
they've
really
been
good
about
listening
to
that
and
they're
wanting
more
feedback.
So,
if
you're
on
the,
if
you're
not
on
the
packets
like
I'd,
definitely
get
on
that,
but
they're
also
on
the
CNF
channel
on
the
cloud
native
slack,
and
if
we
have
ideas
on
what
would
be
good
from
a
cloud
computing
or
cloud
provider
side,
they
want
to
try
to
create
those
configurations.
That
would
be
useful
for
this
industry.
A
A
Most
everything
is
right
now
on
the
development
side,
for
the
CNF
test.
Federico's
is
targeting
q
Khan
and
adding
looking
to
add,
ideally,
network
service
measure,
support
and
working
on
different
types
of
configuration
for
the
the
cloud
native
network
functions
that
are
privileged
and
unprivileged
and
dealing
with
stuff.
Like
that,
we
have
SRV
topics,
we've
been
talking
with
different
folks
and
getting
more
pull.
Requests
from
from
different
people
is
something
bill.
There
I
think
that's
it,
and
does
anyone
have
any
questions
or
topics
that
they
want
to
see.
C
C
Think
something
that
would
be
of
interest
to
the
telcos
would
be
in
the
of
the
test.
That
is
the
right
spot
for
this,
but
we
have
been.
We
had
done
some
initial
work
towards
network
service
match
and
in
a
neutron
based
integration,
where
you
can
call
you
de
Kooning's,
they
system
to
call
OpenStack
and
then
be
able
to
make
a
call
from
open
stacked
into
kubernetes
in
order
to
wire
services
together.
D
So
Taylor,
when
we
were
at
the
onus
last
week,
we
spent
some
time
talking
through
the
set
up
VPP
within
your
testbed
and
I
just
wanted
to
kind
of
go
through
this.
So
what
we
discussed
was
that
today,
it's
mostly
relying
on
a
static
VPP
configuration
for
the
conf
Wow.
It
would
be
good
and
we
already
have
some
work
done
within
NSM
and
by
VPP
I
mean
like
the
CNF
VPP,
like
the
thing
that
runs
inside
the
container
and
implements
the
function.
D
D
A
Sounds
good
yeah
I'd
like
to
continue
that
and
it's
something
that
Michael
and
Denver
and
I
had
been
talking
about.
I
think
there
may
even
be
a
ticket
as
far
as
on
that
side
and
I
think
it
could
be
split
between
creating
ensuring
the
CNS
itself
is
more
generic,
specifically
a
VPP
bassy
enough
and
then
adding
NS
m
or
whatever
stitches.
The
connections
together
can
probably
be
done
separately
so
that
those
can
be
completed
independently.
That
sounds
good.
Why.
F
C
So
my
preference
would
be
that
they're
that
they're
independent,
so
basically
bare-metal
kubernetes
talking
with
a
with
an
independent
OpenStack.
My
understanding
is
that
the
test,
the
test
that
already
has
an
openstack
system
running,
which
is
built
on
top
of
EPP,
and
we
also
member
service
so
much
how
they
can
attach
things.
So
we
have
lovi
people.
C
We
have
VPP
as
a
say,
like
they're,
curing
that
easier,
senator
kubernetes
networking,
and
then
we
can
also
attach
a
VPP
based
network
to
it,
and
then
we
can
grew
Nate,
the
parameter
like
the
X,
LAN
or
or
Gary
or
whatever
you
want
to
use.
So
in
the
OpenStack
side,
if
we
create
a
port
in
Neutron
and
then
we
were
to
take
the
port
and
instead
of
looking
up
at
the
end,
we
were
to
attach
the
X
LAN
port
about
the
upside
I.
Think
we'd
be
a
relatively
easy.
D
C
In
order
to
make
it
interesting
and
then
the
kubernetes
side,
we
could
also
accelerate
that
data
path.
So
we
could
do
shared
memory
from
the
pod
to
VPP
and
then
DV
peak
and
then
use
D
DK
to
emit
the
mix
lam
packet
on
to
its
destination.
So
I
think
we
have
something
that
I'm
not
gonna,
say
it's.
It's
easy,
but
I
I
think
it's
relatively
straightforward
and
certainly
easier
than
other
use
cases
we
could
take
on
that.
Would
that
would
show
some
significant
value
to
do
telcos
pretty
pretty
quickly.
A
F
F
Yeah
yeah
my
plate,
some
should
be
clearable
pretty
soon
here,
so
yeah
I
can
definitely
look
into
the
system
go
ahead
and
almost
on.
F
A
C
A
A
Do
think
that's
a
good
use
case,
and
so,
if
we
can
get
someone
else
to
drive
that
and
add
the
support
the
the
VX
and
stuff
has
been
going
in
as
far
as
the
VPP
networking
stuff.
So
that's
kind
of
been
testing
me
I
know
in
Wells
had
been
giving
some
feedback
on
some
of
that,
but
thats
related
to
add
in
the
adding
this
support
for
Frederick
and
can
follow
up
with
you
outside
and
see.
What's
left.