►
From YouTube: CNCF TOC Meeting 2022-09-20
Description
CNCF TOC Meeting 2022-09-20
B
B
B
Okay,
I
will
I
will
now
kick
us
off
for
four
minutes
in
so
welcome.
Welcome
to
your
Toc
meeting.
Welcome
to
your
normal
LF
antitrust
policy,
news
welcome
to
meeting
Logistics.
You
have
made
it
to
the
meeting
well
done
or
you're
watching
online
later
also
good
doc.
Members
present
today
is
going
to
be
attractive
for
the
TOC
public,
working
Tech
and
here's
our
agenda
today.
A
D
Come
on
in
oh
cool,
yes,
there
you
go:
okay,
okay,
great,
so
pixie
is
a
observability
tool
for
kubernetes,
a
little
emergency
evpf
to
collect
kilometer
data,
we're
currently
a
cmcf
Sandbox
project
at
a
high
level.
You
know:
we've
been
seeing
some
pretty
aesthetic
growth
and
contributions
and
continuing
project
growth
happening.
D
We
are
actively
working
on
adding
additional
maintainers
to
the
project,
both
from
you
know
within
so
stepping
back.
A
second
pixie
was
built
by
Pixie
Labs,
which
was
acquired
by
New
Relic
a
couple
years
ago
and
pixie
was
contributed
over
to
the
cncf.
After
our
acquisition,
the
entire
Pixi
team
is
still
very
actively
working
on
pixie,
and
we
have
quite
a
few
external
contributors
right
now.
D
We're
working
on
expanding
our
maintainers
to
include
folks
from
outside
of
New
Relic,
and
also
to
broaden
our
governance
process
to
include
folks
from
outside
of
Neurology.
So
we
want
to
get
to
be
a
good
vendor
neutral
status
right
now.
Our
governance
is
pretty
much
controlled
by
me
and
Michelle
and
we'd
like
to
make
that
more
Community
govern
over
the
next
few
months,
and
we
plan
to
resume
our
community
meetings.
Monthly.
A
C
Then
couple
of
things
I
didn't
we
didn't
talk
about
the
other
day,
so
you
know
the
calendar.
When
you
do
the
calendar
just
make
sure
it
gets
added
to
the
cncf
calendar
too.
That
would
be
one
that
one
would
be
like
when
you
start
writing
down
things.
If
we
I
don't
know
if
we
talked
about
so
please
talk
to
the
tag
contributor
strategy
to
get
their.
You
know
view
on
like
what
you're
writing
up
in
terms
of
governance.
D
A
E
Hi
yeah,
my
name
is
vinay
vinay,
Kulkarni
and
I
work
for
a
company
called
futureway
Technologies,
which
is
a
subsidiary
of
Huawei
and
cni
Genie.
This
is
this
project
had
been
in
the
back
burner
for
for
quite
some
time
due
to
staffing
issues.
Since
the
2019
government
restrictions
came
down
with
the
for
the
parent
company
and
they
had
to
cut
staff
to
manage
the
project
budget
budgeting
based
on
the
progress
that
they're
going
to
make
after
that.
But
now
it
looks
like
they
asked
us
to
take
it
over.
E
In
all
fairness,
they
asked
to
take
it
over
in
last
one
year
or
so
it's
been,
but
we
haven't
had
the
bandwidth
to
you
know,
give
attention
to
it,
given
all
the
other
things
that
were
going
on,
we
recently
hired
a
new
person
who
joined
us
a
couple
of
weeks
ago,
so
I'm
hoping
to
get
a
reboot
on
this
project.
I
think
the
need
for
this
project
is
there
in
the
sense
that
there
isn't
very
many
solutions
out
there.
E
That
does
multi-home
networking
and
we
have
a
another
project
that
we've
been
working
on
WE
presented
recently
in
various
conferences
and
I.
Think
the
technology
that's
based
on
xtp
is
really
solid
and
scalable,
so
we're
hoping
to
bring
that
with
the
native
isolation.
This
is
some
of
the
ideas
that
I
have
given
put
down
over
there
in
the
annual
review.
E
This
is
one
of
the
very
popular
solution
out
there
and
then
essentially
reboot
it
over
the
next
one
year
or
so
so
we're
gonna
start
doing
the
routine
regular
Community
meetings
and
as
soon
as
our
staffing
is
done
and
we're
gonna
have
there
I
think
there's
another
a
university
in
China,
which
is
potentially
there
are
researchers
there
who
might
who
are
already
working
on
this
project
and
they
might
contribute
to
this.
We
haven't
worked
out
the
staffing,
yet
it
depends
on
what
their
contribution
interest
are.
E
So
the
points
that
I
mentioned
there
are
pretty
much
I
think
the
bottom
line
is
that
yeah
we
want
to
continue
this
project
we
want
now.
We
want
we're
Staffing
up
and,
as
we
get
more
traction
and
see
how
it
goes
we'll,
probably
obviously
we
want
to
get
external
contributors
and
make
it
a
real.
You
know
beehive,
but
that's
going
to
be
a
little
while
I
believe.
E
B
Know
no
I
I
had
unplugged
my
laptop
from
my
normal
microphone
and
when
I
plugged
it
back
in
Zoom
decided
not
to
switch
back.
So
it
was
just
a
microphone
oops.
The
I.
Don't
really
have
any
other
questions
because
they
were,
they
were
covered.
I
mean
cni,
Genie
kind
of
started
to
stall
for
a
while
looks
like
it's
getting
reinvigorated
in
life.
C
C
Exactly
so
thanks
a
lot
for
that
and
I'm
sure
your
guidance
is
going
to
help
here.
So
one
question
like
I
wanted
to
ask:
was
you
were
talking
about
the
technology
itself,
the
ebpf
XDP
itself,
but
do
you
have
a
sense
of
like
what
use
cases
that
you're
going
to
start
with
like
when
the
new
people
come
in?
Like
do
you
want
to
do
exactly
what
they
were
doing
before
or
go
to
a
smaller
use
case,
like
you
were
saying.
E
Yeah,
the
first
use
case
is
essentially
bring
psyllium.
That
should
be
a
relatively
easy
thing
to
do,
for
the
new
persons
who's
coming
in
and
wrapping
up
adcillium,
you
know,
and
to
do
end-to-end
testing
make
sure
that
it
works
in
isolation
and
I.
Think
the
next.
The
big
thing
that
I
really
want
to
do
is
take
our
ebpf,
the
xtp
architecture
that
we
have,
which
natively
provides
isolation.
E
Today,
when
you
I
think
this
came
about,
the
idea
came
about
from
one
of
the
talks
that
I
I
listened
to
in
kubecon
EU.
This
was
I,
believe
Ikea
was
trying
to
use
psyllium
and
multis,
and
they
mentioned
a
bunch
of
issues
that
they
ran
into
and
one
of
the
things
they
were
talking
about
in
that
was.
They
want
to
have
multi-home
networking,
each
of
them
in
their
own
isolated
networks.
E
This
is
something
you
could
do
with
any
other
provider,
but
you
may
have
to
use
Network
policies
that,
if
you're
using
the
same
provider,
in
our
case,
we
do
that
by
at
the
packet
layer
we
use
the
geneve
isolation
geneve
as
the
overlay,
and
then
we
use
the
bnis
for
isolation.
So,
and
we
have
a
demo
of
that
working
for
a
new
architecture
that
we've
been
working
on
where
multiple
it's
a
kubernetes,
you
take
a
kubernetes
cluster
and
then
re-architect
it
so
that
you
have
one
API
server.
E
Talking
to
today,
you
have
one
API
server.
It
has
a
bunch
of
worker
nodes.
What
if
you
had
multiple
API
servers,
sharing
those
worker
nodes?
Now
each
of
these
API
Services
its
own
current
partition,
and
that's
where
our
solution
works
very
well,
I,
don't
think,
there's
any
other
CNA
solution
that
saw
and
our
solution
is
not
a
it's,
not
commercial.
As
you
know,
it's
more
of
a
prototype,
we
don't
have
the
Staffing
here
to
do
commercial
stuff.
E
We
do
a
r
d
and
then,
if
it
becomes
a
strong
need
for
the
parent
company,
they
do
it.
They
have
the
Staffing
to
do
it,
but
our
idea
is
to
essentially
present
the
technology
and
then
get
collaborators
contributors
in
to
work
with
so
it's
ecosystem
project,
essentially
so
that
that
use
case
I
believe
this
can
be
a
very
good
candidate
to
solve
that.
You
know,
provide
natively,
isolated,
multi-ten,
multi-home
networking
to
put
it
in
one
line,
sounds
good
thanks.
G
E
E
We
still
do
better
than
psyllium
in
a
couple
of
metrics
like
TCP
RR
test,
but
what
we
did
see
was
that
when
we
go
to
like
real
line,
speed
like
10
gig
next,
which
is
what
what's
common
and
it's
going
further
further
up
now
we
were,
we
were
capping
out
at
2.3,
2.56
gig
or
something
I
investigated
that
and
found
that.
The
reason
why
that's
happening
is
because
we
use
XDP
in
the
to
at
the
to
terminate
the
container
connection
like
when
a
pod
sends
out
traffic.
E
E
That
is
an
overhead,
so
we're
looking
at
potential
Alternatives
redesigning
the
architecture
a
little
bit,
because
in
the
past
one
year,
when
we
started
this
project,
we
did
not
have
anything
that
would
give
us
something
like
xtp
redirect,
which
you
know
shoots
a
packet
from
one
interface
to
another
one
and
in
the
last
one
here,
I
think
the
psyllium
folks
have
added
a
few
things
like
ebpf
redirect
peer.
These
are
a
few
apis
which
do
what
XD
period
redirect
does.
E
E
E
We
haven't
done
a
comparison
with
the
pdk.
I
would
suspect
that
dpdk
would
probably
perform
a
little
bit
better
because
you're,
essentially
taking
the
packet
and
putting
in
user
mode
and
processing
it
there,
but
then
you
also
have
the
limitations
of
dpdk
right.
It
doesn't
work
everywhere.
You
need
to
have
that
SDK.
You
need
to
have
that
make
or
you
need
to
have
that
the
processor
that
works
with
it
I
don't
know.
B
B
E
I
know
I
know
about
that:
I
haven't
we
haven't,
had
the
chance
to
do
a
comparison
with
dpdk
I
suspect
that
dpdk
will
probably
perform
better
better
than
XDP
or
ebpf
in
most
cases,
but
that's
yet
to
be
seen.
The
main
place
where
it
might
come,
xtp
might
do
better
is
when
we
get
offload.
You
know,
offload
ecosystem
is
still
in
in
its
very
infancy.
Ebpf
offload.
There
is
not
a
lot
of
links
out
there.
E
There
is
one
metronome
which
we
tried
out
for
some
time
last
year
and
we
ran
into
several
issues
that
metronome
on
paper
says
they
do
ebpf
offload.
But
when
you
try,
when
you
go
near
it
and
then
take
a
closer
look,
it
doesn't
support
xtp
redirect
it
does
not
support
tail
calls,
which
is
some
of
the
things
that
we
were
using.
So
there
are
limitations
to
that.
E
But
well
it's
too
early
to
say
there
might
be
some
I
just
had
a
call
with
with
the
CTO
of
another
company
who
is
working
on
hxtp,
which
is
Hardware
xtp
offload
that
was
yesterday
so
I'm
digesting
the
information
I
got.
There
is
potential
for
offloaded
XDP
to
you,
know,
give
perform
on
par
with
report
yeah.
C
A
Thank
you
so
much
for
all
of
that.
Additional
information
for
folks
that
have
more
questions
I
would
recommend
that
you
file
issues
with
the
project
so
that
the
discussion
can
take
place
in
a
recorded
location
and
be
shared
with
everyone
else.
Thank
you
again.
Next
up
is
fluid.
Are
there
any
project?
Members
from
fluid
on
the
call
that
I'd
like
to
talk
about
the
project
services.
G
If
there's
no
one
I
can
go
through
this
thanks,
Kathy,
okay,
so
Floyd
provides
cognitive
native
distributed.
Data
sets
orchestrator
for
data
intensive
applications.
It
is
going
very
well.
The
project
is
very
healthy
and
static.
It
has
very
healthy
and
steady
contributions
that
has
you
know
by
weekly
on
community
meetings.
Consistently,
has
five
maintenance
form
committers
and
non-new
adopters?
Last
year
a
total
is
15
and
has
two
releases
and
the
maintenance
committees
come
from
not
just
one
company
I
think
at
least
four
or
five
different
companies.
G
G
Yet
they
are
trying
to
get
work
meters,
contributors
and
users,
and
then
they
will
consider
applying
for
the
incubation
project,
so
they're
asking
cncf
similar
to
other
projects,
to
increase
promotion
of
the
project
and
basically
more
channels
to
Advocate
their
project,
and
they
would
like
to
collaborate
with
other
parties
in
the
cncf
and
even
it
could
be
out
of
cncf
and
also
they
want
some
relaxing
technical
writing.
Support
for
documents.
A
C
You
know
increase
not
just
contributions
but
also
ad
maintainers,
so
governance
was
something
that
they
are
looking
at
to
do
better.
They
I
don't
think
it
was
mentioned
here,
but
they've
started
like
new
sub-projects
as
well,
based
on
Andrea,
so
using
Andrea
as
a
as
a
layer.
You
know
as
underlying
component
to
and
they've
added
some
contributors
and
maintainers
there
as
well.
So
overall
they
are
doing
okay.
What
they
need
right
now
is
hey.
C
C
You
ended
up
tweaking
that
quite
a
bit
to
see
how
we
could
get
additional
signals
from
different
types
of
folks
and
not
just
the
way
it
was
awarded
before
so
I,
so
based
on
that
I
think
they
they
are
kind
of
ready
for
this.
I
might
have
to
look
at
how
they
are
doing
security,
I,
don't
know
if
they
have
talked
to
tax
security.
Yet
that
might
be
something
that
we
might
have
to
get
them
to
do
so:
security
and
governance.
C
C
C
C
They'll
stay
for
a
couple
of
you
know
PRS
and
issues,
and
then
they
go
away
right
like
because
you
know
they
come
here
because
something
that
they
are
trying
to
do
is
not
working
and
when
they
are
happy,
then
they
move
on
right
so
that
that
is
the
kind
of
challenge
that
they
are
facing,
because
you
know
at
some
point
you
need
to
get
maintainers
you,
don't
you
don't
want
people
who
are
just
doing
features
right.
So
that
is
the
usual
issue.
A
A
A
The
traffic
agents
they're
working
on
making
sure
that
it
does
not
disturb
the
existing
Network
policies
and
checks,
including
providing
Network
policies
for
system
admins
and
specific
constrained,
firewall
requirements
and
instructions
to
configure
load.
Balancers
telepresence
must
they're
also
looking
to
have
it
easier
to
deploy
in
the
context
of
large
clusters
with
multiple
different
kinds
of
applications.
A
They
have
a
few
asks
of
the
cncf
they're,
looking
for
expanding
their
user
base,
with
more
contributors,
tutorials
and
Improvement
of
the
documentation,
and
they
are
looking
to
welcome
contributors
who
have
expertise
with
complex
environments
or
specific
use
cases
and
share
how
they
got
telepresence
to
work.
For
their
particular
use
case,
integrating
with
other
tools,
especially
on
cloud
native
ones,
does
anyone
have
any
questions
associated
with
telepresence.
C
So
one
one
other
thing
that
I
wanted
to
say
here
was
like
just
like
any
other
sandbox
projects.
They
are
having
the
same
issues
around
hey.
How
do
we
increase
our
contributive
base?
How
do
we
publish
tutorials?
How
do
we
talk
to
and
improve
engagement
with
other
CNC
of
projects?
So
you
know
you
know
that
trend.
C
Is
there
right
like
so
it's
the
same
set
of
problems
that
they're
facing
here
as
well,
then
the
other
one
I
think
Amy
you
touched
on
it
already
is
like
when
the
happy
path
everything
is
working
like
if
it's
this,
if
somebody
is
doing
something
different
from
the
happy
part
that
they
have
documented,
then
they
run
into
trouble.
So
that
is
why
they
are
requesting.
You
know
people
to
try
out
telepresence
in
complicated
deployment
scenarios,
so
they
get
feedback
on
that.
C
So
one
thing
I
had
in
mind
here
was
for
these
folks
to
somehow
talk
to
end
user
Community
to
get
that
kind
of
feedback.
Maybe
they
can
do
a
survey
or
something
to
and
engage
somehow
with
the
end
user
Community
to
see
if
folks
are
using
telepresence
and
how
to
get.
You
know
feedback,
so
they
can
update
their
docs
and
whatnot.
A
Yep
makes
complete
sense:
yeah
anytime,
you
were
struggling
with
project
adoption,
It's,
usually
the
edge
cases
that
catch
everybody
and
and
inhibit
that
growth
rate.
So
if
you
have
experience
with
telepresence
or
not
yet
recommend
trying
it
out
or
any
of
the
annual
sandbox
projects
that
we
were
covering
here
today
and
then
providing
feedback
to
the
projects
via
an
issue
or
I'm
reaching
out
to
them
in
their
corresponding
slack
channels.
Any
other
questions
are
on
telepresence.
A
F
Yeah
I'm
here
hello
right
so
for
this
project,
I
think,
overall,
it's
on
the
right
track.
It's
growing
fast,
it's
GitHub
metrics.
It
really
helps
the
pretty
active
and
I
also
check
these
contributions,
and
they
are
doing
really
great
on
in
in
general
and
I.
But
I
do
have
two
suggestions
regarding
to
this
project.
I
think
the
first
is
I
noticed
that
majority
of
the
maintainers
actually
are
almost
only
maintainers
as
well
as
the
contributors
are
from
the
single
company,
which
is
I
think
still
need
to
improve.
F
The
second
part
is
regarding
to
the
adoption
you
know.
Rainbow.
Assembly
is
essentially
a
Cutting
Edge
technology,
so
we
should
not
expect
if
that
super
at
a
super
large
scale,
but
I
do
suggest
that
this
project
work
on
for,
for
example,
maybe
publish
some
user
studies,
especially
those
if
those
blog
posts
can
help
people
understand
how
this
project
is
helping
them
adopting
webassembly,
because
this
project,
the
position
itself
like
it's
well,
it
tend
to
intend
to
make
you
develop,
deploy
wrong
when
possible
applications
faster
either.
F
It
has
also
have
some
killer
features
like
a
very
fast
virtual
machine
for
random
assembly.
I.
Think
all
of
those
Technologies
are
good,
but
I
would
like
to
learn
how
they
are
helping
real
world
customers
or
real
world
users.
I
think
this
is
another
error
in
terms
of
adoption
that
can
improve
and
not
just
the
not
not
super
large
skills
adoption
but
more
like
more
examples.
Helping
people
understand
how
this
project
can
benefit.
You
can
benefit
other
people,
yeah
that
that
are
two
suggestions.
I
can
give
to
this
project.
G
F
B
G
B
F
I
I
do
see
that
I
would
go
to
move
to
incubation
I
think
their
timeline
is
I
think
end
of
this
year,
but
my
personal
suggestion,
as
it
may
be
a
little
bit
fast.
It's
not
not
very
practical.
Considering
for
the
first
issue,
as
I
said,
the
the
maintenance
are
still
from
the
single
company
and
the
adoption
we
need
to
see
a
more
you
know
more
activities
and
Improvement
in
adoption.
Parts
like
another
thing,
the
end
of
this
year
is
a
practical
timeline.
Yeah.
A
H
Yeah
see
everyone
so
I'll,
take
it
so
minorities
sending
memory.
You
know,
data
manager
to
help
with
the
abstracting
kind
of
pseudo
coping
memory
sharing,
and
it's
used
for
distributed
data
tasks,
graphic
analytics
numerical
Computing
certificate,
Computing
machine
learning.
The
project
has
seen
quite
a
lot
of
activity
in
the
last
year
since
it
joined
the
sandbox
earlier
last
year.
So
there
were
end
users
reporting
adoption
in
years
of
scientific
Computing,
research
facilities,
I
think
one
in
France
and
those
of
fintech.
H
H
The
GitHub
activity
shows
and
the
stats
show
quite
significant
growth,
especially
in
the
number
of
individual
contributors
in
total,
in
which
country
it
is,
and
the
number
of
organizations
building
contributions
as
well.
The
main
feature
that
they
highlight
in
the
project
was
they
added
integration
supporting
airflow,
and
this
is
in
addition
to
other
Integrations.
They
already
have
for
machine
learning.
Things
like
this
flow
back
to
our
GMX
net
and
others
and
also
Basque.
So
that's
quite
interesting,
and
now
they
added
support
for
their
phones.
H
They
mentioned
in
their
requests
for
the
cncf
more
opportunities
to
introduce
the
project
to
a
broader
set
of
end
users
and
increase
adoption.
So
this
was
already
mentioned
a
couple
of
times
today.
I
think
it's
it's
a
common
one.
They
specifically
mentioned
the
fact
that
they
submitted
a
few
proposals
for
coupon
and
those
were
rejected.
So
they
see
this
as
an
opportunity
to
get
more
visibility.
H
H
So
maybe
that's
something
to
to
look
at
and
the
other
one
is
that
on
the
website,
it's
not
immediately
clear
which
use
cases
the
project
focuses
on
so
making
it
more
clear
and
Publishing
more
case.
Studies
would
probably
help
with
with
the
the
project
as
well
or
at
least
the
perception
of
the
project
from
from
potential
end
users,
so
I
think
that's
all
I
have
actually
okay.
B
B
H
B
C
B
C
So
the
TOC
issues
918,
you
know
there.
This
was
the
first
time
we
tried
out
the
freeze
and
we
saw
some
edge
cases
as
usual
right,
and
so
we
want
to
iterate.
So
we
are
collecting
feedback
from
you
know
how
we,
you
know,
what
are
the
kind
of
things
that
we
talked
about
on
this
time
for
the
freeze
and
then
how
do
we
do
better
for
the
next
time?
C
A
A
I
would
also
add
that,
for
those
of
you
that
are
interested
in
understanding
more
about
the
level
of
effort
involved
in
going
through
due
diligence
due
diligence
refresh
understanding
the
activities
that
the
talk
participates
in
during
those
processes
for
graduation
or
moving
levels.
So
the
thread
on
that
issue
is
an
excellent
resource
to
understand
all
the
things
that
take
place
that
are
not
otherwise
documented
in
the
TOC
repository
at
this
current
point
in
time.
So
I
encourage
everyone
to
take
a
look
if
you
are
interested
in
pursuing
becoming
a
Toc
member
in
the
future.
B
No,
that's
perfect.
That
actually
leads
me
towards
a
Coming
Attraction,
the
the
Coming
Attraction
for
Toc
elections,
we'll
be
talking
about
this
kubecon
North
America,
but
the
nomination
cycle
kicks
off
in
January
for
Toc
and
seating
in
the
end
of
February.
So,
okay,
if
like
people,
people
are
interested
in
that
that's
coming
up
and
again,
my
last
piece
is
like:
if
you
have
people
that
you
want
to
nominate,
please
nominate
them.
A
A
Well,
if
there's
no
questions
yet
I
do
want
to
point
out
to
everyone
on
the
call
that
there
were
quite
a
few
themes
associated
with
several
asks
of
the
cncf,
primarily
around
getting
new
contributors
and
increasing
awareness
and
visibility
of
a
lot
of
these
projects,
specifically
around
documenting
use.
Cases
and
studies
associated
with
the
functionality
feature
set
of
these
individual
projects.