►
From YouTube: 2016-07-28 Kubernetes SIG Scaling - Weekly Meeting
Description
Public meeting recording of the Kubernetes Scalability SIG (Special Interest Group).
A
B
I'm
gonna
talk
with
corny,
with
the
chorus
guys
and
they've
released
at
CD
3:04,
which
serves
some
of
the
dependency
problems
we
were
having
with
trying
to
update
the
goal,
gets
into
kubernetes
I'm,
still
waiting
on
an
updated
tag,
release
of
rocket
the
reason
why
I
need
that
is
because
it's
a
transitive
dependency,
so
everybody
is
to
level
set
the
protobuf
version
in
order
for
us
to
build
properly
so
I'm
reading
on
that
level
set
to
occur
and
then
I
can
make
changes
in
sign
up
kubernetes,
hopefully,
test
set
as
default.
Prince
III.
B
A
B
A
B
B
C
A
B
A
Well,
our
our
thrust
is
going
to
be
on
the
performance
testing
and
once
we
art
our
hang-up
has
been
that
we
we
have
some
of
our
own
provisioning
tooling,
that
we
use
to
set
up
our
clusters
and
in
our
CI
system,
and
we
we
decided
to
support
a
couple
of
couple
of
these
testing
things.
We
need
to
do
we
needed
to
make
some
changes
so
we're
we're
busy
trying
to
get
that
done.
So
we
can
get
these
numbers
if
these
numbers
turn
through
and
then
we
have
a
few
other
things
on
our
list.
B
B
A
Thanks
to
do
since
we're
somewhat
on
this
topic,
do
is:
is
this
some
change?
Is
there
some
change
improvement
related
to
rocket
support
and
1.4
you're
saying
that
needed
a
rebase
line,
but
it's
the
credible.
A
B
B
A
The
one
thing
I
did
see
was
that
that
references
to
that
work
came
through
on
the
product
management
group
and
it
was
the
contacts
there.
I
mean
I,
think
Wojtek
started
off
with
talking
about
throughput
throughput
improvements,
but
by
which
I
think
he
was
talking
about
the
like.
Take
her
take
a
pod
scheduling,
request
and
schedule
a
bunch
of
them
and
ago,
as
opposed
to
doing
the
loop,
the
loop
through
every
time,
batch
batch
scheduling
of
pods,
so
that
specifically
didn't
show
up.
A
A
B
A
A
And
then
there
was
some
there
was
some
commentary
around
that
that
I
guess
I
was
slightly
confused
to
Wojtek
we're
here
we
could
get
a
clarification
from
him
on
it.
It
sounded
like
when
we
last
chatted
that
the
he
was
working
on
general
purpose
performance
improvements
and
it
sort
of
felt
like
that
that
scheduling
work
was
just
going
to
generally
benefit
from
it.
A
B
We
we've
got
our
OpenStack
aside,
install
up
there's
a
ton
of
networking
issues
inside
of
that
environments
because
they
have
let
the
latest
chipsets
and
not
all
drivers
and
firmware
have
been
settled
yet
he's.
If
you
go
into
that
environment
be
be
aware
that
there
could
be
demons
if
we
had
a
literally
update
the
firmware
and
a
bunch
of
machines
that
we
had
like
drivers
that
haven't
been
fully
vetted
for
that
Hardware.
Yet
I
I.
A
A
We've
we've
been
doing,
we've
been
doing
some
work
a
bit
a
bit
off
to
the
side,
not
directly
related
to
this,
to
this
sake,
thinking
a
pretty
hard
look
at
low
latency
networking
for
one
of
our
customers,
and
so
we've
been
digging.
We've
been
digging
in
on
drivers
and
other
things
here,
as
of
late,
so
I
think
it's
it's
good
advice,
though,
we'll
keep
it
in
mind.
Yeah.
B
A
Indeed,
and
I
don't
think
Rick
is
on
here,
but
there
was
something
our
networking
guy,
but
there
was
a
there
was
something
else
we
were
taking
a
look
at
and
I
can't
remember
what
it
is
off.
The
top
my
head:
oh
I'll,
try
to
remember
to
pinging
offline
just
to
see
if
you
had
a
quick
opinion
on
it
sure
sure
so,
alright,
well
any
other
any
other
topics.
For
today
we
can
call
an
early,
close
there's,
something
else
all
right,
we'll
call
it
a
we'll
call
it
a
wrap
for
today.
Everyone
thank
you
today.