►
From YouTube: Meshery Community Meeting - November 8th, 2019
Description
The @OctarineSec and @nservicemesh adapters move to stable just in time for #KubeCon!
A
A
Yeah
good
morning,
yeah
thanks
man,
I
was
gonna,
say
you
don't
know
how
hard
it
is
for
me
to
say
good
evening
when
it's
it's
this
early
in
the
morning,
oh
yeah,
yeah
well,
well,
I
just
wrote,
Raghav
and
I
were
just
chatting,
and
so
now
I
need
to
give
me
a
moment.
If
you
guys
would
I
need
to
go,
pull
up
meeting
minutes,
organized.
A
Assuming
you
have
access,
please
please
help
me
with
this.
Please
record
your
attendance
and
actually,
as
we
go,
please
feel
free
to
take
meeting
minutes
with
me,
because
I
generally
have
a
hard
time
doing
two
things
at
once.
So
so
your
help
here
will
be
good,
be
good.
I
know,
I've
only
asked
you
know,
I,
don't
know
how
many
times
to
you
know
take
take
meeting
minutes,
but
it's
just
I
can't
get
him
to
do
it.
A
A
B
A
Alright
good
so
we're
just
about
for
after
I
know
we're
there's
gonna
be
a
few
other
folks
who
are
gonna
join
today,
but
but
maybe
let's
get
let's
get
going
so
so
good.
So
this
doesn't
break
a
record,
but
it
still
warms
my
heart
to
both
have
to
two
new
folks
on
the
call.
So
Raghav
is
with
us.
You
know,
and
Dheeraj
is
with
us
as
well
guys.
C
Yeah,
surely,
first
of
all,
thanks
for
inviting
me
to
this
community
hall
and
iris
and
I
work
I'm
representing
Citrix
here,
I
work
on
our
conceit
tricks
and
I
worked
on
this
in
like
creating
this
signal
service
image
which
is
based
on
the
east,
use
one.
So
you
can
actually
go
to
only
go
to
the
github
and
search
for
Citrix
is
to
adapter,
so
that
is
likely.
C
Our
Citrix
is
like
its
own
containerized
ADC,
visual
project,
on
which
we
are
actually
installing
DP
in
the
data
plane
of
the
service
mesh
as
a
sidecar,
as
well
as
the
ingress
gateway.
So
I
have
to
primarily
work
on
that
and
I
look
forward
to
one
point,
our
like
make
sure
you
work
also
the
mesh
area,
the
Provost
it
gets.
Oh
yeah.
A
So
guys
it's
well,
it's
been
a
long
time
since
I've
seen
the
type
of
work
done
that
that
Dheeraj
is
done
and
we'll
have
to
I.
Don't
want
to
put
you
right
on
the
spot
now,
but
we
maybe
maybe
next
week,
we'll
have
to
maybe
even
ask
for
a
demo
from
you
Dheeraj
on
the
deployment
of
the
Citrix
EPX
as
as
the
sidecar
proxy.
It's
been
like
a
year
and
a
half
since
I'd
seen
any
project.
Do
that?
That's
so
that's
pretty
cool.
A
A
The
project
was
called
engine
mesh
I'm
gonna,
try
to
bring
it
up
engine
finish
so
guys.
What
this
was
is
for
sto
when
you
deploy
is
do
today.
The
default
service
proxy
is
envoy
and
it
had
been
shown
a
couple
of
times
in
the
past
that
you
could.
You
know
the
envoy
was
the
default.
That
was
the
battery
included
if
you
will,
but
you
could
take
that
battery
out.
A
You
know,
take
displace
envoy
with
engine
X
or
with
conduit
with
a
different
proxy,
and
the
engine
mesh
project
was
just
that
it
was
deploying,
is
do
but
using
nginx
as
the
proxy
and
that
I'd
gone
around
and
actually
that
this
is
this.
Is
that
was
actually
my
diagram
here
and
I'd,
given
a
couple
of
webinars
on
on
that
project,
but
it
had
been
set
down
and
more
or
less
archived
back
in,
like
the
SEO
dot,
seven
timeframe
and
right
now
we're
on
Mikey
Co
1.3.
A
C
Yeah,
that's
good,
that's
good!
The
cool!
We
know
that
his
to
a
pilot
is
like
exodus
server.
We
came
up
with
like
our
own
Exodus
client,
which
they
are
calling
it
as
base
to
adapter,
and
that
is
essentially,
as
you
rightly
guessed,
it's
an
interface
to
convert
our
like
this
XD
SAT,
as
into
the
sacred
CPX
I
mean
we
do
depends
configure
Citrix
appears
using
the
rest
based
API
nice.
A
Nice
yeah
I
I'd
assumed
that
all
of
the
proxy
providers
had
more
or
less
given
up
on
that.
So
this
is
cool
yeah.
We're
definitely
gonna
have
to
spend
to
spend
some
time
either
next
week
or
the
next
week
just
like
to
go
through
and
see
it.
As
a
matter
of
fact,
you
Roger,
actually
we're
gonna,
see
it
whether
we
want
to
or
not
I
think
as
you
go
to
do
some
work
on
the
measuring
adapter
for
CPX.
We
will
be
seeing
well
we'll
be
seeing
it
in
action
actually
in
the
project.
C
D
A
D
A
Too
bad,
okay
and
then
guys,
if
you,
if
you
can
help
me
with
the
note-taking,
because
I'm
not
going
to
do
a
good
job
anyway,
a
couple
couple
of
quick
topics,
so
in
general,
I
think
I
mentioned
may
had
mentioned
earlier.
So
all
you
know,
the
cute
cons
coming
up
in
a
little
over
a
week,
measure
will
be
presented
at
one
of
the
day:
zero
events
at
NS,
Khanh,
NSM,
Kaunas
network
service
mesh
con
so
far,
who's
in
our
community
I
was
just
chatting
with
her
earlier.
A
A
Another
topic
is
that
yeah
well
actually,
since
we're
on
cube
con,
let
me
also
say
that
I've
just
gotten
done
establishing
sig
network
in
the
CN
CF.
So
some
of
you
have
heard
me
say
this
before,
but
but
I
used
to
share
the
CN
CF
s--,
networking
working
group
and
working
groups
are
kind
of
going
away,
and
so
now
now
I'm
I'll
be
chairing
the
CN
CF
sig
network,
which
at
some
point
is
like
the
the
place
where
we'll
like
whoops,
where
we
will
likely
take
measure
II
through.
So
the
intention
is.
A
A
The
next
thing
is
I
meant
to
drop
this
into
our.
So
all
of
you
hopefully
have
access
to
the
measure.
E
team
drive
the
Google
Drive
I'm
gonna,
okay,
good
good
I'm,
gonna
drop
in
there
a
some
research
that
Forrester
has
recently
done
if
you're
not
familiar
with
Forrester
they're,
an
analyst
firm,
so
they're
like
a
Gartner,
a
4:51
research,
there's
a
an
analyst
firm
called
Forrester.
A
A
Service
meshes
in
jet,
you
know
it
broadly
and
the
the
kind
of
report
goes
through
and
introduces
micro
services
and
etc
and
kinda
talks
about
talks
about
the
landscape
and
they
so
he
Randy
the
VP
who
puts
it
together
this
research,
he
used
the
layer,
5
layer,
5,
I/o
service,
mesh
landscape
to
put
together
an
outline.
You
know
to
put
together
this
structure
that
you
see
before
you,
so
each
of
these
so
Superman,
Anand
and
and
everyone
else,
who's
helped
on
that
landscape.
A
Your
work
is
actually
being
represented
here
as
well,
and
so
my
Sheree
is
also
represented,
then
as
an
open-source
project,
and
it's
also
denoted
as
a
a
manager,
so
he's
a
manager
over
sto,
linker,
D
and
other
service
meshes.
So
it's
so
I
had
spoken
with
the
guy
who
put
this
together,
Randy
Hefner
and
made
sure
that
he
understood
the
place
that
massery
sits
as
a
manager
of
service
meshes
and
so
kind
of
interesting.
A
A
A
A
Some
of
the
status
and
the
mashra
CTL
improvements
for
windows
Anand.
Your
work
was
released
in
2.3.
So
that's
great.
Another
thing
that
we
did
here
that
we
confirmed
that
hadn't
been
confirmed
before
was
the
ability
to
deploy
measure
E
on
the
mesh,
the
ability
to
deploy
measure
II,
specifically
on
STM,
most
of
the
time
that
that
people
are
deploying
missery.
A
A
It's
often
the
case
that
people
want
to
generate
that
load
outside
of
the
cluster,
as
if
it
was
a
user
or
you
know
some.
Some
client
of
the
services
generate
a
bunch
of
load,
make
a
bunch
of
requests
and
send
that
into
the
cluster
send
that
into,
but
sometimes
people
actually
want
to
generate
that
load
from
within
the
cluster.
And
so
there
really
wasn't
any
changes
necessary
here,
but
just
a
validation
and
a
little
bit
of
documentation
around
that.
A
We
had
I
think
it
was
a
hack,
toe,
professed
contributor
had
moved,
had
updated
the
way
that
the
chips
look.
So
let
me
let
me
see
if
I
can
show
you
that
real
quick,
so
so
most
of
you
have
seen
measuring
and
kind
of
been
inside
the
UI.
It
used
to
be
the
case
that
these
chip,
these
little
adapter
chips.
Let
me
let
me
increase
the
they
used
to
show
the
port
number
that
a
given
service
mesh
was
connected
on
anymore.
A
It
just
shows
the
port
number
when
you
hover
over
the
tooltip
and
so
they're
kind
of
done
in
order.
I
think
the
first
adapter
we
had
was
for
is
do
the
second
one
was
for
linker.
D
third
was
for
console
fourth
octarine
and
then
fifth
NSM
and
then
D
Raja
will
see
if,
if
the
CPX
adapter
lands
before
there
are
two
other
adapters
that
are
out
there,
one
is
for
kuma
and
the
other
one
is
for
V.
A
A
A
So,
actually,
on
general
announcements,
maybe
there's
one
other
thing
to
say
in
that
bats:
well,
the
general
manager
for
for
the
Linux
Foundation
for
training
in
the
Linux
Foundation.
He
lives
here
in
Austin,
which
is
where
this
is.
Where
I'm
sitting
today
he's
been
asking
me,
but
in
some
respects
by
proxy
us
to
create.
A
Servus
mesh
training,
much
like
the
training
that
you
may
or
may
not
have
seen
coming
out
of
the
linux
foundation,
around
kubernetes
and
being
at
kubernetes
administrator
they're
asking
for
that
type
of
training
to
be
created
for
service
meshes,
and
so
so
that's
pretty
exciting,
because
the
reality
is
for
the
intro
level
training.
A
So
I
don't
know
that
I've
talked
about
that
on
the
community
call
before,
but
that's
exciting
that
that's
just
another
avenue
for
your
work
to
get
to
be
represented.
If,
if
you
look
at
some
of
the
statistics
for
how
many
people
have
taken
those
courses
like
for
the
kubernetes
course,
if
92,000
for
the
Prometheus
course
it's
a
little
less
than
that,
so
anyway,
we've
got
a
real
chance
for
your
work
to
be
shown
very
prominently.
A
Ok,
so
I
figured
I'd
walk
through
the
the
recent
releases
just
to
make
sure
everybody's
kind
of
seeing
the
fact
that
your
work
is
either
landed
or
see.
Some
work
that
you
haven't
seen
yet
so
so
the
latest
release
is
2.4.
It's
a
couple
of
days
old,
I'm,
sorry
0
to
4,
that
is,
that
brings
out
the
octarine
adapter
and
here
very
shortly
we
need
to
cut
another
release
that
brings
out
the
NSM
adapter.
A
We
also
enhanced
Mestre
CTL
in
the
way
that
you
can
deploy
it.
So
we're
gonna,
let's
demo,
these
things
right
here
and
we
also
got
better
better
script
and
I
think
Pavan.
If
you're
on,
like
you
know,
thanks
to
your
help,
we've
got
a
better
script,
around
measuring
deployments,
interfacing
with
eks
and
with
gke,
and
so
the
new
script
there
has
been
included
in
the
release.
Oh
there
is,
and
so
Ashish
is
with
us
as
well.
A
A
A
Way
of
installing
mastery
CTL,
so
if
you
go
out
to
the
lair
5io
org
you'll
find
a
new
repo.
This
homebrew
tap
repo
there's
been
some
update
to
the
measure
II
documentation
to
explain
how
to
install
measure
CTL
a
couple
of
ways
right
now
to
do
it
on
Mac
or
Linux,
one
of
those
with
homebrew
I
think
homebrew
is
fairly
popular
with
those
that
are
using
Mac
as
a
package
manager.
So
the
steps
are
these
three
right
here,
it's
relatively
straightforward:
let's
give
a
quick
demo
of
that.
C
A
A
A
The
the
three
step
that
we
just
looked
at
the
first
one
was
to
install
to
add
the
layer
v
io
homebrew
tap.
They
basically
add
that
that
registry
add
that
repository
to
your
homebrew
configuration
and
so
that
this
first
command
here
is
brew,
tap.
And
then
you,
you
just
reference.
The
repo
later
v,
io
/
tap
that
first
command.
Like
I
said
it
adds
the
that
repository
to
your
to
your
registry.
If
you
will
to
your
list
of
tap
providers,
the
second
command
here
would
then
install
my
Suri
CTL.
A
A
F
F
I
mean
sure
if
you
guys
want
to
1
meter,
demo
or
like
I
sure
can
but
I'll
need
a
few
minutes
to
just
set
it
up.
I
mean
probably
another
like
Ryu
4
minutes,
I
mean
you
set
it
up.
Yeah
I
guess
but
like
I
can
I
can
get
ready
for
the
demo,
no
matter
what
plus
other
than
that,
like
you
know,
I
do
have
the
other
tiny
thing
like
you
know,
that's
in
progress
too,
like
you
know,
which
is
actually
the
other
load
generator
art.
F
A
Be
great
to
see
maybe
while
you're
bringing
it
up
I'll
just
highlight
the
fact
that
probably
what
I'm,
assuming
last
week,
we
didn't
have
NSM
exposed
in
the
UI.
So
now
now
now
you
can
see
it.
That's
good!
So
maybe
I
won't
won't,
read
em,
oh
the
things
that
have
been
demoed
here,
but
since
we're
talking
about
adapters
Dheeraj,
this
is
well
I.
C
A
A
We've
spoken
of
this
one
a
couple
of
times
but
part
of
it,
helping
advance
that
service
mesh
API
that
that
sort
of
you
know
common
API,
there's
various
partners
out
there
that
are
doing
it,
some
that
are
much
more
engaged
than
others,
but
one
of
those
is
is
measure
II.
We
had
announced
a
partnership
with
SMI.
This
last
cube
can't
last
q,
Connie
you
and
so
the
the
spec
is
now
being
advanced,
some
which
is
good.
A
But
my
point
is
that
even
with
this
spec
they're,
really
each
of
the
service
meshes
themselves
are
unique
and
need
to
be
interfaced
with
you
know,
specific
to
that
service.
Mash
now,
given
CP
X's
deployment
in
context
of
an
sto
deployment,
the
the
nice
thing
is
that
it'll
end
up
looking
a
lot
like
the
sto
adapter
and
doing
most
of
the
same
things
that
the
east?
U
adapter,
does
it
just
you
know,
there'll
be
a
slight
tweak,
so
so
a
couple
of
things
that
adapters
are
capable
of.
A
So
not
only
do
they
speak
the
language
you
know
to
the
specific
api's
or
to
the
specific
capabilities
of
a
given
type
of
service
mesh,
but
each
of
these
adapters
are
capable
of
interfacing
with
more
than
one
deployment
of
a
given
mesh,
so
you
might
deploy
SDO
in
one
cluster
and
maybe
you've
got
miss.
Do
a
second
is
do
in
a
different
cluster?
You
you
don't
necessarily
need
to
deploy.
A
Multiple
adapters,
in
order
to
interface
with
those
clusters,
or
rather
I,
should
say
the
design
is
such
that
you
don't
have
to
today
today,
as
a
user.
You
probably
would
you
would
deploy
two
adapters
point,
one
at
one
and
one
at
one
cluster,
one
at
the
next,
so
you
can
have
multiple
adapters
of
the
same
type
deployed
at
some
point,
though,
will
make
it
much
better
in
the
UI
to
let
you
know
that
you
can
just
have
one
adapter
of
a
given
service
mesh
deployed.
A
So
in
the
case
of
CPX,
just
like
one
CPX
adapter
deployed
and
then
you
could
still
interface
with
any
number
of
different
environments
like
any
number
of
different
clusters
that
are
running
these
thdo
CPX
for
each
of
the
adapters.
If
you
click
on
them,
it
measure
will
just
do
a
quick
ping
test,
just
a
quick
test
to
say
that
measure
itself
can
talk
to
its
own
adapter,
because
those
are
deployed
in
independent
containers.
So
it's
just
a
quick
test.
A
When
people
go
to
run
measure
for
the
first
time
they
end
up
configuring
it
or
you
know
configuring
how
my
Sheree
is
is
running.
Usually
that
starts
by
pointing
it
at
a
kubernetes
cluster.
You
either
hand
it
your
cube,
config
file,
or
it
will
auto-detect
your
local
cube,
config
and
load
it
up
for
you
and
depending
upon
whether
or
not
you're
running
missery
inside
your
kubernetes
cluster
or
outside
your
community's
cluster
you'll
have
a
different
experience.
You'll
go
over
and
load
up
your
adapters.
A
If
you
need
to,
if
measure
we
didn't
autodiscover
them,
which
it
should
be
contestant
load
them
up.
If
you've
got
Gravano
or
Prometheus
running,
you
can
point
my
Cherie
at
those
endpoints
and
it'll
pull
back
a
lot
of
performance
data
and
performance
is
a
whole
separate
area
within
measure
e,
letting
people
do
tests
and
understand
the
performance
of
the
meshes
that
this
performance
area
is
not
real,
is
G
almost
entirely
not
built
into
the
adapter.
The
adapter
doesn't
need
to
consider
this
there's
just
a
small
bit
of
it.
A
A
A
And
so
yeah
there's
one
thing:
that's
unique
to
there's
two
things
that
are
unique
to
the
SD
adapter.
That's
a
bit
more
powerful
than
some
of
the
other
adapters.
Most
of
the
other
adapters
just
have
the
ability
to
install
the
mesh,
install
out
sample
applications
and
apply
custom
configuration
if
you're
looking
to
push
over
a
given
set
of
configure
given
manifest
is
to
use
a
bit
more
powerful
or
this
adapter
is
a
bit
more
powerful
because
it
has
these
two
other
things:
the
ability
to
apply
some
config
granted.
A
C
A
Good,
okay,
let
me
stop
sharing
and
ask
Grisha
if,
if
you
happen
to
be
in
a
position
to
show
some
new
things
for
performance
testing
for
load
generation,
so
the
load
generator
so
so
Grisha
is
on
mute.
But
hopefully,
while
he's
getting
ready,
the
load
generator
that's
built
into
mastery
today
is
called
for
I/o.
A
And
if
for
those
who,
we've
talked
on
this
call
in
the
past
about
some
analysis,
some
research
that
the
buoyant
team
had
commissioned
a
consultancy
called
Ken
Volk
to
test
to
do
some
performance
tests
of
linker
D
and
they
used
a
different
load
generator.
The
load
generator
that
they
had
used
it
was
is
called
w.
F
So
yes,
whatever
release,
that
is
right,
and
so
unfortunately,
this
piece
is
not
yet
completely
integrated
with,
but
I
can
give
you
a
high
level
direction
as
to
like
you
know
water.
What
will
be
my
next
steps
now
to
to
give
you
guys
an
overview
like
you
know,
so
we
have
a
project,
it's
called
eco
wrq
to
and
if
you
guys
have
looked
in
the
layer,
five
I
will
get
hub
org.
You
will
also
find
another.
You
know
repo
call
Toby
RT
to
odo
be
argued
to
wish
you
guys
go
inside.
F
You
will
see
that
it's
a
fort
from
the
Keene
walls
of
your
key
to
just
like
what
Lee
said
with
some
minor
changes
like
you
know,
which
were
mostly
in
the
Lua
script.
Now
here
is
Igor
double
your
key
to
looking
at
which
I
made
pricing
changes
like
you
know,
up
against,
like
in
a
work,
was
done
by
Sacco.
F
The
first
thing
was
actually
to
make
it
more
like
a
library,
so
here,
if
you
guys,
can
see
I
do
make
it
much
more
readable,
hopefully
so
fundamentally
I'm
trying
to
keep
it
simple,
but
you
know
so
there
are
two
methods
like
you
know
so
or
two
functions.
One
is
wrt
run
and
the
other
one
is
a
transform
double
for
IO.
F
F
Now
the
first
function
here
like
that'll,
be
our
key
run.
Essentially,
that's
what
it
does
it
executed
and
then,
like
you
know
so
here,
I
have
a
structure
like
you
know,
that's
defined,
which
is
a
mapping
for
the
results.
So
you
can
see
all
the
fields
that
are
extracted
from
that
I'll
be
our
key
results
and
then,
like
you
know,
the
percentile
information
is
noted
here
as
an
array.
Now
what
this
does
is
it
executes?
F
It
gets
a
JSON
which
is
with
the
help
of
the
Lua
script
and
then,
finally,
like
you
know
what
we
get
here
at
this
stage,
will
actually
be
at
JSON.
You
know
in
the
form
of
a
string
and
then
what
we
do
next
is
actually
try
to
marshal
it
into
a
struct,
because
it's
much
easier
to
work
with
this
trunk
and
again,
like
the
struct
definitions
right
here,
and
it's
of
course
like
no
much
more
readable.
F
Supporter
result
formats,
like
you
know
what
is
rate
if
you
run
a
result
and
the
other
one
is
gr.
Pc
o
WR
k
works
for
you
know,
as
far
as
I
know,
it
works
with
a
schooi.
So
what
I've
tried
to
do
here
is
you
know,
transform
the
results
into
the
appropriate
format.
So
so
here
is
actually
a
sample
for
dial
result
working
on
the
right.
This
is
like
the
traditional
for
dial
results
like
you
know,
so,
for
a
particular
example.
F
Here
we
result
and
then
I
have
this
other
file,
like
you
know,
which
I've
been
working
with.
You
know,
I
call
this
the
analysis,
dot
txt,
essentially
so
what
this
does
is
I've,
taken
that
we
arty
out,
but
I've
tried
to
map
it
to
the
respective
fields
and
then
I'm
trying
to
see
like
how
I
can
you
know?
This
is
how
I
worked.
The
whole
thing,
and
yesterday
like
I,
had-
and
here
is
the
rest
of
the
code.
F
I
can
always
just
I'm
not
going
to
walk
you
through
the
rest
of
the
code,
but
rather
I
have
a
set
up,
but
you
know
which,
in
order
to
make
my
life
a
lot
easier
to
work
with
yesterday,
like
I
I,
have
created
a
fundamental
base
image
with
w
RK
bundled
and
go
install
the
same
one.
So
I
don't
have
to
have
that
your
local.
F
So
if
I
say
Nick
docker
run
which
what
it's
going
to
do
is
it's
going
to
preload
the
image
with
W
R
K,
which
is
already
you
know,
be
built
on
the
local
and
then
it's
good
load.
The
go
code
like
you
know,
which
I
just
showed
you
guys
and
then,
if
you
guys
can
see
it
like
it's
created
as
a
module,
you
know
like
a
go
module,
so
you
know
it's
easier
to
work
with
that
way
and
and
I
also
have
a
make
file.
You
know
very
simple:
it's
not
fancy.
F
Like
you
know
the
first
one
is
the
get
w
RK,
because,
fundamentally
for
this
one
to
work
like
at
all,
I
think
I
already
mentioned
you
eat
a
relative
but
you're
at
a
tool
instance.
So
so,
which
is
the
reason
why
I
can
I
have
the
WR
kit
to
repo
clone
here.
So
if
it's
are
not
already
there,
you
know,
I
have
a
simple
make
file
command.
What
you
know
which
helps
you
to
get
it
and
then
the
base
build
image
is
actually
the
one
that's
going
to.
F
Actually
you
know
pull
the
paste
of
your
key
image.
I
created
a
docker
file
for
that,
like
you
know,
which
is
right
here,
it's
a
tsoukalos,
you
know
take
its
pre
installed
system
for
building
WR
k
at
that,
along
with
that,
it
will
set
up
close
at
the
make
command,
will
actually
fool
the
wrt.
That's
that's
pretty
much.
What
this
efficiently
does?
It
go
run
anything
it
just
preloads.
F
To
top
your
k,
sorry,
it
builds
top
Yorty
inside
the
container
and
also
has,
though
it's
at
all-
and
I
was
my
next
step,
which
is
the
doctor
rock.
What
this
does.
Is
it
actually
takes
that
base
image
that
essentially,
like
you
can
see,
I
have
a
volume
mount
which
pretty
much
loads.
My
cocoa
inside
there's
not
much
further
coat
here
like
so,
which
is
the
reason
why
it's
it'll
be
much
faster
and
then
I
like
I'm
running
it
like
so
every
time.
I
do
this.
F
It
just
runs
a
container,
but
you
know
at
that
preload
to
like
in
all
the
libraries
that
are
needed.
Of
course
like
this
can
be
optimized
like
an
hour,
but
for
me
this
was
done
just
yesterday,
so
yeah.
This
is
where
I
have.
So,
as
you
can
see
like
you
know,
the
test
did
run
and
let
me
just
show
you
the
main
file
here,
to
give
you
some
relevance
so
keeping
it
simple
again.
F
You
know
using
a
library,
so
you
know
so
that's
what
your
people
can
actually
see
it
in
the
command
line.
So
when
I
just
ran
it
or
like
you
know,
it'll
look
like,
in
other
words
a
lot,
but
but
it's
interesting
to
actually
process
the
whole
thing.
I'm.
Just
going
to
you
know,
maybe
rerun
it
like,
so
it's
it's
just
easier,
alright,
so
so
the
make
run,
you
can
see
that
you
know
it'll
first
get
the
modules
which
for
now
this
is
this
is
not
too
bad.
F
So
we
have
the
file
loaded.
So
you
can
see
here
this
first
part
that
we
see
the
output
is
actually
that
we
get
from
the
WR
k
you
know,
but
there
you
can
see
the
result
is
JSON
formatted.
We
have
a
lot
of
data
coming
out,
I
mean
earlier.
We
did
not
like
you
know.
I
tweaked
it
a
bit
to
get
a
little
more
data,
so
we
can
actually
chart
it
nicely
essentially.
F
So
so
that's
what
this
whole
section
is,
and
then
it
ends
here
and
then
this
this
line
here
essentially
like
you
know
it
tries
to
map,
it
sorry
tries
to
unmarshal
it
to
a
struct
and
that's
what
this
is
well.
Finally,
here
is
my
for
IO
result
which
I've
dumbed,
so
you
can
see.
This
is
actually
the
fourth
for
iOS
HTTP
run
or
results
format,
and
you
can
see
how
the
data
is
actually
loaded.
F
There
definitely
are
some
empty
fields
which
you
know
I'm
trying
my
best
to
extract,
but
you
can
see
all
the
other
fields
like
you
know
which,
which
have
been
populated
with
values.
So
the
reason
why
we
have
to
do
this
was
that
make
sure
he
is
actually
I
mean,
especially
the
user
interfaces
is
built
off
of
for
iOS
charting
framework,
and
then
we
extended
it
and
that
customized
it
so
like.
No,
but
but
fundamentally,
the
underlying
data
format
we
readily
accept
is
actually
a
four
io
one.
F
So,
in
order
to
make
our
life
a
lot
easier,
I
thought
it's
much
easier
to
do
the
transformation
on
the
back
rather
than
senator
DUI,
and
have
a
separate
branch
of
code
to
deal
with
it.
So
essentially
you
can
see
here
here
we
have
the
code
again,
which
does
the
transformation
and
it's
ready
to
pretty
much
be
accepted
in
the
UI
now.
So
all
I
have
to
do
now
is
that
the
library
is
ready.
F
I
have
to
you,
know,
load
it
into
my
Cherie
and
then
connect
the
dots
a
bit
and
hopefully
in
the
next
few
hours
like
I'll
have
something
functional,
so
we
can
start
using
WR
k
as
a
load
generator
option
right
front.
So
that's
that
pretty
much
concludes.
My
demo
I
think
I've
already
spoken
a
lot
about
it,
but
that's
that's
pretty
much
it
for
the
your
key
demo.
Anything
anything
else
you
guys
want
to
see.
Let
me
know.
A
F
So
one
of
the
interesting
things
with
trouble
urk
is
especially
the
one
which
the
kinwa
bikes
work
from
is.
Actually
there
is
a
concept
called
the
coordinated
emission,
and
you
know
rather
than
me
explaining
it
I
think
like
you
guys,
would
be
better
off
reading
about
it,
but
the
way
the
coordinated
omission
works
is
that
it
actually
tries
to
better
simulate
a
real-world
scenario
where,
in
you
know,
when
a
lot
of
requests
come
in
the
server.
F
Actually
you
know,
sometimes
it
actually,
you
know,
puts
them
in
a
queue
sort
of
way
of
processing
it
like,
as
a
result
of
which
the
first
route,
the
first
query.
You
might
actually
take
10
milliseconds,
but
the
second
per
you
might
take
20
milliseconds
and
then
the
problem
is
the
rest.
Will
actually
you
don't
have
a
consequent
impact
on
the
way.
F
You
know
the
responses
come
back
from
the
server,
so
coordinated
emission
kind
of
respects
that
when
it's
actually
doing
a
load
test,
while
the
other
load
generators,
like
you
know,
kind
of
you
know,
you
know
in
a
heartless
manner.
They
try
to.
You
know
hit
the
server
with
requests
based
on
whatever
is
requested,
so
you
know
it's
like
based
on
the
number
of
queries
per
second
number
of
threads
number
of
connections.
F
All
that
put
together
like
it
actually
tries
it's
best
to
you
know,
do
its
job
fairly
well,
but
coordinator
emission
is
probably
not
considered
in
most
of
those
low
generators
like
so,
for
example,
with
for
IO.
As
far
as
I
know,
I
mean
I'm.
It's
been
a
while,
like
I
went
back
to
for
iOS
documentation,
but
as
far
as
we
know,
like
you
know,
a
Ford
I/o
doesn't
consider
coordinated.
F
A
So
that's
cool
one
thing
about
the
WR
k
to
work
is
that
I
know
the
IBM
group
that
works
on
East
iou's
the
performance
and
scaling
that
they're
highly
interested
in
the
notion
that
measure
e
now
becomes
that
the
load
generator
has
now
become
swappable
because
they
have
one
that
I
think
that
they'd
like
to
use
within
measure
e.
So
that's
exciting.
The
last
thing,
maybe
that
I
showed
this
earlier
on
the
call,
but
I,
don't
know
how
many.
A
If
everyone
that's
on
now
was
on
then
I
said
that
Forrester
a
principal
analyst
there
had
used
the
layer,
5
service,
mesh
landscape
as
a
reference
for
writing
up
a
report
and
then
also
for
giving
a
webinar,
and
these
are
the
slides
from,
and
so
the
webinar
was
plan,
a
service
master
for
your
micro
services,
future
the
Forrester
from
mr.
Hefner
he's
in
Dallas
Texas.
A
But
point
is
again:
I
just
wanted
to
show
you
to
share
with
you
all
that
your
work
is
being
represented.
That
measure
he's
here
as
an
open
source
project
measure.
He
is
here
as
a
manager
of
multiple
meshes,
which
was
a
bit
of
a
new
concept,
I
think
for
the
researcher.
He
had
considered
that
that
measure
it
was
like
a
service
mesh
itself
and
so.
A
A
Anyway,
that
that
group
said
that
they're
gonna
have
an
announcement
at
cube
con
about
service
mesh
hub
and
I
believe
the
announcement
is
that
it'll
be
multi-tenant,
so
they
they
actually
pre
announced
that
at
all
things
open
a
couple
of
weeks
ago.
So.
A
Good,
let
me
shut
my
Yap
err
and
say:
hey,
you
know
so
one
just
stoked
that
Dheeraj
is
here
and
stoke
that
Raghav
is
here
Raghav.
If
you
have
more
time
after
this
call,
we
can
continue
our
conversation.
This
is
good,
also
provide
okay,
good,
good,
good,
good,
Pavan,
first
time
that
weird
that
your
PR
has
landed
so
Pavan
you've
been
here
a
really
long
time,
but
but
yeah
I'm
happy
for
you
stoked
that
you're
you're
enhancing
the
project
like
that
around
eks
support.
A
A
So
I
think
we
have
the
mailing
addresses
for
most
of
you
and
so
I
might
find
myself
at
the
post
office
for
a
few
hours
trying
to
get
you
guys
have
copies
kiss,
because
again
your
work
is
represented
in
there,
I'm,
not
sure
he's
represented
in
there
a
couple
of
times,
so
I
think
you
should
be
proud
of.
It
would
be
cool
when
you
guys
have
got
it.
A
C
So
I
personally
wouldn't
be
gender
cube,
con
I'm
in
Bangalore,
but
Tara
deep
and
probably
purple
of
isometric
should
be
there.
Oh.
A
Cooper
these
days
in
India,
I
think
in
Bangalore
like
a
little
over
a
month
or
so
ago,
but
but
this
cute
Khan
in
the
is
in
North
America
it's
in
San
Diego
in
just
about
a
week.
So
not
not
this
coming
week,
but
the
following,
and
so
so
yeah
it's
just
it's
just
around
the
corner.
I
gotta
tell
you
of
the
various
open
source
groups,
the
communities
that
I'm
involved
in
there's
been
a
lot
of
conference
driven
development
happening
so
I.
Do
people
try
to
get
things
done
before
the
conference,
so
I
didn't?
A
E
A
A
Okay,
so
the
things
is
that
I'm,
hoping
to
finish
up
a
conversation
with
fourth
Raghav
I'm
Jewish.
Could
could
you
message
Raghav
and
ask
him
to
come
back
on
this
call.
Raghav
wants
to
participate
in
the
community
through
google
Summer
of
Code.
So
that's
awesome.
It's
great
Nuno.
The
reason
I
was
and
the
reason
I
was
calling
out.