►
From YouTube: Layer5 Community Meeting (June 12th, 2020)
Description
Welcome @smhayg!
B
B
A
Like
on
teams,
what
you
can
do
is
like,
if
you're
not
part
of
the
the
entity,
you
can
still
set
it
to
accepts
automatically
like
permitting.
So
it's
not
like
a
normal
feature,
but
permitting
you
can
set
it
like.
Okay
accepts
automatically
anyone
that
joins
as
guests,
which
is
using
the
web
webpage.
So
I
don't
know
if
Google
Hangouts
may
be
permitting,
though,
can
do
the
same?
So
you
don't
have
like
to
accept
yeah.
B
Yeah
yeah
good
I
checked
on
those
settings
a
couple
of
times.
It
seems
as
if
that
your,
that
is
that
behavior
is
only
acceptable
for
those
who
are
part
of
the
same
domain
like
the
same
G
suite
domain
unless
I'm
missing
a
setting
I
know.
Actually
this
is
a
good
question
kind
of
while
we're
waiting
for
those
that
join
other
Google
meet
calls,
whether
you're
hosting
one
or
just
joining
one.
You
recall
if
okay
yeah
now
I
think
we
looked
at
this
before
that.
B
B
A
number
of
you
aren't
directly
invited
to
the
call,
rather
you've
been
added
to
the
community
mailing
list,
and
so,
by
virtue
of
being
a
member
of
the
mailing
list,
you're
invited
to
the
call
and
I
don't
think
that
Google
meet
recognizes
that,
like
it
doesn't
open
up
the
mailing
list
and
yeah.
So
when
you
join
you're
still
asked
to
be
admitted
so
anyway,
so
context
for
everyone.
That's
listening
to
this
that
we've
been
there's
a
little
bit
of
a
bothersome
thing,
the
behavior
with
when
people
join
the
call.
B
You
have
to
wait
to
be
admitted
and
that's
not
really
the
intention
of
the
call
you
should
to
be
able
to
freely
come
and
go,
and
so
we've
been
considering
using
zoom
instead
of
Google
meat.
For
that
reason,
and
for
the
reason
that
the
zoom
account
that
we
have
will
allow
us
to
livestream
these
meetings
so.
A
B
Other
than
I
think
the
slight
delay
in
some
of
our
zoom
conversations
just
because,
just
because
it
takes
a
little
while
the
latency
as
the
Chinese
are
listening
in
that
that
that
round-trip
takes
a
little
while.
So
how
did
that?
That
was
a
that
was
a
crack
at
all
of
all
of
our
Indian
community
members.
D
B
B
E
Yep
so
I'm
Ishita
Kumar
I'm,
a
master
student
at
Indiana
University
in
Bloomington
right
now
and
I'm
just
so
excited
to
be
a
part
of
the
neophyte
community.
I
was
actually
a
previous
user
of
measure.
Ii
and
I
just
became
a
part
of
this
open
source
organization
and
I'm
so
excited
to
be
here
and
yeah.
C
B
B
The
meeting
minutes
are
a
community
curated
thing,
so
please
jump
into
them
slap
your
name
down
on
a
list.
If,
if
we
didn't
get
you
just
yet
and
with
that
I'll
say
it's
fantastic
to
have
some
new
faces,
we've
got
a
few
other
folks
that
joined
this
week,
which
is
very
nice,
hopefully
they'll,
be
able
to
see
today's
recording
a
little
bit
later.
B
We're
excited
to
have
each
of
you
there's
a
couple
of
you
are
already
placed
into
a
bucket
if
you
will,
in
the
in
a
good
way,
I
think
Dheeraj
and
Shivam
are
or
shabam
rather
are
interested
in
some
of
the
work
that's
going
on
with
SMI
and
the
SMI
conformance
tool.
That's
actually
good
topic
and
I
won't
go
through
the
rest
of
the
list,
but
it's
great
to
have
all
the
rest.
B
The
folks
that
have
joined
today
on
the
community
calls
by
the
way
for
those
that
are
pit
knew
we
end
up
talking
about
well
any
number
of
things,
but
sort
of
broadly
across
the
ecosystem
and
in
what's
happening
in
service
mesh
land
and
then
specifically,
what's
going
on
with
a
number
of
the
initiatives
that
each
of
you
are
participating
in
or
early,
and
so
we
do
have
other
weekly
calls
and
their
agendas
are
slightly
different.
And
so
so
this
one's
a
bit
broader
in
nature.
B
A
Let's
say
that
we
put
every
week
and
first
then,
based
off
at
some
point
in
time,
so
just
to
not
make
like
it's
a
Cinderella
movie.
It's
like
been
what
three
years
four
years
that
I've
been
like
blogging
a
lot
tweeting
a
lot.
So
it's
like
getting
now
kind
of
the
results,
but
that's
not
what
I
I
do
for
it
and
likely.
So
we
do
it
for
the
community.
And
then,
when
we
had
these
nice
announcements,
then
it's
always
okay,
warms
the
hearts,
because
it's
not
only
the
community,
that's
easy!
A
But
it
goes
beyond
so
like
some
of
the
the
members
that
were
here
e
at
the
fishery
and
layer,
five
community
calls
and
helping
uki
I'm
seeing
them
more
and
more
in
Lindon
that
they
they
joined
like
companies
and
like
interns
first
and
then
maybe
they
will
join
another
company.
So
really
that's
yeah.
That's
incredible!.
B
He
is,
it
really
is
pretty
neat.
There's
you
know,
I
gotta,
say
I
actually
had
to
ask
you
earlier
just
to
try
to
keep
track
of
all
the
ways
in
which
your
being
your
work
is
being
recognized.
A
Yeah
but
I
would
say,
I
never
wanted
to
be
one
captain,
because
I
have
too
much
respect
for,
let's
say
the
legendary
captains
Lee's
one
nigel
is
another
Alton,
and
so
on
and
Stefan
share.
Also
so
that
you
are
my
captains
and
for
me
it
was
more
like
if
I
can
hang
out
with
these
guys,
like
just
talk
with
them
and
like
these
community
meetings
having
access
to
you,
it's
like
already
like
a
win.
So
the
first
thing
is
like
I,
never
set
captain
as
a
goal,
but
I
always
set
myself
to
learn
from
them.
A
Now:
captains
and
the
CN
CF
ambassador.
They
come
on
nomination,
so
work
hard,
like
my
I,
did
kendo
for
ten
years,
so
martial
art,
Japanese,
martial
arts
and
my
kendo
professors
and
say
it
was
always
saying
like
shine
as
much
as
you
can,
and
when
you
think
that
you
are
not
in
it.
Let's
say
in
not
in
the
place
that
you
think
you
deserve
shine
even
brighter.
Eventually,
someone
will
see
you
and
like
take
you
out
of
the
place.
You
are
and
bring
you
to
somewhere
else.
Okay.
A
Now
that
is
like
pure
philosophy,
but
actually,
if
you
want,
if
I
can
relate
to
my
experience
right
now,
it's
like
it's
been
like,
like
I,
said
four
or
five
years
into
the
let's
say
container
world
kubernetes
even
less
to
be
totally
sincere,
but
cloud
native
I
went
and
touched
to
everything
right.
So
it's
just
just
go
like
that,
be
passionate
and
then
eventually
we
will
see
you
other
people
outside
here
will
see
you
and
that's
how
you
get
the
denomination's.
A
It's
based
on
the
work
you
achieved,
not
on
the
work
that
you
will
do
like
the
work.
You
will
do
it's
it's
it's
important
and
will
be
then
even
more
impactful,
thanks
to
the
voices
and
the
rules
that
you
have,
but
at
first
don't
go
for.
I
want
to
be
captain,
maybe,
but
that
that
was
me.
Okay
again,
oh.
B
If
you
will
into
that
that
new
title
that
into
that
new
role,
really
is
what
you
know
is
saying
is
not
a
not
a
promotion
into
something
that
that
you
can
now
start
doing.
Rather
it's
just.
It
should
be
just
a
recognition
of
what
you're
already
doing,
depending
upon
the
role
that
can
be
really
difficult,
but
yeah
anyway,
so
so
new
nose
being
recognized
for
helping
and
sharing
with
others.
B
Sharing
with
others
there's
a
couple
of
upcoming
events,
ant
Weiss,
who
recently
wrote
a
blog
post
about
mesh
real
anding
inside
the
CN
CF
landscape,
which
we
should
include
here,
has
also
formed
a
web
assembly.
Meetup
I
think
you
know
sort
of
based
in
Israel,
but
I
expect
it's
probably
a
virtual
meet
up
and
I.
Does
anyone
know
if
his
first
meetup
already
happened?
B
If
it
hasn't
I
recommend
attending
I?
Think
he
just
did.
Maybe
he's
got
a
recording,
though
this
is
for
some
of
you.
This
is
very
front
and
center
in
terms
of
your
focus
in
terms
of
some
of
the
things
that
you're
doing
in
the
community
here,
for
others
of
you,
you're
saying
what
is
webassembly
and
what
is
wasum
and
if
you're
asking
that.
Please
ask
that
in
the
slack
channel,
because
most
of
us
probably
have
that
question
and
it's
a
good
piece
of
technology
to
talk
about.
B
A
I,
it's
I'm,
just
like
in
in
the
weirdest
spots
again
like
here
so
I'm
just
before
Brendan
burns,
so
no
pressure
and
for
for
the
ones
just
that.
Okay,
maybe
note
the
name
but
don't
know
really
who
is
is
like
one
of
the
tree
that
synthetic
burn.
It
is
so
yeah
again
no
pressure
and
I'm
going
after
my
dear
friend,
Ralph
Squillace,
which
is
like
going
for
crust.
A
B
B
Her
that
makes
sense
that
they
would
use
hers.
It's
people
more
people
might
actually
join
instead
of
using
my
my
ugly,
the
but
anyway
the
reason
that
I
said
people
might
chuckle
is
that.
Well,
there
are
those
in
the
service
mass
community
that
really
don't
think
that
SMI
has
legs,
and
so
so,
as
an
inside
joke,
I
get
a
chuckle
out
of
it.
The
crustless
stuff
has
anybody.
Looked
at
crosslet,
I.
A
A
It's
one
of
the
rare
things
that
you
can
bundle
or
cluster
with
the
doctor
desktop
kubernetes,
so
cross
that
can
be
bundled
together.
So
you
have
like
a
kind
of
two
nodes,
so
you
have
the
darker
desktop
nodes,
kubernetes
nodes,
which
is
a
one-note
only
and
then
you
can
attach
cross
right
or
closer
eyes
cluster
it
with
it.
But
I
never
went
too
far
because
again
I'm
not
like
a
rust
guy,
but
the
team
is.
If
you
build,
there
was
okay
or
web
application
in
in
rust.
A
B
A
No,
this
one
is
specialized
for
us
I
think
the
the
principle
might
be
the
same
as
the
visual
colet's,
but
yeah
yeah.
B
Good
yeah-
and
that
was
kind
of
the
analogy
I
was
trying
to
draw-
is
that
had
the
scheduling
of
yeah
I
had
lawfully
I,
don't
know,
I've
been
long
awaiting
VMware
too
well
do
do
tenzou,
I,
guess
more
or
less
for
some
time
where
they've
been
doing
scheduling
for
quite
some
long
time,
but
it's
been
about
a
particular
type
of
entity
of
VM,
so
kubernetes
for
containers,
but
but
yeah.
So
it's
very
interesting
to
see
the
scheduling
of
webassembly
programs
in
this
fashion.
B
So
it's
something
for
those
that
are
focused
in
this
area
to
reflect
on
some
of
the
use
cases
that
we've
discussed
in
this
community
have
been
a
little
a
little
bit
of
a
different
motion.
They've
been
about
running
web
assembly
programs
in
the
service
master
data
plane,
but
those
programs
are
portable
and
so
for
those
of
you,
who've,
looked
at
the
service
mesh
performance.
B
Well,
it
may
not
have
all
the
information
that
it
needs
to
calculate
mesh
mark
and
so
running
that
same
program,
but
running
it
in
a
note
in
a
kubernetes
node
might
make
sense,
or
is
it
opens
up
some
interesting
things,
I
think
it's
it's.
We
should
do
some
thought
fair
enough
or
a
couple
of
upcoming
events
there's
another
one
that
open
ships,
Commons
gathering
is
happening.
Those
those
will
typically
be
scheduled
kind
of
alongside
cue
cons.
B
B
But
one
of
those
is
establish
a
new
standard
for
how
to
measure
the
performance
of
a
mesh,
and
while
we've
had
some
success
in
doing
that
within
this
community,
we
were
going
to
go
ahead
and
bring
that
those
set
of
discussions
to
the
CNC,
F
and
host
those
discussions
coming
in
under
that
umbrella.
If
you
will
and
I
think
you
know
under
the
under
the
hopes
of
garnering
some
additional
contribution,
some
additional
involvement
so
really
there's
not
much
left
to
do
other
than
schedule
a
meeting
time
and
invite
people.
There
is
essentially
a
an
agenda.
B
B
Yeah
and
actually
so,
I
I
chuckle
I
give
a
smile,
because
that's
a
great
I
like
where
your
mind
is
going,
we've
kind
of
talked
about
those
kinds
of
things
before
some
of
those.
We
talked
about
some
of
the
use
cases
that
were
very
similar
to
what
you
just
said
were
lighter
weight,
maybe
in
nature
in
terms
of
actions
that
or
how
the
infrastructure
should
be
affected.
B
Maybe
maybe
it's
a
certain
account
or
a
certain
tenant
versus
others.
So
a
number
of,
like
a
number
of
actions
to
take
a
number
of
ways
to
in
real-time,
redirect
or
reconfigure
the
infrastructure,
including
a
use
case
like
you,
know,
executing
some
care
form
so
so
seeing
it.
Maybe
if
you
would
say
some
more
that
that's
a
what's
an
example
of
how.
F
F
For
example,
whenever
you're
doing
polls
from
your
blob
storage,
you
can
measure
based
on
the
public,
a
public
API
is
are
being
pulled.
For
so
say
one
cluster
is
getting
hammered.
Another
cluster
is
not
being
reached
at
all.
You
know
that
the
routing
is
not
doing
a
good
job
there
or
a
lot
of
times
like
this
happens
more
in
function
than
compute
a
lot
of
times.
B
So
somebody
will
interrupt
me
hopefully,
but
hasin.
Let
me
ask
maybe
just
a
little
bit
further
in
that
last
example
about
overloaded
functions
and
and
the
effect
of
information
that,
maybe
is
that's
cached,
maybe
that
they're
using
or
that
say
a
little
bit
more.
If
you
would
I
mean
there's
some
concerns
around
functions
and
they're
they're
cold
start
time.
Some
that
are
maybe
kept
up
is
part
of
the
concern
that
some
of
these
are
kept
up
too
long
and
that
you
would
desire
for
their
cache
to
be
invalidated
or.
F
Longer
than
you
want
you're
going
to
have
something
that
I
would
call
a
cache
creep
where,
like
there's
a
lot
of
data
that
you're
holding
on
to
that
you
shouldn't
be,
that
should
have
died
already
and
then
sometimes,
especially
when
you're
doing
like
any
hash-based
kind
of
activity.
You
get
into
problems
of
all
the
writing.
Some
hashes
and
things
like
that,
which
is
not
really
good,
because
you're
doing
you're
performing
the
same
operation
for
the
same
type
of
request
over
and
over
and
it
is.
B
Yeah,
that's
interesting,
I
think
you
know
what
gets.
Maybe
if
you
take
that
ones,
because
some
of
these
challenges
are
surmountable
and
in
various
ways
like
in
the
absence
of
the
service
mesh,
it's
possible
to
address,
you
know,
cache
creep
and
making
sure
that
your
you
know
your
data
is
fresh,
but
in
the
presence
of
the
mash
you
might
be
able
to
I.
Think
you
you
might
as
an
architect
or
as
someone
you're
running
running
the
infrastructure.
B
You
might
step
back
and
say:
well,
hey
will
you
know
which
one
of
these
approaches
is
what
I
want
to
lean
into
which
one?
What
I
want
to
do
and
why
I
think
that
an
argument
for
using
a
mesh
in
this
way
that
there's
there's
probably
a
few
arguments
but
I'll
just
choose
one.
That
is
that
you
might
be
able
to.
B
Well,
avoid
like
on
an
ease
in
an
easier
way,
avoid
statically,
defining
your
thresholds
for
something
or
even
like
for
the
hash-based
example,
where
you've
got
these
activities
that
are
going
on
and
you're.
You
know,
and
at
some
point
you
reach
this.
You
reach
this
point
by
which
you're
overwriting
things
that
you
you
shouldn't
be.
B
F
Helps
for
manual
life
and
I
know.
This
is
true
for
other
some
other
companies
as
well,
especially
companies
for
deal
with
sales.
You
sometimes
have
seasonal
notes,
where,
like
there's,
Black
Friday
various
like
the
Cyber
Monday,
or
something
like
that,
where
the
amount
of
resources
you
need
to
deploy
is
not
very
easy
to
determine
you're
going
to
have
to
be
ready
to
deploy
a
lot
more
than
you
normally
do,
and
sometimes
you
can't
reach
the
demand.
F
So
there
are
a
lot
of
strategies
that
people
use
so
that
they
can
lower
that
threshold
of
not
being
able
to
meet
the
demand
so
that
we
hopefully
don't
get
to
a
four
three
or
four
for
you.
You
wanna
make
sure
that
services
are
as
available
as
possible,
so
some
companies,
for
example,
I
know
this-
is
something
that.
B
B
For
you
know
the
the
last
weeks
average
you
know
account
for
some
amount
of
seasonality.
Maybe
it's
Friday
at
10:00.
It
understands
or
can
reference
the
statistics
from
the
week
prior
or
whatever
any
way
that
you
you
know,
II
don't
need
to
go
through
all
the
specifics
of
the
example,
I
guess
other
than
suffice
to
say
that
hey,
there's
a
novel
approach
to
some
things
that
people
are
solving
separately
and
this
approach
like
some
of
the
benefits
of
a
mesh
itself.
B
In
terms
of
being
you
like
uniform
and
ubiquitous
uniform
in
the
way
in
which
these
things
are
being
identified
and
potentially
handled,
yeah
pretty
interesting,
there's
a
lot
yeah,
even
just
in
so
much
as
the,
if
nothing,
if
the
mesh
itself
didn't
reconfigure
itself,
but
just
in
so
much
as
it
was
to
signal
or
identify
that
those
signals
are
there
and-
and
you
know,
webhook
something
else.
You
know
to
invert
and
bocoe
workflow
or
something
that.
F
The
simplest
thing
that
could
happen
is
that
you
would
get
reports
in
your
New
Relic
and
then
it
goes
to
your
graph
and
you
get
historical
data.
And
then
you
would
have
probably
some
sort
of
machine
learning
going
on
where
you
extract
that
historical
data
into
some
sort
of
phone
graph,
where
you
then
can
try
to
match-
and
this
could
especially
be
useful
when
you're
doing
DDoS
mitigation
as
well,
because
then
you
get
to
compare
it
to
historical
patterns
and
see
what
works.
F
B
B
F
I
think
the
best
approach
to
this
would
be
an
API
based
solution,
so
that
you
can
allow
kind
of
separation
of
concerns
and
kind
of
an
inversion
of
control
in
place
where
the
other
party
can
develop
their
own
interface
on
how
they
can
use
the
service
rather
than
you
having
to
hook
into
their
service.
Because
then
it
would
a
lot
of
dependencies
that
don't
need
to
be
there.
Yep.
B
And
then
yeah,
then
you
kind
of
get
in
the
way
of
business,
specific
logic
or
environment,
specific
logic
that
people
are
gonna
want
to
do.
Based
off
that
info
that
yeah,
you
sort
of
shortchange
you
yeah.
That
makes
a
lot
of
sense.
Yeah,
like
in
some
respects,
you
could
imagine
macine
I'm,
not
sure
if
you've
gotten
to
run
Mary
just
yet,
but
even
even,
if
not
like,
you
could
imagine
that
within
there
it's
because
it
is
deployed.
Maybe
it's
deployed
an
additional.
B
You
know
it
is
a
sorry
deployed
some
additional
intelligence
to
the
data
plane
and
is
watching
for
some
of
these
signals
or
has
the
right.
It
is
performing
that
machine
learning
on
the
fly
that
that
Mary
is
just
exposing
exposing
those
signals
or
providing
a
web
hook
to
subscribe
to
or
and
letting
people
adjust
the
the
sensitivity
in
the
various
settings
of
that
our
algorithm,
that's
running
or
just
the
static
threshold.
B
F
B
You
know
Thanks
next
one
was
a
topic
around
some
recent
work.
That's
gone
on
into
the
mesh
readapt
er
for
the
citrix
service
mash,
the
maintainer
of
that
Dheeraj
had
fixed
an
issue
that
have
been
kind
of
long
standing,
and
so,
if
next
time,
you're
running
ma
cherie
go
ahead
and
if
you
would
give
an
attempt
to
provisioning
the
citrix
service.
Mash
have
mastery
provisioning
right
now
that,
at
my
sri
adapter
is
in
a
beta
phase
and
we'd
like
to
move
it
to
a
stable
and
a
couple
of
bugs
opened
recently.
A
Sorry,
Lee
I
just
have
liked
it
one
question
because
I
will
be
again
testing
all
the
service
meshes
very
soon
for
before
I
go
to
Camp
cloud
native.
Actually
so
next
week
expect
some
calls
from
from
me.
But
do
we
have
like
a
statist
page
if
the
adapters
are
working
with
misery,
because
remember
every
time
that
I
show
you
like
link
Rd
and
steel?
Of
course
they
work
great
and
then,
when
I
go
a
little
bit
on
the
others
console.
But
you
saw
that's
octarine
also
at
some
point
in
time.
A
B
B
Not
to
the
degree
no
I
mean
partially,
but
not
in
the
most
helpful
way,
is
not
in
the
way
in
which
is
really
what
you're
looking
for,
which
actually
calls
to
some
of
the
things
that
we're
trying
to
get
done
inside
of
the
continuous
integration
working
group
where,
whereas
builds
are
done
and
releases,
are
made
of
each
adapter
that
there
need
to
be
integration,
tests
that
will
come
will
validate
the
successful
provisioning
of
a
mesh
validate
successful
provisioning
of
the
that
adapters
sample
apps
and
we
are
there's
work.
That's
been
done.
B
We've
used
kind
a
little
bit
inside
of
github
actions
too,
and
so
I
think
the
the
one
that's
farthest.
So
the
adapter
that's
furthest
along
in
that
regard
is
the
linker
d--
adapter.
It
was
so
randomly
chosen
as
the
one
that
we
were
gonna
go.
Do
that.
Let
me
take
a
note
here
and
push
actually
since
you're
on
the
call,
and
if
you
could,
you
need
to
take
a
note
about
this
in
Monday's.
A
And
just
for
us
who
are
consuming
more
on
the
site,
yeah
just
see
like
okay,
this
adapter
is
working
this.
This
is
working
and
maybe
just
what
what
is
the
intro
so
that
you
are
using,
of
course,
being
on
SS
l2,
all
the
time
I
get.
I
get
it's
special
for
from
time
to
time.
That's
for
sure,
but
yeah.
Thank
you.
C
B
D
So,
basically,
what
we
have
done,
liquor
yeah
so
Elektra
first,
given
update
on
SMI
conformance
tool
first,
so
there
has
been
a
significant
amount
of
work
like
since
the
last
call.
So
what
we
have
done
is
we
have
made
like
Ganesha
has
made
his
own
fork
of
kurtal
and
there
he
has
edited
like.
Basically,
we
have
hooked
into
cuddle
and
we
have
leveraged
some
of
their
like.
We
have
leverage
some
of
the
functionality
and
made
it
according
to
our
test
cases.
D
D
We
can
provide
some
function
like
a
separate
function
for
each
test
case,
which
we
can
then
so
in
that
way,
what
we
are
thinking
is
like
just
keeping
track
which
function
has
been
completed
and
then
maybe
using
my
series
load
generator
or
fort
I
like
we
have
a
built-in
support
for
that
to
generate
some
load
and
get
the
IP.
We
do
use
to
client
here
in
Missouri,
so
we
can
get
the
IP
of
the,
and
we
can
do
lot
of
things
with
that.
D
We
can
do
similar
things,
so
what
we
can
do
is
we
can
leverage
the
we
can
leverage
our
own
Fork
of
cartels
and
we
can
say:
okay,
you
deploy
the
business
using
my
series,
install
installation
command
in
a
cool
in
a
in
a
like,
maybe
in
Cuba
netizen
docker,
and
then
maybe
apply
this
like.
We
can
specific.
D
We
can
use
petal
and
say:
okay,
deploy
this
app
and
then
mesially
will
load,
generate
and
perform
the
load
generation
and
see
you,
okay,
whether
other
things
are
working
or
not,
and
for
every
test
case
that
that
has
passed.
We
can
eventually
say:
okay,
these
many
tests
is
these.
Many
test
cases
have
passed
and,
like
the
build
works,
fine
we
can
leverage
that
actually
like.
Since
it
is
mostly
declarative
testing,
it
would
just
be
some
Yambol
files
to
deploy
them
and
we
can
use
our
own
sample
app
to
deploy
like
for
each
business.
B
B
The
the
context,
by
the
way,
I
think
half
of
you
have
the
context
and
half
don't,
and
that's
for
those
that
don't
very
briefly.
Smi
is
a
relatively
new.
Well,
it's
a
project.
That's
been
around
for
a
year.
It
is
if
I
can
bastardize
this
analogy.
It
is
what
terraform
is
to
multi-cloud.
Smi
is
to
service
meshes,
or
rather
it
is
what
CNI
is
for
container
networking
that
that's
a
it's
an
abstract
specification
to
have
a
single
set
of
API
is
to
interface
with
any
of
the
service
messages.
B
That's
in
part,
that's
part
and
parcel
to
a
portion
of
measure.
E's
goal
and
messy
has
been,
was
a
launch
partner
with
SMI
a
year
and
something
ago,
and
in
that
time,
Naveen
and
Shiva
and
Kanishka,
and
some
others
in
the
community
have
been
working
a
bit
toward
helping
a
bit
toward
making
measure
EA
validation
tool
for
SMI
to
validate
whether
or
not
each
of
the
service
mesh.
Specific
implementations
are
adhering
to
SM
eyes,
spec
and
part
of
that
validation.
Work,
making
measure
e
a
conformance
tool
for
that
specification
means
that
measure
e
we're
automating.
B
B
Their
comments
on
the
update
that
Naveen
gave
there
are
two
interesting
things
that
I'll
highlight,
and
one
is
just
you
know
whether
or
not
we
we
had
hosted
a
an
online
Meetup
I
organized
the
cloud
native
Austin,
the
the
scene,
CF
meetup
here
in
Austin
and
we'd
had
a
couple
of
folks
from
d2
IQ.
They
presented
cuddle
while
they
presented
Kudo
and
part
of
that
was
presenting
cuddle
anybody
using
cuddle
or
familiar
with
it.
Beyond
what
Naveen
just
said,.
B
In
some
respects,
cuddles
a
little
bit,
you
know
it
has
some
utility
a
little
bit
like
Sona
boy.
Does
there's
a
bit
of
a
relationship
there
if
you're
familiar
with
Sona
boy.
The
other
thing
I
thought
that
was
really
cool,
pretty
interesting.
That
Naveen
was
just
highlighting
is
the
notion
that
it
was
well
a
couple
of
weeks
ago
at
docker
con
when
Kanishka
presented
some
webassembly
work
that
the
community
has
done.
He
did
so
using
a
new
sample
app.
The
new
sample
app
is
called
image
hub.
B
It's
basically
a
you
know
a
ripoff
of
docker
hub
and
it's
it
doesn't.
It
doesn't
work
in
that
regard.
It's
just
a
sample
app
for
purposes
of
demonstrating
web
assembly
and
and
some
things
interesting
things
you
can
do
with
love
assembly
and
service
meshes
and-
and
it
was
great
to
get
a
new
sample
app
and
of
what
novena
just
said
that
more
or
less
and
means
essentially
announcing
the
fact
that
there's
another
sample
app
in
flight
right
now
and
that
it
was
just
created.
B
There's
a
lot
of
potential
use
cases
for
it,
both
as
a
learning
device
to
teach
people
about,
but
basically
a
better
app,
maybe
a
lighter
weight,
more
interesting
app
than
book
info
or
emoji
voto
or
the
other
very
popular
people.
Apps.
That
service
messages
have
is
to
come
forthwith.
Well,
hopefully,
a
better
one
but
part
of
coming
forth
with
that
a
universal
one,
and
maybe
a
bit
better
part
of
coming
forth
with
that-
is
that
that
this
sample
Apple
will
be
used
in
context
of
the
conformance
testing
I.
A
Like
that,
then,
just
again,
sorry
being
so
talkative
today,
can
the
sample
app
be
available
for
all
the
service
missions.
Then
remember
at
very
first,
we
tried
to
have
like
a
comparison
between
the
service
meshes
somehow,
and
the
problem
is
like
book
info
was
maybe
a
link
or
D
and
on
steel,
but
not
on
the
others.
So
if
the
sample
app
can
be
like
available
from
all,
then
we
have
like
very
nice,
let's
say
at
least
comparative
app
across
all
the
service
meshes.
D
Yeah
sorry
certainly
go
ahead,
no
go
ahead,
yeah!
So
actually,
oh!
Yes,
this
sample
app
is
not
click
integrated
with
any
seriousness
and
is
independent.
So
in
that
way
it
can
be
like
integrated
with
every
service
mesh,
because
every
solutions
for
like
well
for
most
of
them
they
either
require
a
liberal
level
or
an
annotation.
So
what
we
are
doing
in
SMI
conformance
is
we
will
take
like
an
input
from
the
user.
D
Ok,
if,
if
the
user
is
not
executing
like
predefined
pre
supported
service
message,
so
if
the
there
are
lists
of
the
business
that
we
support
by
default.
So
if
the
user
is
testing
a
new
service
mesh,
so
what
they
can
do
is
they
can
say.
Ok,
these
are
the
labels
and
annotation
that
I
require
the
container
or
the
pods
to
have,
and
then
my
service
mesh
will
take
care
the
rest
of
the
deployment
process.
D
E
D
D
A
E
D
D
B
So
it's
a
wonderful
suggestion.
It's!
It
was
actually
an
action
item
for
us
to
determine
that
Iran
that
it
was
gonna,
be
compact
and
so
a
great
suggestion.
I'd
thanks
for
following
up
on
a
tuna
being.