►
From YouTube: 2020-07-24 meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
A
A
B
C
D
D
D
Just
wait
another
minute
I
was
hoping
some
people
from
microsoft.
They're.
Sorry,
not
microsoft
from
aws
would
be
showing
up
since
elita
from
aws.
D
Was
going
to
be
championing
this
specification
working
group.
E
I
am
the
one
for
aws
today
ted
my
peers
sandra
who
runs
our
efforts
on
sdk
and
developer
experience.
She
is
out
of
office
today,
so
I
I
know
she
committed
some
action
items,
I'm
not
quite
sure
if
she
has
completed
yet,
but
but
I
can
cover
part
of
our
agenda
on
behalf
of
aws
here
today.
B
D
Likewise,
I
believe
I
was
not
here
last
week
I
was
out,
but
when
I
last
left
our
heroes,
I
believe
the
goal
was
the
different
groups
who
were
attending.
This
meeting
were
going
to
review
the
current
sampling
sdk
api
and
try
to
get
a
sense
of
whether
the
way
samplers
are
currently
set
up.
What
capabilities
they
currently
have
at
open
telemetry?
D
So
I'm
curious
is
there
for
the
people
who
are
on
the
call
today.
Is
there
anyone
coming
back
with
an
assessment
like.
D
F
D
F
Just
let
me
know
from
microsoft
yeah
so
from
our
we
are.
We
have
some
needs
and
sampling,
and
I
I
didn't
know
that
what
you
guys
are
doing-
and
I'm
here
today
to
learn,
but
from
our
side
I
can
come
up
with
such
an
assessment
by
the
next
week.
D
Perfect,
thank
you
so
much
yeah,
that's
the
the
main
thing
we
want
to
get
right
for
ga.
Just
for
clarity,
for
other
people
who
might
be
new
is
to
just
longer
term.
It
would
be
great
to
standardize
some
sampling
concepts
like
sampling
priority
and
things
like
that.
D
But
for
ga,
the
primary
thing
we
want
to
make
sure
we
get
right
is
that
the
sampler
api
in
the
sampling
facilities
within
the
sdk
are
sufficient
for
people
to
at
least
port
their
current
samplers
over
to
open
telemetry
and
have
their
current
back
ends,
work
with
open
telemetry
without
needing
any
kind
of
standardization.
D
Hi
doing
well,
I
was
just
giving
a
quick
review
of
what
the
goals
of
this
sampling
working
group
were,
and
since
we
do
have
some
new
people,
I
wonder
if
we
could
just
maybe
just
do
a
round
of
introductions
just
why
people
are
here
and
what
they're
hoping
to
get
out
of
the
sampling
working
group.
D
H
Sure
kevin
brock
off
I'm
an
it
services
consultant,
so
I'm
the
customer
and
I'm
doing
a
bunch
of
work
right
now
on
the
collector,
getting
it
all
the
tracing
stuff
bugs
fixed
and
improving
usability
and
stuff.
So
I
was
really
interested
in
catching
up
on
what's
going
on
in
the
sampling
thing,
because
we're
trying
to
implement
tail
based
sampling
in
there
so
just
want
to
catch
up
with
all
that.
D
Okay,
elizabeth
heinlein
nice
to
meet
you.
C
D
Great
to
meet
you
elizabeth
and
then
we've
got
davey,
who
I've
now
gathered
from
aws.
E
Yeah
hi
david
mendez
is,
I
lead
the
engineering
for
the
x-ray
back-end.
We
have
some
back-end
implementation
for
sampling
rules
for
vending
sampling
rules.
We
have
some
sampling,
coders
and-
and
I
led
particularly
that
implementation-
I'm
representing
aws
in
that
sense,
nice
to.
I
How
are
you
doing
jonah,
cowell
cto
at
logs
io
and
I'm
mostly
listening
and
have
past
experience
as
someone
that
spent
a
lot
of
time
at
appdynamics
and
and
it
worked
in
various
commercial
apm
tools
that
have
sophisticated
sampling,
so
just
trying
to
make
sure
that
some
of
that's
addressed
here
thanks
all
right.
G
Yeah
hi
hi
everyone,
I'm
alolita
sharma,
from
aws,
I'm
working
specifically,
and
my
team
is
working
on
the
metrics
implementations
for
open
telemetry,
specifically
improving
and
adding
metric
support
for
the
c-plus
plus
sdk,
as
well
as
building
out
a
series
of
exporters
that
are
useful
for
prometheus
for
cortex
and
for
you
know,
validating
the
collector
and
and
sdk
pipelines
and-
and
we
have
been
involved
with
the
project
for
about
three
months
now-
really
super
excited
that
the
face
that
the
project
is
moving
at
and
continue
hoping
to
get
more
involved
in
the
sampling
sake.
D
D
Now
we've
got
lumira
from
microsoft,.
F
Yeah
hi,
I'm
ludmila
I've
been
involved
into
distributed
tracing
comfort
for
quite
a
while.
At
this
point
I
was
off
for
a
few
months,
so
now
I'm
back
from
microsoft
sites.
We
are
really
interested
in
sampling.
The
immediate
need
is
head
based
sampling,
but
we
need
to
open
telemetry
algorithm
to
interrupt
with
our
previous
algorithm
and
we
want
like
the
graceful
migration
to
open,
telemetry
algorithm.
So
we
are
mostly
interested
in
consistent
sampling
across
clouds
and
everything
yeah.
So
I
I
will
be
driving.
F
I
will
try
to
drive
the
sampling
priority,
call
request
further
and
in
general
we
are
interested
in
everything
else.
The
tail
based
sampling
is
obviously
something
everyone
is
interested
in,
of
course,
yeah.
D
Okay,
let
me
get
josh
josh
here
from
lightstep.
A
Hi
yeah-
I
I'm
here
not
to
represent
lightstep
as
much
as
to
keep
my
ear
on
the
conversation,
particularly
looking
at
metric
sampling
these
days,
and
I
think,
there's
my
interest
in
metric
sampling
is
that
I
believe
we
can
approach
high
cardinality
metrics
by
supporting
sampling
for
metric
events,
and
that's
why
I'm
here
so
I
should
just
be
listening.
I
don't
have
a
lot
to
say
about
chasing
well,
I
do,
but
I
won't
how's
that
come
on
all.
D
Right
and
we
got
yuri
as
well.
J
Jagermeister
hi,
I'm
sorry,
I
might
have
to
drop
any
time
because
I'm
waiting
for
a
call
but
yeah.
I
mostly
want
to
make
sure
that
jaeger
capabilities
are
still
possible
in
open
telemetry,
something
yeah.
D
And
then
shenzhen
lee.
B
D
Okay,
okay,
so
maybe
kicking
this
meeting
off
then,
and
it's
nice
to
meet
you
all
and
I'm
glad
to
see
new
faces
here
this
week.
The
the
homework
from
before
which
I
would
then
ask
also
of
of
microsoft
and
anyone
who's
got.
An
existing
sampling
regime
has
been
to
make
sure
that
you
could
get
your
existing
sampling
algorithms
working
with
the
current
sdk.
D
So
I'm
curious
who's
come
back
this
week
with
a
report
on
that
front
that
be
you
davey
davi.
Sorry.
E
Yeah
no,
like
we,
we
had
sandra
mcmullen
that
she
is
out
of
office
this
week
and
chang.
Do
you
know
whether
you
guys
have
completed
the
action
item
that
we
took
two
weeks
ago.
B
No
we're
still
working
on
that
yeah,
but
basically
we
we
didn't
find
any
inconsistency
anymore,
so
no
blocker
at
this
moment,
but
we
need
the
time
to
complete
the
doctor.
That's
positive.
H
Okay,
I
got
a
question
for
you
guys,
so
you
what
you're
talking
about
is
the
exact
same
stuff
that
you
could
do
with
the
aws
api
gateway.
As
far
as
configuring,
the
sampling
rules
in
case.
E
Yeah
so
so
kevin
we
spoke
about
that
with
ted
and
jonah,
and
others
like
on
the
first
meeting
yeah.
I
think
we
are
first
looking
like
what
is
the
basic
sampling
functionality
that,
in
general,
the
each
exporter
and
sampling
have
to
have,
and
we
will
have
extensions
to
support,
like
the
use
case
that
each
vendor
provides.
In
that
case,
what
we
call
the
the
managing
sampling
rules
through
the
service
or
centralized
sampling
rules
will
be
an
extension
that
we
will
develop
further
later.
D
D
I
was
wondering
if
someone
from
this
call
would
be
willing
to
take
that
issue
as
an
assignment,
so
that's
basically
reviewing
what's
currently
in
the
spec
doc
about
the
sampling,
sdk,
there's
a
first
step,
just
making
sure
it's
accurate
with
what
what's
currently
being
implemented
or,
if
there's
some
gap
there,
but
more
importantly
than
making
sure
that
if
we
do
need
to
change
that
sampling
it
api
in
this
spec,
you
know
taking
feedback
from
this
group
as
people
try
to
implement
samplers
and
updating
the
spec
to
match
what
we
end
up.
D
G
I
think
ted
this
is
alita,
we
are,
we
are
actually
an
intention.
Maybe
you
can
give
an
update
there
again
sandra
and
I
had
discussed
about
you
know
walking
through
the
spec
and
ensuring
that
it
actually
syncs
up
with
our
understanding
and-
and
you
know
our
requirements
right
so
will
we
do?
You
know
sanction
if
we
are
doing
that
right
now?
Have
we
started
that
process
at
all.
B
We
we
are
actually
more
like
focusing
on
the
compatible
between
open,
parametry
and
aws,
but
now
we're
looking
for
the
outdated
aspect
on
the
open
telemetry.
So.
G
D
Okay,
great
alolita,
could
I
assign
this
to
you
since
you're
running
that
group
that
be
okay.
G
Yes
sure
sure
I
mean
I'll
sync
up
with
sandra
and
and
make
sure
that
we
can
at
least
report
back
what
we
see
as
gaps
or
not.
D
Okay,
great,
thank
you
so
much.
That's
super
helpful.
How
are
things
going
at
microsoft
in
this
front.
F
Well,
I
it's
my
third
day
in
the
office.
I'm
sorry,
I
don't
welcome.
D
Back
yeah,
but
I
know
you
you,
you
mentioned
the
sampling
priority
issue,
so
you
were
gonna
pick
up
trying
to
get
that
standardization
in
there
as
well
great
awesome,
but
is
microsoft
currently
building
a
a
micros
like
an
azure,
specific
sampler?
I
assume
that's
something.
You're
gonna
need
right.
F
J
D
D
Okay,
great
welcome
back
likewise
elizabeth
for
new
relic,
I
assume
new
relic
already,
does
some
sampling
of
some
form
and
you're
going
to
need,
in
addition
to
a
new
relic
exporter,
probably
the
new
relic
sampler
of
some
kind
do
yeah
what
the
requirements
are
there.
C
I
unfortunately
don't
have
those
I'm
kind
of
a
last
minute
edition
and
I
didn't
get
do
all
my
homework
before
this,
but.
C
D
Well,
maybe
we
can
call
this
meeting
short
again
then,
because
it
seems
like
we've
got
newcomers
who
are
interested,
but
we
do
need
to
to
get
everyone.
Who's
gonna
need
to
do
some
kind
of
sampling.
D
D
Hopefully,
people
you
know
will
eventually
support
what
open
telemetry
has
out
of
the
box,
but
you
know
in
the
meantime,
people
have
to
ship
their
own
exporters,
probably
with
a
lot
of
environment,
variables
and
other
things,
and
they
may
want
to
ship
a
combination
of
exporter
plus
sampler,
plus,
possibly
something
else
we're
seeing
in
some
cases.
D
You
know
if
users
need
to
especially
install
more
than
one
plugin
to
make
a
system
work
correctly
with
a
back
end.
You
wouldn't
want
your
users
to
get
stuck
in
some
some
half
state
right
where
they've
installed
installed,
say
the
exporter,
but
they
haven't
installed
the
sampler,
and
so
something
crazy
is
going
on.
D
Yes,
yes,
this
is
this
is
the
thing
sampler
sampling
in
particular
is
pernicious
right.
That's
you're
only
going
to
notice
down
the
line
that
you've
not
had
the
correct
data
for
a
while
if
that
gets
messed
up
or
you
get
your
bill
back
and
you're
very
surprised.
D
D
We're
writing
we're
trying
to
use
open
telemetry
out
of
the
box,
so
we're
not
going
to
install
an
exporter,
but
but
we
do
have
some
like
configuration
parameters
that
we
need
to
make
sure
that
they
set
like
we
have
an
access
token,
for
example,
and
you
can
set
that
as
a
grpc
header
on
otlp,
but
that's
like
a
kind
of
weird
obscure
setting.
D
So
it
looks
kind
of
ugly
if
you
just
have
all
of
these
settings
out
there
in
a
pile,
so
we're
making
a
little
wrapper
called
hotel
launcher
that,
just
like
cleans
up
a
lot
of
that
configuration
so
that
it's
a
lot
simpler.
So
I
just
wanted
to
put
that
idea
in
people's
brains,
not
that
we
need
necessarily
some
one
official
distro
framework
or
something.
But
people
should
consider
looking
at
ways
they
could
package
up
open
telemetry
with
their
own
stuff
and
make
it
like
the
install
path
a
little
bit
cleaner.
F
F
Basically,
yes
from
our
site,
we
will
be
heavily
invested
in
doing
everything
in
when
you
looked
until
I
watched.
E
To
add
to
that,
like
it's,
it's
and
I
messaged
that
on
the
first
meeting
that
we
had
like
it's
important
that
we
close
days
back
for
for
sampling
super
soon,
because
one
thing
is
to
ship
the
samplers
and
and
and
the
implementation
exporters
for
customers
to
consume.
But
in
the
case
of
x-ray,
for
instance,
most
of
the
what
we
call
centralized
sampling
rules
functionalities
are
actually
embedded
on
the
managed
surface.
E
So
if,
if
they
spec
chains
a
little
bit
like
now,
you
have
to
propagate
that
change,
while
those
peer
teams
that
are
not
necessarily
involved
on
this
effort
and
as
we
all
know
like
this
crossover
facilitation,
doesn't
have
that
that
fast
right.
So
if
in
case
of
a
caving
for
instance,
that
was
bringing
the
case
of
api
gateway
like
we,
we
need
to
concentrate
that
effort
with
them
and
then
push
those
chains
globally
in
all
regions
that
take
a
long
cycle.
H
E
So,
for
me,
one
of
the
things
that
I'm
looking
for
from
here
is
in
and
before
I
used
to
let
x-ray
integrations
across
aws
as
well.
So
it's
it's
when
we
have
a
fine
award
on
days
back
to
go
and
talk
to
those
teams
to
make
sure
they
are
also
supporting
the
propagation
of
the
champion
decision
and
and
can
actually
respect
that
sampling
decision
on
at
the
head
that
the
that
is
performing
the
sampling
call.
E
D
E
I
mean
in
our
experience
like
based
on
a
telemetry
like
most
of
the
customers,
using
any
kind
of
a
vendor
specific
sampling.
In
our
case,
actually,
specific
sampling
rules
are
using
through
the
managed
services.
They
are
not
actually
doing
that
sampling
decision
on
their
application
they're
either
using
alb
to
create
the
trace
header
or
using
api
gateway
to
manage
the
sampling
rules
and
make
the
first
sampling
decision
and
propagate
downstream.
D
D
That's
a
little
bit
more
like
how
do
we
standardize
some
of
these
concepts
like
priority,
but
one
of
the
questions
with
the
sdk
right
now
is:
can
the
sampler
actually
control
things
in
the
way
that
is
necessary,
or
is
there
some
piece
of
data
or
some
facility
or
hook
that
it
would
need
in
order
to
do
this
correctly?
One
thing
that's
been
proposed,
for
example,
is
being
able
to
change
the
sampling
decision
say
at
a
later
date.
D
D
And
then
a
parallel
track
that
maybe
ludmilla
is
interested
in
taking
on
is,
is
how
do
we
then
standardize
some
of
these
concepts
like
priority,
so
that
would
be.
I
think
the
follow-up
question
is,
if
you
wanted
to
ditch
your
custom
sampler,
what
would
open
telemetry
need
to
have
provided
within
the
sdk
as
a
more
standard
concept
in
order
to
allow
you
to
ditch
your
sampler
entirely,
so
those
are
the
two
big
questions
like
does
the
sampler
have
the
the
right
hooks?
D
Does
it
have
the
right
control
features
to
do
the
things
you
want
as
a
custom
sampler
today
and
to
what
what
you
know?
What
would
we
need
to
standardize?
Is
it
a
piece
of
information
like
what
what
would
open
telemetry
need
to
bake
into
the
sdk
so
that
you
could
drop
having
a
sampler,
and
so
I
think
I
want
to
to
put
those
down
as
homework.
D
For
people
and
then
we'll
re-meet
next
week,
friday
and
hopefully
hopefully
with
some
answers
and
then
we
can
get
some
issues
like
assigned
and
moving.
How
does
that
sound
to
everyone.
G
Sounds
good,
I
think
you
know
again
as
we
review
and
and
and
dig
through
the
spec.
I
agree
with
you
ted
that
you
know
again.
Implementation
is
one
of
you
know,
one
of
the
better
ways
of
verifying
what
what
works
and
what
doesn't
on
this
fact
today.
So
we'll,
definitely,
you
know
sink
back
up
in
terms
of
the
the
work
that
we
need
to
do
on
our
end,.