►
From YouTube: gRPC Community Meeting 5-24-18
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).
B
A
B
So
we
actually
sent
out
the
form
to
JVC
at
I/o
and
I
think
we're
going
to
have
it
in
the
dock.
The
link
to
the
forum
for
people
are
interested
to
sign
up
for
package
build
of
your
PC
core.
This
applies
basically
to
c-sharp
Python
Ruby,
PHP,
no
and
we're
gonna
have
that
soon.
So
this
is
a
forum
for
people
who
are
interested
in
having
early
access
to
that
repo.
Of
course,
this
is
an
addition
to
or
released,
and
this
process,
which
will
already
have
an
ask
for
a
while.
B
I
might
have
a
small
thing
so
high
like
that.
No
one's
ever
seen
you
before,
okay,
so
hi
I'm
young.
Oh,
so
we've
been
working
on
improving
the
infrastructure
for
GR
PC.
You
know
over
the
last
few
quarters
and
then
one
thing
that
you
might
have
realized,
but
you
also
might
not.
Is
that
the
way
the
test
results
are
presented
on
the
for
requests?
B
Oh,
like
all
the
testers,
are
not
public
and
you
have
access
to
like
the
structured
test
results
which
we
hope
is
going
to
improve
like
the
experience
for
the
contributors,
because
before
it
wasn't
easy
to,
you
know
access.
You
know
those
test
results
and
I
understand
what
they
mean
so
right
now,
when
you
create
a
pull
request
against
a
repo
like
you're
gonna,
see
all
those
publicly
and,
as
I
said,
there's
like
their
presence
in
a
way
that
so
hopefully
much
more
understandable
that
it
used
to
be
in
the
past.
D
D
D
D
Yeah
so
anyways
yeah,
planet-scale
observability
with
open
senses,
4g
RPC
and
go
java
and
python.
Who
am
I
just
introduced
myself
and
have
a
manual
key
a
decade
and
I'm,
an
avid
open
source
consumer
and
producer
I
work
in
the
go
programming
language,
as
well
as
open
senses
and
distributed
systems
at
a
rich
tech.
Inc
I
really
enjoy
using
gr
PC.
Why
open
census,
low
latency
overhead
provides
you
more
than
instrumentation
its
micro
services
friendly?
D
We
have
the
mechanism
called
context,
propagation,
which
I'm
sure
many
of
you
are
aware
of
that
allows
you
to
propagate
information
across
our
PC
boundaries.
Ok,
open
census
is
vendor
agnostic
clouded
agnostic.
I
mean
you
know
you
could
write
in
your
computer,
AWS
digitalocean,
Google
Cloud,
add
her
and
from
that
you
could
export
an
extract,
explore
traces
and
metrics
send
them
to
Prometheus
x-ray
stackdriver
signal
effects
in
stanag
Zipkin
jäger,
you
name
it.
We
have
a
really
simple
API
and
in
order
to
exports,
you
know
in
order
to
export
to
a
back-end.
D
All
you
need
to
do.
Is
you
implement
the
exporter
interface
register
in
your
in
your
main
program
and
that's
it?
The
rest
is
handled
for
you.
These
are
validated
validated
ideas
that
come
from
running
at
scale
at
Google
for
the
past
10
years,
I'm
internally,
they
have
a
system
called
Census
and
open
census.
Is
the
open
source
rewrite
right?
All
the
work
is
done
entirely
in
the
open,
and
we
do
that
github.com
/
census
in
instrumentation,
open
census,
is
implemented
in
go
C++,
Java,
Python,
Ruby,
PHP,
Arland
c-sharp.
D
You
know
other
language,
other
language
implementations
will
be
coming
soon.
You
know
you
don't
need
distributed
systems,
expertise,
so
I've
gone
through
these
at
startups,
where
you
don't
need
to
roll
a
product
out,
but
as
you
keep
building
products
that
run
at
scale,
you
encounter
all
sorts
of
crazy
problem.
So
with
open
census
you
don't
have
to
be
a
distributed
systems.
Expert.
D
Quick
primer,
tracing
tracing
is
a
mechanism
by
which
you
track
the
progress
and
behavior
of
a
remote
procedure.
Call
or
it
doesn't
even
it
I
mean
it
could
be.
Even
a
function
call
basically
track
any
behavior
through
a
system
monitoring.
That's
the
mechanism
by
which
you
collect
quantifiable
metrics.
All
of
your
system,
for
example,
server,
latency,
client,
latency,
number
of
file,
descriptors,
open
context,
propagation,
which
I
introduced
previously.
That's
the
mechanism
by
which
you
send
distinguishable
information
about
a
request,
of
course,
RPC
boundaries
tags.
D
Then
the
next
concept
is
tag.
Sorry
key
value
pairs
attached
to
metrics
to
help
you
uniquely
identify
whatever
it
measure
it
could
be.
You
know:
host
post
is
equal
to
some
IP
user
agent
you're
coming
from
a
programming
language
then,
but
tags
are
tags
are
used
for
metrics.
So
that's
that's
an
important
thing
to
note
and
then
for
traces.
We
have
annotations
so
annotations
help.
You
describe
the
state
of
a
trace
and
you,
you
know
the
state
of
swarm
your
system
with
the
ten
that
you
you
made
that
annotation
and.
D
D
We've
launched
a
new
app
that
allows
you
to
search
YouTube
that
allows
you
to
search
for
videos
in
YouTube
and
after
we've
launched
it
it's
running
at
scale.
We
need
to
figure
out
what
exactly
is
going
is
going
on
without
a
way
to,
without
a
way,
to
clearly
track
without
a
way
to
clearly
track
metrics
and
also
figure
out
how
trusted
tracers
propagate
you
most
of
the
times
it
flying
blindly
and
even
worse,
is
the
fact
that
any
interaction
between
microservices
becomes
really
complex.
D
B
D
B
D
Visit
that
in
here,
it
shows
a
trace
of
what
exactly
happened
so
to
give
a
bit
of
background.
This
is
this
is
an
app
that
I
instrumented
me.
D
D
Yeah
so
I
just
remitted
caddy
as
well
and
here's
a
code
for
it.
It
sends
it
also
uses
Prometheus
to
collect,
to
collect
sorry
it
exports
metrics
to
per
meters
as
well,
but
anyways
here
is
what
here
is
what
request
that
was
cached
looks
like
when
we
view
done
stackdriver
tracing,
so
we
could
see.
You
know
we
received
a
request
here.
It
went
through
paddy
HTTP,
the
HTTP
server
and
then
in
here
this
is
actually
where
we
started
processing
the
request.
D
So
here
you
know,
we
have
an
a
ssin
that
we
made,
we
say,
check
cash.
If
queries
present
right-
and
we
noted
that
this
is
a.
This
is
a
driver
that
will
be
using
its
manga,
we're
using
MongoDB
and
go,
and
we
noted
that
we
had
a
cache
hit
here
at
the
very
end,
but
this
is
how
its
propagated-
we
extracted
a
query.
Parsed
it
after
that,
we
searched
on
one
would
be
be
cache
to
find
out
what
this
has.
Its
core
had
been
seen
before.
D
D
When
taking
a
look
at
this
request,
we
note
that,
when
search
manga,
we
need
to
find
one
here
when
search
through,
we
didn't
find
anything,
and
then
we
we
made
an
RPC
made
a
D
RPC
call
we're
made
of
your
RPC
call
T's
back
in
the
perf
M
search,
and
in
here
you
could
see
in
search
it
a
process
trace.
Let
me
show
the
code
for
this.
B
D
D
D
We
know
that
surgit,
okay
in
here
we
had
a
cache
miss
and
we
had
to
actually
search
that
YouTube
API
made
a
remote
who
made
an
HTTP
call
to
YouTube's
API.
To
search
for
that
query,
and
in
search
for
that
video
and
in
here
you'll
notice,
our
user
agent
for
the
go
API
client
and
our
agent
here
was
open
sans.
Let's
go.
D
B
D
D
C
B
C
D
D
D
D
And
oh
I
haven't,
haven't
shown
you
what
it
looks
like
on
AWS,
so
you
know
this
is
a
cache
hit.
This
is
what
it
looks
like.
This
is
the
exact
same
data
that
we
saw
in
stock
driver
but
exported
to
AWS
x-ray
right-
and
this
is
what
I
was
telling
you
about-
gives
a
nice
for
you.
We
made
a
search,
went
to
KDE
went
down
to
the
actual,
the
actual
back-end
right.
We
had
a
front
end
server
good
proxy
to
the
backend,
and
then
you
know
the
extra
extracted
the
query
extracted.
D
And
show
you
what
that
shows
the
full
request
and
then
a
nice,
a
nice
thing
with
tracing
it
with
distributed
tracing
is
all
of
a
sudden.
You
get
a
map
of
how
services
are
propagating
of
our
request,
has
propagated
and
how
your
services
interact.
So
this
is
a
really
simple:
it's
a
really
simple
back-end.
It
just
has.
D
D
Yeah
even
looks
like
and
then
anyways
that
that's
so
far,
the
demo
I
wrote
in
in
the
docs
created
an
introduction.
It's
a
primer
to
using
G,
RPC,
sorry
open
census
with
G
RPC
I
made
the
Python
developer
and
this
for
Java
developers
and
for
go
developers
and
if
click
in
here
it's
just
basically
running
through
a
bunch
of
the
steps
that
that
I
provided.
These
are
quick
introduction
here.
D
Is
an
RPC
definition
for
a
service
that
capitalizes
content?
You
know
you
send
in
kanji,
capitalizes
it
and
sends
it
back
out.
Capitalized
shows
yeah
I
generate
this
is
using
go,
and
here
we
make
our
plain
server,
which
would
be
the
normal
case
before
you
learn
about
open
senses.
This
is
typically
what
your
server
will
look
like
38
lines,
and
then
you
know
we
have
a
go
client
as
well.
D
Equally
as
large-
and
this
is
you
know,
these
are
the
interactions
of
what
it
would
look
like,
so
you
send
in
food,
it
echoes
back
food,
capitalized,
say
what
that's
a
drill
open
for
business
said
you
is
the
problem
that
I
mentioned.
You
know
you
just
launched
a
new
service,
you
launched
a
new
service
and
all
of
a
sudden
gets
really
popular.
You
don't
know.
What's
breaking
things
are
taking
very
long.
B
D
B
D
D
Occasionally,
you'll
have
something:
that's
really
an
outlier.
So
if
you
clicked
in
that,
you
can
show
you
what's
going
on
exactly
and
I
believe
stackdriver
will
be
rolling
out
with
a
feature
where
you
could
link
a
trace
to
a
metric
tour
trace.
So
you
do
a
reverse
search.
So
in
here
you
could
was
it
we
could.
Click
on
that
point.
Ask
hey
viola,
could
I
find
traces
related
to
this,
because
in
go
we
associate
tracing
and
metrics
using
the
same
context.
So
that's
a
feature
that
closet
you
will
become
now.
D
Certainty
is
back
to
that
tutorial
said
this.
This
shows
this
is
the
diff
on
how
to
add,
to
add
open
census,
open
census,
your
instrumentation
key
ago
app
here
we
have
an
exporter
for
AWS
x-ray.
We
also
have
one
for
Canisius
and
stackdriver
in
there
remember
I
mentioned
this
is
how
we
started
trace.
This
is
how
you
start
a
trace.
This
is
how
you
end
it.
D
D
The
aggregation
here
is
by
rate,
so
this
is
at
this
time
the
service
is
really
popular.
We're
hitting
4000
requests
per
second,
and
then
these
are
the
stock
driver,
metric
s-sir
choices
that
we've
seen
before
and
also
Google
stock
driver,
and
that's
a
client
instrumented
to
give
you
pretty
much
the
scene,
you
see
pretty
much
the
same
metrics
and
choices
overall,
you
could
instrument
open
census,
starting
from
your
very
starting
from
the
request
until
until
your
back
end,
sir
starting
from
your
client
until
you
back
end
and
also
your
instrument.
D
Database
drivers
like
I
did
for
MongoDB
goes
to
instrument
your
server
itself.
Science,
fermented
caddy
and
all
this
work
is
open.
Source
I
also
made
a
demo
for
Java
gr
PC
users,
it's
its
equivalent
of
the
go
tutorial.
Please
check
it
out
and
also
I
made
up
one
for
Python
and
with
that
sorry,
for
all
the
rock
I'm
going
to
I
was
just
too
excited,
but
thank
you
very
much.
That's
that's.
All
I
have.
A
D
Joined
Microsoft
is
on
board
with
open
census
and
they
planted
rolling
out
a
c-sharp
implementation.
But
there
is
a
company,
forget
I,
believe
its
pivotal
related.
They
rolled
out
a
c-sharp
implementation
already,
but
they
haven't
given
us
and
guarantees
that
it
is
stable
and
if
they'll
be
releasing
each
open
census,
but
the
official
one
will
be
coming
pretty
soon.
D
A
B
A
Yeah,
if
anybody
didn't
see
the
coop
con,
China
and
North
America,
which
is
in
Seattle
this
year,
the
call
for
presentations
just
opened
up
this
week
for
both
of
those
so
I
definitely
check
those
out
I'm,
not
sure
how
I
don't
know.
If
we
thought
a
lot
of
folks
trying
to
go
to
the
one
in
China,
but
I
would
imagine
the
North
America
one
usually
gets
a
a
big
turnout
as
well.
A
A
C
Yeah,
it's
I
think
it's
a
quick
question,
because
you
might
know
that
there
is
an
open
issue
in
github
that
I
think
was
open
at
two
years
ago.
I
think
it's.
There
was
some
some
chats
there,
but
nothing
concrete
and
I
wanted
to
ask.
If
there
is
any
initiative
to
finally
set
up
when
I
think
you're
crying
for
well
I,
think
you're
super
for
jira
PC
that
allows
us
to
run
non-blocking
forums.
C
B
C
D
C
C
B
C
C
C
B
C
Because,
yes,
I,
think
the
you
manager
that
is
exported
by
the
library
or
having
also
the
note
G
s
leap
over
yo
manager,
I
think
the
implementation
for
I
think
a
yeah.
It's
not
it's
not
rocket
science,
it's
not
peanuts,
but
it's
something
that
that
can
be
done
so
will
be
a
good
good
and
with
advantage
for
the
Python,
a
synchronous
ecosystem.
So
this
is
something
that
this
me
and
I
was
also
talking.
C
B
B
C
Saying
that
it's
it's,
this
is
gonna,
be
something
important
for
you,
the
next
also,
you
have
to
say
that
for
us,
I
will
this
for
me
and
maybe
also
for
my
company.
It's
important.
We
also
are
taking
on
put
some
efforts
here.
So
I
don't
know,
maybe
in
the
next
weeks,
if
this
initiative
is
still
not
started,
maybe
I
can
present
something
related.
How
could
be
implemented
because
maybe
we
that.
C
Yeah
exactly
that's!
That
was
the
point,
because
I
remember
that,
one
year
ago,
when
I
was
visiting
the
same
issue,
I
didn't
realize
it
about
the
implementation
of
the
abstraction
of
the
human
eye.
Sure
maybe
two
years
ago,
wasn't
there
no
no
GS
and
then
I
saw
in
communism
for
GRN
and
my
mind's
my
mind
was
saying:
wow.
This
can
be
employment,
it
also
forcing
you
using
a
pure
pure
and
that
if
Brian
yeah.
B
A
A
A
Right,
thank
you
all
for
joining
us
say
I
will
get
the
recording,
at
least
at
the
very
least
of
the
demo.
We
got
that
it
looks
like
so
we'll
get
that
up
on
our
YouTube
channel
and
can
share
that
out
with
everyone
later
and
if
you
want
to
demand
mo
anything
or
share
anything
on
our
next
meeting.
Please
reach
out.
Thank
you
so
much.
We
will
chat
soon.