►
From YouTube: Service Mesh Introduction, Recording 9 Dec 2020
Description
CNCF, Cloud Native El Salvador, Tarun Pothulapati, Bouyant
B
B
B
B
So,
let's
get
started.
So
what
is
the?
What
is
the
need
for
a
service
mesh
in
any
environment
so
currently
with
them?
With
the
move
from
monolith
to
microservices?
There
is
an
inherent
complexity
associated
with
it
right
now.
Now
we
are
dividing
our
big
monolith
services
into
multiple
microservices,
like,
for
example,
in
the
in
the
exam
in
the
diagram.
Here
we
will
have
we'll
have
a
a
microservice
for
newsfeed,
a
microservice
for
so
so.
B
If
we
consider
a
social
networking
platform
we'll
have
in
microservice
for
news
feed
chat,
messaging
friend
requests
a
lot
of
things
right
now.
This
is
very
useful.
The
the
break,
the
change
from
monoliths
to
microsoft.
This
is
very
useful
because
it
it
allows
teams
to
to
proceed
independently
without
having
to
without
having
to
have
tight
dependencies
between
each
other.
Each
microservice
can
can
update
independently
can
scale
independently
and
things
like
that,
so
the
the
benefits
are
huge,
even
though
there
is
inherent
complexity.
B
Service
mesh
is
one
way
that
allows
you
to
manage
this
complexity.
The
networking
side
of
the
networking,
observability
and
security
side
of
microservices
in
an
efficient
manner,
so
we'll
see
how
that
is
done.
So,
first
we'll
see
the
problems
in
this
microsoft
in
in
this
microservices
world,
especially
these
microservices
now
can
be
written
in
any
languages
like
go
java
or
whatever.
B
C
B
How
does
one
service
recognize,
whereas,
where
another
services,
what
is
the
ip
or
dns,
essentially
because
in
this
microsoft,
world
things
are
very
dynamic
like
there
is?
There
is
some
microsoft.
There
is
one
instance
of
the
microsoft
is
going
down.
One
is
one
instance
of
the
microservices
I
mean
coming
up,
etc.
So
there
is,
there
is
a
disclaimer.
There
is
a
discovery
problem,
inherently
next,
obviously
secure
communication
in
a
monolith
there
is
no.
B
There
is
no
need
for
secure
communication,
because
it
is
just
one
binary
right,
one
single
binary,
where
you
call
one
one
function
to
another,
so
there
is
no
need
for
security
because
the
boundary
is
inside
itself,
but
now,
with
with
with
network
now
with
every
now
with
network
calls
everywhere,
there
is
an
inherent
need
for
secure
security
right
security
in
transit.
Essentially
now,
whenever
a
service
is
calling
another
service,
you
need
you
need
it
to
be
you
you,
you
don't
want
the
information
to
leak
to
someone
to
or
anything
like
that
right.
B
B
B
Traffic,
shaping
you
can
have
canary
deployments
right,
like
you,
can
a
b
testing
like
when
the
you
can
have
multiple
versions
of
a
micro
services
and
talk
and
send
five
percent
of
the
request
to
a
new
version,
and
things
like
that
to
check
before
you
deploy
it
fully
authorization
right
like
you
want
to
authorize
which
service
users
with
service-
and
things
like
that.
So
this
is
a
photo
from
from
a
blog
post
by
monzo.
Monzo
is
like
an
international
banking
plan
like
it's,
it's
a
european
banking
service.
I
I
think
it's.
B
This
is
a
real
diagram
of
their
microservices
deployment.
They
are
1500
micro
services
each
talking
to
like
hundreds
of
different
others.
Each
each
dot
here
is
a
single
micro
service,
and
you
can
see
how
complex
the
whole
architecture
is
and
how
complex
it
could
be
for
for
systems
at
high
scale,
like
you
can
understand,
like
monza,
handles
a
pretty
high
scale
traffic,
especially
considering
that
it's
a
banking
platform,
the
same
problem
is
with
uber
too
right.
This
is
a
photo
from
a
talk
from
uber.
B
As
you
can
see,
they
have
like
thousands
of
micro
services
talking
to
hundreds
of
different
other
micro
services.
Each
one
talking
to
there
is
a
real
problem
here
right
to
handle
them
consistently
to
have
a
way
to
observe
them,
consistently
reliability
and
a
lot
of
things.
So
this
is
where
a
service
mesh
really
shines
essentially
like
having.
B
If
you
have,
if
you
have
like
two
three
micro
services,
you
may
not
need
a
source
on
the
whole
secure
communication
things,
but
if
you,
if
you,
if
you
correct
for
a
deployment
like
this
it,
it
becomes
very,
very
hard
without
without
anything
like
service
mesh.
So
we'll
see
what
is
those
machines?
A
service
is
essentially
a
two
components
like
on
the
higher
level,
the
control
plane
and
the
and
the
data
plane.
B
The
control
plane
is
a
set
of
components
that
that
are
useful
for
user
interaction
and
they
talk
to
the
data
plane
and
configure
them
in
some
way
as
the
user
wants.
So,
for
example,
if
the
user
wants
a
specific
configuration,
it
will
talk
to
the
control
plane
through
the
command
line
or
through
the
api.
The
control
plane
makes
sure
that
that
configuration
is
spread
across
the
data
plane
or
does
it
data
plane
involved.
A
data
plane
involves
a
set
of
proxies
like,
for
example.
B
Here
we
have
three
microservices
and
in
each
microservice
we
have,
we
have
the
proxy.
The
proxy
is,
is
a
is
a
sidecar
container,
essentially
in
the
kubernetes
world,
the
sidecar
container,
every
request
that
goes
into
the
that
one
that
goes
to
the
microservice
will
have
to
go
through
the
proxy.
Essentially,
the
proxy
is
like
a
it's
like
a
security
gate.
It's
like
a
gate
to
your
house
like
in
a
in
a
layman
terms.
Whenever
you
want
to
you
want
to
you,
want
to
get
inside
the
house.
B
You
want
to
go
through
the
security
and
then,
after
after
whenever
you
want
to,
you,
want
to
come
out
of
those.
You
want
to
go
through
the
security
gate
right.
It's
the
same.
The
proxy
acts
like
it.
Whenever
a
service
he
wants
to
talk
to
a
service
say
it
first
has
to
talk
to
its
proxy
and
the
proxy.
B
Then
talks
to
the
proxy
on
the
other
side
and
the
proxy
then
sends
the
request
to
services.
So
there
is
essentially
an
another
layer
of
the
proxy
layer
which,
through
which
the
communication
happens,
the
control
plane
makes
sure
that
the
configuration
is
sent
to
all
the
proxies
like
that.
So
what
do
you
expect
from
a
service
mission?
We
saw
the
problems
in
the
polyglot
services
world.
What
do
you?
What
does
the
service
may
solve?
So
there
are
like
multiple
themes
in
which
service
helps
the
security
you
get
mtls.
B
You
get
mutual
tls
like
here
here
as
in
the
example
here,
whenever
service
c
doesn't
doesn't
have
to
encrypt
its
connection.
It
just
talks
to
it
just
tries
to
send
the
request
in
a
normal
manner,
without
having
to
do
encryption
or
an
encryption
or
anything.
The
proxy
here
updates
the
request
and
encrypts
it
and
then
sends
the
request
to
other
proxy,
which
then
decrypts
it
and
sends
to
its
application.
So
essentially
the
proxy
knows
about
everything:
the
security
observability
side
of
things.
B
You
don't
have
to
configure
your
application
specifically
to
do
mtls
to
do
observability
and
things
like
that.
So
we
get
security
out
of
the
box,
mtls
cert
management,
rotation
and
authorization.
You
get
visibility
so
for
each
service,
you
essentially
have
some
some
golden
metrics
right,
like
like
the
latency
of
a
service,
so,
for
example,
if
your
newsfeed
is
loading
slowly,
you
want
to.
B
B
B
B
Sure
yeah
it's
a
pretty
active
project.
It
is
very
widely
used
in
production
at
banks
and
a
lot
of
other
things
in
a
lot
of
other
places,
we
have
a
slack
channel,
which
is
pretty,
which
is
pretty
pretty
active,
like
there
are
a
lot
of
questions
back
and
forth
between
maintainers
and
users.
So
it's
pretty
awesome
that
we
have
that
connection
with
users,
it's
pretty
famous
on
github
and
we
have
weekly
edge
releases
and
we
do
a
stable
release
every
six
to
eight
week
cadence.
B
So
what
does
liquidity
do?
The
problems
that
I
mentioned
previously
essentially
solves
the
observability.
It
gives
you
observability
out
of
the
box,
you
don't
have
to
you,
don't
have
to
configure
your
application
to
send
some
metrics
or
anything
like
that.
You
can
get
all
this
all
the
golden
metrics
right
away.
You
also
get
reliability
because
essentially
retries
timeouts
load,
balancing
circuit,
breaking
etcetera
security.
Obviously
you
get
transparent,
mtls
and
everything
with
all
of
these
features
without
without
any
additional
overhead.
That
is.
B
The
main
selling
point
is
what
I
see
you
get
all
these
features
without
having
to
do
a
lot
of
maintenance
work
for
the
service
mesh
itself.
So
this
is
the
architecture.
As
we
said,
the
control
plane
involves
some
components
and
the
data
plane
is
essentially
the
your
applications
injected
with
our
proxy.
Essentially,
here
we
have
multiple
controller
components
like
public
api,
tab,
destination,
id,
etc,
and
then
the
data
planes
are
side
of
things.
We
have
your
application
and
the
proxy
here.
B
B
B
B
So
what
is
service
discovery
and
road
balancing
using
service
discovery,
essentially
map
service
name
to
ip
addresses
like,
for
example,
if
a
service
say
wants
to
talk
to
service
b,
it
is
just
ask
the
proxy.
I
want
to
talk
to
service
b,
it's
the
proxy's
job
to
to
map
the
service
names
into
concrete
ip
addresses
that
are
currently
alive.
The
ip
addresses
of
the
essentially
the
the
instances
of
the
service
b.
It's
the
proxy's
job
to
to
to
map
the
service
name
into
hypothesis
and
also
to
load
balance
them
across
these
instances.
B
A
B
B
I
think
you
get
a
lot
of
visibility
from
from
a
service
mesh
you
get
a
graph,
etc.
I
think
I
spoke
a
lot.
Let's
see
a
demo
after
this
demo,
we'll
see
the
other
side
of
things
like
like
open
source
and
how
to
get
involved
side
of
things.
Let
me
show
a
demo.
I
think
I
think
I
have
a
better
I'm
more
of
a
demo
person
than
here.
I
have
a
kubernetes
cluster.
B
If
I,
if
I
do
cubicle
get
ports,
you
should
be
able
to
see
that
you
should
be
able
to
see
some
parts
running
the
system
pods
without
any
without
any
other
bots.
As
you
can
see,
it
is
just
a
plain
plain,
kubernetes
cluster,
with
no
with
with
nothing
installed.
Let
us
install
linker
d
to
install
linkedin.
You
have
a
you,
have
a
script
that
we
use,
call
and
install.
B
If
you
do
and
pass
it
to
sh
you
get
a
linkedin
binary
install.
Essentially
once
you
once
you
do
that
command.
You
have
you'll,
you
can
use
link
already.
The
linkedin
cli
is
installed
it.
You
did
not
touch
your
kubernetes
cluster,
yet
once
you
do
that,
you
can
do
linkedin,
install
and
and
then
pipe
the
output
to.
B
Go
to
cubicle,
so
essentially
we
are
installing
a
lot
of
components
now
for
off
lingerie
the
control
plane
side
of
things,
as
I
mentioned
now,
if
we
do
a
cubicle
get
parts,
you
can
see
a
lot
of
lot
more
components
here
we
see
we
see
a
lot
of
control,
plane,
compress
being
installed
the
controller
destination.
You
also.
B
B
So
let's
wait
for
them
to
initialize.
They
are
getting
initialized
so
so
before
we
do.
That,
we'll
also
see
one
more
thing.
This
just
like
we
saw
in
the
slides
actually
just
like
we
saw
in
the
slides
each
application
should
have
the
proxy
right.
Essentially
I
mentioned
that.
So
how
do
you
do
that
here?
If
you
see,
if
you
see
for
linkedin,
there
are
essentially
zero
by
two.
What
is
the
two
here,
for
example,
here
we
have
all
one
right.
B
Essentially,
there
is
a
single
component
for
everything,
but
for
the
linguistic
components,
even
for
the
control
plane,
we
inject
our
proxy.
So
just
like
your
applications,
even
the
control
pin
has
the
proxy
injected.
So
here
the
two
in
the
two
shows
that
one
is
the
application
container
itself
so,
for
example,
webex
center,
the
other
is
the
proxy.
We
can
see
that
actually.
So
if
I
do
linker
describe
archibat
in
the
linkedin
namespace
describe,
deploy,
meant
and
we'll
see
the
linkedin
web
deployment.
B
We
can
actually
see
two
containers
here
now.
The
first
container
is
the
web
container.
This
is
here
it
is
the
dashboard,
but
it
can
be
your
application
and
then
we
see
another
container
called
the
proxy.
This
is
the
linkedin
proxy.
That
is
like
a
gate
into
your
application
into
your
application
to
for
communication
into
and
outside
outside
of
your
application.
B
So,
as
I
mentioned
like,
there
are
two
things:
the
web
container
and
the
the
web
container
and
the
proxy
container,
essentially
so
they're
running
right
now
we
also
have
a
linkedin
click
command.
That's
made
that
that
is
run
some
checks
across
on
the
cluster
and
make
sure
the
counter
plane
and
checks
for
all
the
configuration
in
the
control
plane
to
make
sure
that
it
is
ready
and
works.
It
works
and
is.
B
B
B
B
But
it
is
a
sample
application
here
we
have
three
applications
and
we'll
see
and
we'll
see
how
that
works.
Here
we
have
three
applications,
etc.
So
we'll
apply
that
to
our
cubic
tl,
essentially
so
before.
Applying
that
all
you
have
to
do
is
that
you
have
to
pipe
that
into
linkered
inject
actually
finger.
The
inject
is
an
other
command
thinker.
Indirect
is
another
command
that
takes
you
in,
takes
your
yaml
manifest
of
your
application
and
just
injects
another
annotation.
B
Now,
if
you
do
linker
to
inject,
we
can
see
it's
the
same
ml
manifest,
but
with
an
extra
annotation
called
linkedin,
your
slash
inject
enabled
yeah.
Essentially,
your
application
is
essentially
getting
another
annotation
which
which,
which
tells
the
control
plane
to
inject
the
proxy
into
it.
So
essentially,
your
application
does
not
have
a
proxy
here.
It
has
a
single
container,
but
because
of
this
annotation
being
there,
the
controller
injects
another
container
called
the
linkedin
proxy.
B
The
control
brain
will
do
that.
So
now,
if
we
pipe
this
into
into
cubic
fl,
we
essentially
we
essentially
get
two
containers
instead
of
one,
as
you
can
see
here,
we
have
two
again
it's
if
we
see
if
we
see
it.
If
we
do
a
describe
on
the
pod
on
the
application
part
describe
part,
let's
do
a
describe
on
the
mojo
auto
part.
B
B
Application
is
like
I'm
port
forwarding,
so
the
application
is
so
this
amplica
application
is
essentially
for
four
things.
Actually,
there
is
a
there
is
a
web
dashboard.
There
is
a
voting
and
emoji
auto
mic,
so
this
is
a
microservice
application.
Example
right,
as
I
mentioned,
there
are
multiple
services.
There
is
vote
bot,
which
is
which
is
like
a
simulation
of
a
user.
It
keeps
voting.
B
It
keeps
voting
just
to
simulate
the
whole
network
right
to
see
how
things
are
happening
it
it
talks
to
the
web
web
application,
which
then
talks
to
the
voting
and
emoji
order
to
vote
to
emojis
and
also
store
the
state.
Now,
if
we
see
the
application,
if
we
open
the
application,
localhost
3000,
you
can
see
a
application
here.
We
can.
We
can
go
to
the
dashboard
and
vote
for
things
essentially
to
emojis.
Actually,
so
I'm
voting
for
this,
you
can
view
the
leaderboard.
B
There
are
a
lot
of
quotes
happening
because
there
are
lots
of
words,
because
the
vote
word
is
the
one
voting
it's
a
simulation
of
the
user,
which
keeps
randomly
voting
on
stuff.
Now,
if
we
vote
on
other
stuff,
you
can
see
the
votes
have
increased.
Essentially
because
it's
the
word
about
doing
our
job,
we
can.
We
can
keep
voting
and
things
like
that.
So
now
we
open
the
control
plane.
Now
there
is
another
command
called
linker
dashboard.
B
Now,
if
I
open
the
linkedin
dashboard,
this
is
the
linkedin
dashboard,
where
you
can
manage
your
applications
and
things
like
that
now
we'll
switch
to
the
emoji
voter,
namespace
and
we'll
see
the
deployments
here.
We
have
four
deployments
emoji,
as
we
saw
emoji
board,
etc.
Now,
if
I
click
on
the
web
deployment
without.
B
That
is
sending
requests
into
the
web
application
and
the
web
deployment
is
talking
to
voting
and
also
to
emoji.
There
is
also
prometheus,
which
is
just
scraping
stuff,
so
that
is
also
there
too,
and
then
we
also
have
tab.
This
is
a
control,
plane
component,
so
ignore
them.
You
only
have
to
consider
about
the
vote,
but
the
word
body
is
talking
to
them,
which
is
then
talking
to
voting
and
emoji,
as
you
can
see,
without
with
zero
application
changes
right,
we
did
not.
B
Talk
rich
tv
through
grpc
and
everybody
understands
that
protocol.
You
get
all
these
metrics
right
away.
What
all
do
we
get
here?
You
also
get
live
calls.
Essentially,
this
is
a
live
count
of
the
calls
happening
and
you
can
also
tap
so
if
I,
if,
if
you
click
on
the
tab
here,
if
I
click
on
emoji
water
and
if
I
do
start
essentially,
we
are
seeing
all
the
requests
happening
in
live.
B
So,
for
example,
here
we
can
see
we
we
can
consider
one
of
these
requests
here
it
is,
it
is
going
from
so,
for
example,
if
I
want
to
have
more
filters,
so,
for
example,
I
want
to
see
all
the
requests
that
are
that
are
for
the
web.
If
I
do,
if
I
do
this,
essentially
we
are
getting
all
the
requests
that
are
coming
to
web
and
going
from
there.
So
here
we
can
see
there
are
requests
from
boardwalk
and
to
voting
and
other
things.
B
So
essentially,
all
all
of
this
you
get
all
of
this
visibility
with
zero
configuration
in
your
application.
How
cool
is
that,
so
that's
how
a
service
mesh
will
be
really
helpful
to
your
applications
with
being
being
without
any
configuration,
but
doing
a
lot
of
things
like
here.
We
get
here
not
only.
B
B
Useful
for
debugging
right
here
you
can
see
that
your
web
is
having
only
94
percent
of
the
request
successful
what
are
happening
to
the
other
other
seven
percent.
It's
because
you
you
want
to
understand
the
problem
right
and
then
you
can
see
that
mojo
is
all
positive.
Emoji
service
is
all
positive,
so
it's
mostly
the
voting
deployment.
That
is
wrong
essentially
because
that
is
where
the
problem
is
essentially
because
you
are
getting
a
red
side,
so
it
is
very.
D
B
For
debugging
you
can
also,
but
you
can
also
open
the
voting
and
see
so
see
what
is
wrong
here.
You
can
see
because
of
once
you
open
here.
You
can
also
see
for
each
request
what
is
happening
you
can
see
which
requests
are
negative.
So,
as
you
can
see,
the
donut
request
is
the
one
that
is
failing
whenever
you.
B
B
Linkard
is
pretty
open
source
and
pretty
active
project.
We
have
a
very
active,
very
active
community
slack
and
all
these
things
we
have
monthly
community
calls.
We
just
had
a
question.
We
just
had
a
comment
called
yesterday
too,
that
is
there.
So
we
have
we
are
we
have
mailing
lists,
etc.
Just
like
any
other
project.
We
are
very.
B
So
if
you
are
interested
to
contribute,
you
are
more
than
welcome
to
join
us.
We
are
more
than
help
you.
We
are
more
than
happy
to
help
you
get
started,
so
let
us
see
what
the
open
source
side
of
things.
So,
what
is
cncf
essentially
cmg
is
a
collection
of
technologies
and
practices
that
are
on
the
cloud
side
of
things.
Victor.
A
B
A
B
Onward,
based
essentially
the
benefits
with
linkedin,
we
make
sure
that
you
get
all
these
benefits
with
zero
configuration,
as
we
saw
in
the
example.
We
did
not
change
any
of
the
application
or
we
did
not
even
create
any
services
or
any
other
crds
right.
All
we
did
was
just
add
an
annotation
to
your
your
deployment
and
boom.
You
got
everything
right
out
of
the
box,
so
we
focus
our
focus
is
very
much
on
simplicity,
simplicity
and
and
maintainability.
We
don't
want
you
to
have
like
a
lot
of
configuration
to
so
that
you
fall.
B
A
lot
with
a
lot
of
flexibility
also
fall
into
a
lot
of
a
lot
of
complexes.
Also,
so
we
make
sure
that
every
feature
is
created
in
a
way
that
works
out
of
the
box.
That
is
maintainable
and
a
lot
of
things,
and
the
other
benefit
is
that
the
proxy,
the
proxy
is
written
in
rust
and
custom
made
for
the
service
mesh
use
case.
It
is.
B
As
you
saw
10
mb,
essentially
because
you're
putting
the
proxy
everywhere
right,
every
application
has
to
be
injected
with
the
proxy.
It
is
important
that
it
is
ultra
lightweight
and
is
safe,
so
by
by
writing
in
rust
and
using
all
the
all
the
cutting
edge
stuff
available.
The
all
of
those
are
possible.
It
is
ultra
lightweight,
safe
and
secure,
so
so
that.
A
B
B
Services
that
then
the
control
plane
obviously
will
need
a
lot
more
configuration
right
lot,
more
resources
because
it
is
processing
more
requests.
It
is,
it
is
handling
more
stuff,
but
but
it
is
not,
it
is
not.
It
is
not
complex,
that
is
for
sure,
maybe
and
even
on
the
resources
side,
the
defaults
are.
The
defaults
that
we
give
are
pretty
much
are
pretty
much
fine
unless,
unless
you
are
scaling,
etc,
we.
A
It,
okay,
I
don't
know
if
someone
else
has
another
question,
and
maybe
you
can
give
a
few
minutes.
The
other
thing
that
I
was
thinking
about
is
like
the
major
difference
between
the
function
of
the
ingress
controller
like
and
with
the
load
balancer
that
you
have
with
the
service
mesh.
So
I
don't
know.
B
B
Takes
care
of
the
communication
after
that,
so
english
is
like
an
entry
point
into
your
cluster
and
then
for
all
the
internal
communication
and
communication
across
clusters.
We
also
do
multi-cluster.
We
also
have
multi
cluster
support
there.
You
can
have
services
talking
from
one
cluster
into
other
cluster
for
all
this,
for
all
the
intra
inter
cluster
and
intra
cluster
communication,
you
would
use
a
service
mesh
not
for
the
not
for.
A
B
A
B
B
So
so
so
I'm
expecting
a
lot
of
questions
here,
especially
because
I'm
more
than
happy
to
encourage
all
of
you
in
getting
involved
in
open
source,
both
in
contribution
and
also
usage.
So
if
you
have
any
questions
here,
I'm
more
than
happy
to
help
you
don't
have
to
consider.
You
don't
have
to
feel
bad
that
it's
a
beginner
question,
I'm
more
than
happy
to
answer
any
question
yeah.
So
so.
B
Cloud
native
computing
foundation
is
essentially
a
foundation
that
hosts
a
lot
of
technologies
around
the
cloud
native
landscape.
These
these
technologies
are
available
across
clouds.
That
is
why
the
whole
cloud
native
you
don't
have
to
consider
a
specific
cloud
etcetera.
These
are
available
across
clouds.
The
aps
are
available
across
cloud,
so
these
are
so
these
technologies
help.
You
run
your
applications
in
clouds
and
also.
B
Not
just
cloud
but
the
whole
the
whole
in
the
in
the
deployment
and
management
side
of
things
is
where
the
foundation
is
focused
on.
So
the
the
foundation
also
has
a
lot
of
programs
to
help
you
get
involved
in
open
source,
essentially
both
for
students
and
non-students,
we'll
see
what
those
programs
are.
B
So
so
all
these
pro
throughout
all
this,
even
if
you're
part
of
any
of
the
program
you're
essentially
considered
the
mentee
cncf
mentee
cncf
is
obviously
a
very
welcoming
and
vibrant
and
vibrant
community.
We
got
kubernetes
prometheus
lingerie
and
hundreds
of
other
technologies
that
are
pretty
much
productionized
etcetera.
So
it's
a
pretty
awesome
foundation
with
a
lot
of
impact
and
scale.
B
Most
of
these
users,
like
you
can
say
any
high
scale
company
over
any
of
this
any
of
this
any
of
these
companies,
and
they
would
have
definitely
deployed
something
from
cncf
at
high
scale.
So
and
also
most
of
these
companies
also
contribute
back.
That
is
it
most
of
these.
Maintainers
are
also
pretty
much
experienced
in
the
coordinated
computing
space
and
also
they
know
they
their
stuff.
It's
it's
a
generalized
statement,
but
it's
pretty
too
for
most
projects.
B
So
what
are
the
programs?
There
are
four
programs,
like
four
programs,
currently
the
google
summer
of
code,
which
is
a
which
is
a
program
focus
for
students.
Essentially,
the
google
summer
of
code
essentially
is
for
students
who
wants
to
get
involved
in
open
source
through
through
google's
summer
of
code.
You
can
contribute
to
a
lot
of
projects,
not
just
cncf.
There
are
like
linux
foundation
every
open
source
foundation.
B
A
B
Was
I
was
contributing
through
google
summer,
of
course,
to
link
thirty
project
itself?
It
was.
It
was
a
specific
feature
that
I
was
working
on
and
google
pays
you
for
contributing
to
open
source.
The
money
is
pretty
good.
It
is
just
like
any
other
internship
internship
and
it
is
pretty
pretty
pretty
famous
too.
If
you
are
in,
you
also
get
a
nice.
You
also
get
a
lot
of
career
opportunities.
B
A
B
Through
the
program
you
have
to
get
selected
for
this
too,
and
then
there
is
outreachy
similar
program,
similar
like
like,
like
google,
summer
of
code,
etc.
So
how
do
you
get
started?
We
saw
all
these
programs
that
that
will
pay
you
and
also
we'll
have
here.
We
will
have
a
coordinated
period
to
contribute
and
also
a
mentor
to
for
that.
But
the
program
is
not
all
right,
like
not
everyone
that
contributed
to
open
source
open.
This
came
through
these
programs.
B
Like
I
came
through
these
programs,
and
I
know
some
people
who
did
but
they,
but
most
of
this
most
of
the
maintenance
and
contributors
are
not
through
these
programs,
so
you
don't
have
so
that
is
not
the
only
way.
So
how
do
you
get
started
so
so
the
way
to
get
started
is
to
choose
a
project
right.
Obviously,
there
are
like
there
are
a
lot
of
projects,
so
first
you
have
to
select
a
project
that
you're
interested
in.
B
This
could
be
a
project
that
you
that
you
have
used
in
your
company
or
you
like,
or
it's
some
or
or
it
has
a
feature
that
you
like
or
it's
it's
something
that
you
know
more
on
or
something
like
that.
So
essentially
you
get
you
want
to
choose
a
project.
You
can
choose
multiple
of
them
too.
It's
not
yeah.
There
is
no
rules.
B
There
are
no
rules
here
for
sure
so
pick
the
project
that
you
love
and
have
been
using
anything
and
then
play
with
the
project
like
contribute
issues
like
one
of
the
great
way
to
add
into
any
project
is
to
is
to
show
the
contributors
to
show
the
maintenance
what
they're
missing,
essentially
it.
This
is
essentially
an
issue
request
if
you,
if
you
know
that
something
is
missing,
please
raise
an
issue
request
on
any
project
like
most
of
these.
B
B
That's
how
they
respond
so
so
so
it
is
very
important
that
to
raise
issues,
if
you
find
anything
a
lot
that
is
also
contributing
to
in
a
manner
because
you're
asking
what
is
missing
and
it
helps
maintainers,
prioritize
and
think
of
features
that
are
missing
picking
the
project
like
choose.
One
of
the
projects
from
the
so
picking
the
project
is
different
when
you're
focusing
on
the
on
the
programs.
Essentially,
if
you
want,
if
you
want
to
contribute
through.
B
E
Yeah,
I
guess
so:
okay,
sorry
guys
that
this
is
totally
unexpected
well,
anyway,
I
think
this
is
a
good
example
of
the
skills
salvadorians
we
need
for
inserting
ourselves
in
the
in
the
war
forks,
I
mean
the
international
workforce
right,
so
I
think
this
is
a
good
practice
and
and
that's
why
it's
for
you
to
participate
in
this
kind
of
events
right.
So
the
idea
is
that
we,
we
will
continue
doing
this
in
the
future.
We
also
want
to
hear
with
you.
E
In
there
we
have
the
the
page
of
the
community
and
you
can
get
into
that
and
at
the
same
time,
in
there
you
will
have
a
the
other
members,
and
you
can
also
send
us
your
suggestions
right.
We
will
really
like
to
hear
from
you
about
the
experience
of
the
event
in
english
and
and
the
time
as
well
right.
We
decided
to
move
this
to
the
night,
but
we
would
love
to
hear
from
you
about
the
experience.
E
B
B
So
we
we
were
in
the
picking
project
kind
of
thing,
so
essentially
when
you,
when
you
want
to
pick
a
project
when
you're
going
through
these
mentorship
programs,
essentially
have
to
pick
a
project
that
is
right,
essentially
projects.
Don't
there
is
no
rule
for
the
projects
to
participate
in
these
programs.
It
is,
it
is
all
entry
if
they
want
to
encourage
and
most
of
the
popular
projects
they
do
that,
because
it
helps
them
too.
B
B
B
You
have
to
not
only
choose
a
project,
but
also
a
particular
feature
that
you
want
to
work
on,
so
they
have
issues
around
how
these
projects
are,
what
issues
they
want
you
to
work
on,
and
things
like
that
you
can
choose
from
them
and
apply
for
it
for
a
specific
for
a
specific
feature
in
a
project
like
it
works
like
that,
so
so.
So
what
is
the
usual
process
through
which
a
a
contributor
gets
selected
to
a
program?
So
essentially
you
have
to
some.
B
This
is
the
core
part
of
your
application.
Essentially,
you
have
to
write
a
design
doc,
explaining
that
how
we
would
work
on
the
specific
feature.
This
design
dock
should
not
be
work
in
isolation.
You
have
to
talk
to
the
maintainer,
essentially
through
the
slack,
and
ask
some
questions
around
the
design
decisions
that
you
will
take,
etc.
So
you
have
to
make
sure
that
you
you
are.
You
are
working
on
the
thing
that
is
important
to
the
maintainer
and
also
in
a
way
that
that
the
maintenance
would
be
happy
with
in
in
so
so.
B
A
lot
of
interaction
would
be
really
helpful.
Here.
You
have
to
try
to
try
to
design
dock
with
implementation,
details
explaining
the
different
approaches
and
choosing
a
one.
That
is
that
should
be
the
that
you
think,
is
the
answer
and
goes
well
with
the
with
the
current
way
of
doing
things
and
the
maintenance
has
expressed
interest
about,
and
then
each
of
this
each
of
these
projects
also
have
community
calls
where
you
can,
where
you
can
further
discuss
about
the
feature
that
you'll
be
working
on
and
how
you
will
be
working
on,
etc.
B
So,
on
one
line
you
have
to
work
on
the
application
for
the
program
and
the
other
line.
You
have
to
also
take
up
a
good
first
result.
So
so
you
have
written
a
great
design
doc
that
works.
Well,
etc,
but
it's
also
important
that
important
to
gain
the
trust
of
the
maintenance.
So
how
do
you
do
that?
You
essentially
can
do
that
by
contributing
to
the
project
you
can
there.
Each
of
these
projects
also
have
good
first
issues
that
you
can
contribute
to,
so
you
can
do
that.
B
B
C
C
B
So
so,
as
I
mentioned
you
have
to
so,
it
is
important
that
you
gain
the
trust
of
the
contributors
and
you
and
you
do
that
by
contributing
the
project.
Each
of
the
project
also
has
good
first
issues
and
issues
that
you
can
take
up.
So
that
is,
that
is
how
you
show
that
you're
you're
committed
to
the
project
by
by
taking
up
issues.
B
B
B
A
B
So
so
so
how
do
you
do
it
right?
So
essentially
it
is.
You
have
to
primarily
target
the
the
feature
that
you
are
that
you
were
that
you
were
selected
on
right.
Obviously,
you
have
to
work
on
it
yeah.
It
is
important
that
you
are
pretty
active
in
the
slack
and
pretty
communicative
about
what
the
current
status
of
the
project
is
and
what
you're
working
every
day.
So
it's.
B
Because
you're
getting
paid
for
it
and
things
like
that,
so
it's
it's
essentially
another
job
and
you
have
to,
and
and
how
do
you
work
on
a
job?
You
have
to
make
sure
that
it
is
the
same
like
communicating
through
maintainers
to
the
maintenance
about
the
project,
what
the
status
every
day,
what
are
they
working
on
and
things
like
that?
B
B
It's
important
that
you
follow
the
style
guide
of
the
project
right
essentially,
because
most
of
these
projects
are
pretty
scalable
are
pretty,
are
pretty
much
using
production.
So
it's
important
that
the
maintenance
expect
expect
a
very
good
bar
on
the
styling
and
efficiency
and
and
operations
of
your
code.
It
it.
It
does
not
mean
that
you
have
to
be
a
working
professional
or
your
skill
has
to
be
hundred
percent.
Just
that
you
have
to
be.
B
You
have
to.
You
have
to
be
pretty
pretty
pretty
communicative
about
it
like
asking
the
maintenance
how
to
do
specific
things
and
how
would
they
do
a
specific
feature
and
things
like
that.
So
it's
it's
not
it's
not
about
your
skill
or
talent,
but
it's
about
how
communicated
you
are
and
how?
How
and
how
open
you
are
when
you're,
when
you're
getting
stuck
and
asking
questions
etcetera,
it's
more
about
that
than
your
skill
yeah.
So
it's
important
that
you
also
write
docs
and
blogs
on
your
work.
B
These
are
they
helped
me
a
lot
in
getting
a
lot
of
career
opportunities,
etc.
So
it's
important
doing.
Work
is
awesome,
yeah
and
you
have
to
do
it's
also
important
to
showcase
your
work,
especially
that
you're
you're
getting
started
in
open
source
and
you're
pretty
much
early
in
your
career
like
at
least
through
some
of
this
program.
It's
important
that
you
show
off
your
work
through
documentation
through
blogs,
etc.
B
If
it
did
not
happen,
a
lot
of
cases
like
as
there
is
a
lot
of
competition
for
these
projects,
especially
that
they're
in
remote
internships.
Essentially,
you
can
work
from
your
home.
I
did
I.
I
was
really
fortunate
to
work
with
folks
in
the
us
from
india.
It
was
pretty
awesome
if
it
did
not
happen,
you
don't
have.
B
Like
like
most
of
these
students
who
get
selected,
they
they
go
through
multiple
items
like
I
did
try
go.
I
did
try,
google
somewhere,
of
course,
the
previous
year,
and
I
wasn't
selected
and
I
got
selected,
so
it
works
like
that.
Like
any
other,
any
other
thing
right
you
have
to
you
have
to
try
it
multiple
times
to
get
it
right.
B
So
so
you
yeah.
So
if
you
did
not
get
selected,
you
also
did
a
lot
of
project
on
the
project,
all
the
work
on
the
project
around
documentation,
contributing,
etc.
It's
important
that
you
stick
to
the
project
like
even
even
just
contributing
to
open
source
without
any
of
these
programs
he's
also
actually,
okay.
Here,
people
love,
like
I'm
sure,
I'm
sure
a
lot
of
companies
would
love
for
people
who
have
already
contributing
to
a
project
or
who
already
did
their
contribute
to
a
project.
So
it
does
it.
B
It
is
not
just
about
the
program
and
the
way,
but
it's
it
also
helps
you
further
in
your
career,
like
just
contributing
to
open
source
without
any
specific
program,
and
things
like
that,
so
it
is
very
important
that
you
stick
to
contributions
in
open
source
if
you
are
in.
If
you
are
interested,
if
you
are
early
in
your
career
and
things
like
that,
you
can,
you
can
keep
applying
to
the
next
cycle
right
these.
These
programs
happen
if
or
two
times
every
year,
most
of
these
programs.
B
So
you
have
a
pretty
good
chance
and
and
timeline
to
get
selected
the
next
time
post
mentorship.
So
now
you
guys
are
going
to
complete
your
project,
one
of
the
important
things
that
a
lot
of
students
or
people
through
this
program
misses
that
they
they
don't
contribute
to
the
project
after
the
program.
I
think
this
is
as
important
as
the
program
itself
like
if
you
have
any
other
commitments,
etc.
Sure
you
don't
have.
There
is
no
rule
that
you
have
to
do
that,
but
but,
but
you.
A
B
And
it
it
gave
me
a
chance
to
to
work
on
to
to
work
on
the
product
feature
full
time
now
I
get
paid
to
do
the
same
open
those
work.
So
there
are
a
lot
of
benefits
and
I
would
suggest
everyone
to
to
keep
contributing
the
goal.
The
contribution
does
not
have
to
be
a
code
right
because
you
essentially
have
a
full-time
job.
You
can
you
can
answer
questions
in
the
slack
and
things
like
that.
There
are
like
100
different
things
that
you
could
that
you
could
do
for
a
project
reviewing
pre-orders
bug.
B
C
B
Folks
to
do
to
continue
with
the
project
yep,
so
all
the
all
the
slides
that
have
been
there
are
specific
through
the
projects,
and
I
know
a
lot
of
people
are
not
students
or
can
go
through
these
programs
for
them.
I
would
suggest
them
to
not
go
through
these
programs,
but
but,
but
in
a
similar
timeline
in
a
similar
framework.
Right
essentially
pick
a
project
that
you
are
interested
in.
B
And
then
tell
your
interest
in
this
lag
right,
go
tell
them
that
you're
interested
and
express
your
interest
and
then
ask
them
what
is
the
right
thing
to
work
on
to
get
started
on
like
issues
documentation?
So
it's
a
pretty
act.
It's
a
pretty
pretty
back
and
forth
thing
rather
than
a
single
isolation
like
you,
even
even
like
contributions
for
maintenance.
It's
a
it's
a
very
collaborative
thing
rather
than
an
isolation
thing.
So
it's
important
that
you
talk
to
other
people
and
et
cetera,
and
most
of
these
people
are
pretty
open
and
kind.
B
So
you
know
there
is
no
most
of
the
cases.
You
would
get
a
very
nice
reply,
it's
just
that
you
don't
have
to
waste
their
time.
That
is
what
they
expect
yeah.
So
so
you
have.
You
can
follow
the
same
framework
on
choosing
your
project
getting
started
with
it
and
then
taking
up
smaller
issues
and
then
the
bigger
ones
etcetera.
So
it
works
like
that
and
and
most
of
the
cnc
projects
are
pretty
open
about
it,
and
most
of
these
maintainers
have
also
started
in
the
same
way.
B
A
A
B
B
B
A
B
This
was
a
problem
for
me
to
like,
essentially
in
college,
you
don't
have
beefy
laptops
and
it
was
pretty
like
like
now
that
I
moved
into
a
work,
laptop,
etc.
I
understand
how
problematic
it
was
back
then
like
when
you're
living
with
the
with
the
normal
laptop,
and
you
don't
have
a
lot
to
think
about.
I.
B
Get
work
done
somehow,
but
now
I
can't
go
back
essentially
because
it
is
that
comfortable
in
a
beefy
laptop
right.
So
there
are
tools
essentially
that
I
love
you
to
do
that
kind,
and
things
like
that,
as
as
victor
mentioned,
and
most
of
these
cloud
providers
also
have
student
credits
right,
I
used
to
use
a
lot
of
azure
student
credit.
You
know
myself
whenever
I
used
to
I
whenever.
B
Whenever
I
wanted
to
create
production
is
the
workloads,
there
is
github
student
back.
There
is
azure
student
credit.
There
is
aws
educate
program,
so
make
sure
to
use
all
these
programs
and
and
get
benefits
out
of
them.
I'm
sure
there
are
like,
like
the
like,
the
github
student
pack
has
like
100
different
benefits.
You
can
get
jet
brains
all
these
ideas
for
free
all
the
professional
stuff
for
free.
So
it's
pretty
pretty
cool
that
students
can
access
that
and
you
have
to.
A
Yeah,
the
other
thing
that
I
haven't
tried
and
let
me
paste
in
the
chat
yeah
so
that
there
is
this
tool
where
suppose
that
notifies
you
the
new
new
issues.
So
obviously
you
can
monitor
like
all
these
long,
long,
low
organ
fruits
you
can,
but
I
haven't
tried.
I
mean
I
have
to
do
that
knowledge
that
I
haven't
tried
but
yeah.
I
guess.
B
Yep
yep
there
is,
there
is
obviously
the
concern
about.
It
could
be
language,
it
could
be.
It
could
be
your
laptop
and
I
had
constants
too,
and
I
would
not
say-
and
I
would
not
generalize
it
that
easily,
but
but
I
I
but
for
a
student
who
wants
to
get
started,
I
don't
think
these
are
as
big
as
the
problems
as
for
maintenance.
For
example,
if
I
was
a
maintainer
to
the
project,
I'm
sure
the
the
laptop
problem
could
have
been.
The
laptop
thing
could
have
been
a
problem,
but
for
a
getting
started
person.
B
A
B
D
D
Okay,
the
executive
director
of
the
cloud
native
computer
foundation-
then
he
passed
away
on
november
1st
this
year,
okay,
but
last
year
in
november,
I
had
a
chance
to
meet
with
him,
and
so
one
of
the
things
that
we
were
talking
about,
it
is
how
where
the
contributors
are
coming
from,
and
so
most
of
the
contributors
for
these
projects
are
from
europe
from
the
us
and
also
asia.
So
the
question
is:
how
can
we
increase
the
the
the
number
of
contributors
from
latin
america
and
so
then
immediately
thought
about
this
program?
D
D
I
know
that
maurizio
and
victor
can
pitch
in
and
share
more
about
how
we
are
doing
these
meetings,
some
of
the
things
that
we
have
come
making
a
priority.
For
example,
we
want
to
share
with
you
how
people
can
set
up
businesses
related
to
kubernetes,
community,
kubernetes
and
and
open
source
and
and
cloud
name.
There
are
over
200
companies.
Startups
tyrone
worked
for
one
of
those
companies
he
this
company.
D
D
The
cloudberry
computer
foundation
has
at
least
103
000
contributors
over
500
companies
members,
so
the
biggest
companies
in
the
world
you've
got
jp
chase.
You
got
comcast,
you
got
facebook
intel,
you
name
it,
okay,
so
by
by
joining
the
the
community,
you
start
you
start
kind
of
coming
in
contact
with
with
other
friends.
You
know,
people
that
you
meet
through
qcon,
right
and
so
slack
makes
it
easy
to
communicate
with
people
all
over
the
world.
Okay,
we
wanna
know
about
your
interests.
Your
questions,
so
please
share
we're.
D
D
C
D
D
E
Now,
well,
anyway,
term.
I
think
it
would
be
wise
to
hear
from
you
any
advice
that
you
might
have
for
for
for
young
professionals
here.
You
know
we
have
a
combination
of
students
for
university
people
who
are
starting
just
to
give
you
some
background
here.
What
we
have
we
are
basically
using
the
technologies.
Universities
in
here
are
still
teaching
like
the
classic
dotnet
javas.
C
E
It's
we
have
like
a
technical
depth
in
here.
Let's,
let's
call
it
like
that,
okay,
so
I
think
it
depends
pretty
much
on
what
the
attitude
of
each
of
us
is
regarding
what
kind
of
things
we
should
go,
look
for
and
to
land
by
ourselves.
So
I
would
like
to
hear
your
word
regarding
this:
how
how
do
you?
How
do
you
do
it?
Many
people
in
here
have
the
idea
that
indians
are
born
with
a
laptop
under
their
arm.
E
You
know,
and
and-
and
there
are
a
lot
of
probably
misconceptions
between
and
images,
and
you
know
so.
I
would
like
to
hear
from
you
about
that
experience.
What
what
brings
you
to
to
the
career,
how
how
you
actually
move
into
that?
We
hear
about
your
very
beginning,
but
you
know
in
a
more
let's
say,
active
way.
A
B
There
is
obviously
the
the
top
like
in
any
any
other
university
programs
like
it
is
the
top
layer,
the
top
the
top
tier
colleges,
and
you
get
a
lot
of
opportunity,
but
but
but
in
india
we
see
a
lot
of
people,
including
going
out
of
their
way.
I
mean
it
could
be
because
of
I
mean
density
population
or
whatever,
like
I.
I.
B
Wasn't
kubernetes
there
wasn't
there
there
weren't,
even
containers
is
that
the
college
or
any
other
university
program
will
give
you
the
the
the
fundamental
building
blocks
on
what
a
run
time
is
and
what
a
garbage
companies
and
things
like
that.
But
you
have
to
build
knowledge
on
top
to
make
sure
that
you're
you're
in
you're
in
touch
with
the
current
state
of
things
right.
The
college
stuff
is
on
basics
and
the
previous
state
of
things,
but
you
have
to
understand.