►
Description
An unstoppable shift is underway. Just as the way we consume software has changed forever; the way we create, deploy and manage software is reconstituting away from centralised monolithic IT silos - to small, hyper-efficient, specialised delivery squads who don't just understand the business, they are the business.
This radical yet obvious decentralisation mirrors the breakdown of software from monolithic applications to discreet microservices, affording a level of speed and agility not witnessed before. It comes with it's challenges, however. Join Brad Drysdale, APAC Field CTO from Kong as we dissect this evolution and hear stories, challenges and opportunities from the front line of software delivery.
A
But
I
guess,
if
you
stand
back
from
the
picture,
there's
essentially
three
mega
trends
that
we
see
are
powering
digital
transformation
and
and
kong
in
that
regard,
and
and
really
it's
all
about
cloud
native
as
we
call
it
so
you
know
the
first
one
is
and
I'm
sure
you're
all
aware
of
this.
You
know
moving
from
virtualization
to
containers,
to
docker,
to
openshift
to
kubernetes,
etc.
You
know
to
get
that
flexibility
in
that
movement
at
the
sort
of
infrastructure
level.
If
we
come
up
a
level,
we
see
that
applications
are
clearly
changing.
A
You
know
we
used
to
have
application,
servers
and
databases
and
and
clearly
that
middle
piece
is
fragmenting
and
becoming
you
know,
hundreds
of
apis
and
microservices
containers
and
controllers,
etc,
and
then
really
the
third
mega
trend
is
that
multi-cloud
is
happening.
You
know,
I
think
everyone
would
ideally
like
to
align
with
a
single
cloud
provider
and
some
are
lucky
to
do
that.
But
you
know
our
customers
are
telling
us
that
they
have
to
deal
with.
You
know
all
these
different
cloud
providers
and
they
have
you
know
on-prem
hardware,
etc.
A
So
you
could
say
that
that
kong
has
excelled
in
its
ability
to
address
this
sort
of
full
api
and
life
cycle
management
requirement
around
apis
and
having
apis
at
the
center
of
your
strategy.
So,
whether
that's
you
know
through
design
deployment,
management,
security
measurement,
etc,
cong
is
the
the
world's
number
one
api
gateway.
A
That's
in
use
today,
so
open
source
at
its
heart
and
there's
been
over
250
million
downloads
of
kong
so
that
that's
where
sort
of
kong
you
know
you
could
say
rapidly
spread
through
the
market
and
then
since
then,
we've
been
on
this
big
evolution
journey.
I
mean
the
con.
The
company
has
doubled
in
size
in
the
last
10
months.
Something
like
that.
A
So
I've
been
on
board
for
three
months:
brad's
been
on
board
for
for
not
much
more
and-
and
you
know
it's
just
testament
to
kong,
moving
into
this
sort
of
full
service
cloud
native
connectivity,
company,
and
so
really
the
secret
sauce
of
kong
is
to
to
be
independent
of
any
platform.
Any
protocol
or
any
approach
that
you're
using
to
build
your
applications
so
that
you
know
you're
not
locked
into
a
tech
stack
and
yeah
kong
essentially
runs
anywhere
and
runs
on
everything
which
really
allows
companies
to
embrace
their
own
complexity.
A
And
so
what?
What
that
really
means,
in
a
nutshell,
is
that
we
help
companies
to
innovate
faster.
We
help
them
to
capture
new
digital
opportunities
and
we
help
them
reduce
the
cost
and
improve
efficiency
in
their
operations
and
just
to
show
a
few.
You
know,
organizations
that
we're
working
with
here's
organizations
such
as
tal,
the
insurer,
national
australia,
bank,
australia,
post
eubank,
commonwealth
bank
qantas
loyalty.
A
You
know
transport
for
new
south
wales,
so
we're
seeing
a
sort
of
growth
in
government
and
the
likes
of
university
of
auckland
in
in
new
zealand,
and
so
you
know-
and
all
of
that
has
happened-
you
know
rapidly
in
a
short
space
of
time,
and
so
with
that
I
I'd
just
like
to
introduce
brad.
So
I've
actually
worked
with
brad
for
quite
a
number
of
years
now
and
he's
a
veteran
you
could
save
the
connectivity
problem
he's
spent
time
at
bea.
He
worked
on
the
first
application
server.
There
spent
time
at
tibco
at
accenture.
B
This
presentation
is
really
only
the
first
step
in
the
start
of
that,
because
yeah,
as
you
could
imagine
decentralization,
is
the
the
process
that
manual
the
journey
that
every
organization
is
going
through,
as
they
look
for
ways
to
innovate,
faster
and
to
capture
new
markets
and
to
do
that.
They're,
breaking
up
the
monolith
and
they're
they're
starting
to
become
engineering
organizations
themselves
to
become
software,
led.
So
there's
lots
of
topics
to
talk
about
I'll
touch
on
an
overview
of
decentralization
and
an
overview
of
kong.
B
B
First,
is
faster
innovation,
so
they're
able
to
innovate
not
only
faster,
but
also
keep
the
quality
of
innovation
very
high
and
that's
often
been
a
very
difficult
conundrum
to
manage.
Do
you
move
fast
and
sacrifice
quality,
or
do
you
keep
the
quality
of
the
assets
you
build
really
really
high
at
the
detrimental
speed
and
so
finding
ways
to
innovate
and
bring
agility
into
the
organization
quickly
with
higher
quality
becomes
critical.
B
How
do
you
reduce
the
cost
space
within
an
organization
and
as
well
as
I
increase
the
efficiency
as
well
as
capturing
new
market
opportunities
and
a
study
that
we
did
fairly
recently
suggested
that
you
know
51
of
the
enterprises
that
we
surveyed
said
that
if
they
would
fail
or
they
would
be
acquired
within
three
years,
if
they
are
unab
unable
to
keep
up
with
the
pace
of
innovation,
so
we
as
digital
consumers
and
digital
customers?
We
put
a
lot
of
pressure
on
every
organization
to
be
better
and
to
deliver
a
personalized.
B
So,
as
you
go
on
this
journey
of
creating
modern
distributed
applications
to
provide
that
agility
for
the
business
you'll
find
that
the
applications
will
consist
start
to
consist
of
an
increasing
number
of
smaller
and
smaller
services.
As
you
start
to
break
down
those
large
monolithic
applications,
you
adopt
more
cloud.
You
adopt
more
sas.
You
lean
into
innovation.
From
an
engineer.
B
This
decentralization
and
the
increasing
complexity
and
the
required
diversity
of
technology
platform
protocols
and
patterns
leads
to
an
incredible
amount
of
opportunity.
The
opportunity
to
innovate
faster
and
the
opportunity
to
innovate
with
far
higher
quality,
and
what
we're
seeing
in
the
marketplace
is
that
the
modern
enterprise
is
the
organization
that
is
embracing.
All
of
these
factors
is
embracing.
Decentralization
is
in
embracing
the
distribution
and
the
complexity
and
the
diversity
of
the
tools,
the
platforms,
the
protocols
and
choices
of
technology
and
they're,
very
much
leaning
into
that
to
embrace
the
opportunity
that
comes
from
it.
B
B
And
so
those
three
different
connectivity
types
all
have
technologies
which
implements
and
underpins
each
of
these
and
those
technologies
are
actually
starting
to
get
very
diverse
and
very
complex.
We
see
edge
api
gateways,
fulfilling
the
need
for
edge
connectivity.
We
see
internal
api
gateways
as
well
as
ingress
and
egress
controllers
for
for
managing
that
cross-application
connectivity
and
we're
starting
to
see
a
rising
popularity
around
service
measures
to
manage
the
service
of
service
connectivity
and
security
within
the
app.
So
the
the
intra-app
style
connectivity.
B
B
First
mindset
is
critical
to
the
modern
enterprise
succeeding
going
forward,
and
so,
as
you
can
see,
the
the
qualities
of
organizations
differ
from
historically.
What
was
a
very
centralized
view
of
applications,
a
very
centralized
view
of
an
I.t
organization
to
something
now
that
is
very
distributed.
B
So
here
are
the
things
that
we
at
kong
think
are
critical
for
you
to
be
able
to
deliver
upon
this
decentralization
and
the
journey
to
decentralization
as
you
go
on
it,
and
so
there's
end-to-end
service
connectivity
across
those
three
patterns
that
we
mentioned
earlier.
You
need
to
have
support
for
architectural
freedom.
Developers
can
choose
the
right
languages
to
build
the
right
pieces.
They
need
the
best
protocols.
Perhaps
graphql
makes
more
sense
for
some
apis
than
just
plain
wrist
starts.
B
Perhaps
an
event-driven
architecture
makes
more
sense,
so
the
requirement
to
be
able
to
support
all
of
that
a
requirement
for
hybrid
and
multi-cloud.
Where
does
this
stuff
get
to
deploy?
Do
you
deploy
this
microservice
in
gcp
for
their
ai
benefits,
the
machine
learning
capabilities
that
perhaps
are
better
service
up
there?
B
Or
are
you
deploying
some
of
those
capabilities
into
azure
because
you're
more
of
a
micro
sort
of
a
microsoft
led
organization,
or
perhaps
both
or
one
today
and
a
different
one
tomorrow,
most
organizations
are
looking
at
ways
to
use
containers
for
mesos
or
kubernetes
style,
orchestration
environments.
So
how
do
you
embrace
all
of
that
technology?
And
how
do
you
do
that
in
a
way
that
is
not
only
native
to
those
tools
but
very
natural
to
the
users
and
the
operators
of
those
tools
and,
of
course,
performance,
scalability,
throughput
and
elasticity
become?
B
You
know
the
qualities
that
any
organization
wants
to
be
able
to
deliver
in
the
modern,
decentralized
environment,
extensibility
the
ability
to
to
be
able
to
make
products
do
more
and
to
bend
and
flex
with
the
shape
of
the
business
and
the
requirements
of
the
business
rather
than
being
rigid
and
fixed,
and
what
I
believe
is
probably
the
most
important
piece
around
all
of
this,
which
is
automation.
You
can't
do
all
of
this.
B
Connect
is
the
platform
for
kong
that
delivers
all
of
that.
It
embraces
all
of
those
technologies,
platforms,
tools
and
protocols,
and
it
allows
you
to
deliver
the
end-to-end
service
connectivity
across
edge
api
gateways
across
ingress
and
egress
controllers
across
service
mesh
style
architectures
within
any
cloud
across
clouds,
multiple
and
hybrid
cloud,
and
to
be
able
to
do
that
in
a
highly
performant
fashion
in
a
very
extensible
manner
and,
most
importantly,
fully
automated
fully
automated.
In
a
way,
that's
native
to
those
tools
and
platforms
that
we
spoke
about
earlier
and
automatable
in
a
very
natural
way.
B
In
order
for
you
to
continue
on
that
journey
to
decentralization,
and
so
I
want
to
stop
there
with
the
slides
and
give
you
a
bit
of
an
overview
of
some
aspects
of
connect
to
introduce
you
to
connect
and
to
show
you
what
it's
capable
of
today
at
a
very
high
level.
Knowing
that
in
further
webinars
we'll
be
going
into,
you
know
deeper
aspects
and
various
different
pieces
of
some
of
the
technology
I
mentioned
earlier.
B
B
Every
single
one
of
those
services
that
I
mentioned
earlier,
whether
it's
a
micro
service
with
insider
kubernetes
or
a
product
led
api
that
you're
delivering
to
an
api
gateway
or
apis
that
are
exposed
by
your
applications
or
various
sas
applications
that
you're
embracing
can
be
registered
inside
of
service
hub.
The
goal
here
is
to
be
that
catalog
of
all
of
the
services
to
give
you
a
level
of
visibility
and
control,
and
so
we'll
go
ahead
and
we'll
create
a
new
service
and
show
you
what
that
looks
like
also
in
here
there's
this
concept
of
connection.
B
So,
as
you
know,
the
the
community
you're
going
to
start
building
around
developers
and
users
and
consuming
applications
of
those
various
different
services
developers
will
get
registered
in
here.
There's
this
concept
of
a
developer
portal
and
I'll
show
you
what
one
of
these
looks
like
as
well.
So
if
you
have
apis
that
you're
developing
that
you
wish
to
document
and
make
discoverable
to
other
users
within
the
organization
or
other
developers
across
the
organization
or
publicly
exposing
out
to
the
broader
api
economy
for
for
people
to
discover
and
consume,
then
a
developer
portal.
B
So
whilst
this
control
plane
component
here
that
I'm
showing
you
in
this
form
is
sas
hosted
by
kong,
you
can
have
run
times,
be
it
api
gateway,
runtimes,
ingress,
controller,
runtimes
or
sidecar
microservice
driven
mesh
runtimes.
You
can
have
them
all
eventually
registered
for
to
be
managed
and
controlled
and
secured
and
protected
inside
of
the
kong
environment.
B
And
so
that's
what
I'm
going
to
do.
First,
I'm
actually
going
to
create
a
runtime
on
my
laptop
here
and
attach
it
with
this
connect
control
plane
so
that
we
can
start
to
play
around,
and
so
it's
really
easy
to
configure
a
new
runtime.
I'm
going
to
go
with
the
quick
start,
docker
setup
here,
because,
as
you
can
see,
I've
got
docker
running
on
my
mac.
B
B
B
B
B
B
Click
create,
so
what
that
does
is
it
creates
an
entry
in
my
service
hub
as
a
pet
store
api,
there's,
no
metrics
or
anything
of
social
analytics
associated
with
the
api.
Yet
because
we've
just
registered
it,
we
haven't
tied
it
to
any
particular
back-end
implementation
and
obviously
there's
no
traffic
going
through
here.
B
B
I
want
to
proxy,
through
my
api
gateway
here,
for
management
and
control,
and
so
in
order
to
do
that,
we're
going
to
register
that
as
a
version
implementation
that
api
already
existed.
But
if
you
were
taking
a
design
first
approach
to
apis-
and
you
were
building
this
from
scratch,
then
you'd
probably
be
using
the
insomnia
tool
that
we
provide
for
doing
that
insomnia
is
that
design
first
tool
which
allows
you
to
build
the
swagger
specification
and
get
visual?
B
Insomnia
is
fully
automatable,
so
you
can
synchronize
this
with
git.
Obviously,
there's
a
repository,
but
underneath
all
of
this
there's
an
insta
command
line
tool
so
that
using
your
ci
cd
tools,
you're
able
to
do
everything
you
can
pretty
much
do
from
a
development
perspective
inside
of
somni
insomnia,
you'll
be
able
to
do
on
a
command
line
level
in
an
automated
fashion
as
well.
B
So
let's
go
ahead
and
add
an
implementation
for
that
version.
One
of
that
api.
So
I
click
on
version.
One
and
you'll
see
here.
I
have
the
ability
to
register
a
new
application
as
well
as
upload
a
spec.
If
I
have
one
so
I'll,
go
ahead
and
create
a
new
implementation
and
I'm
going
to
copy
in
that
url
for
that
pet
store
which
we
tested
on
the
back
end.
B
And
now
I
register
a
route,
so
you
can
have
multiple
routes
pointing
to
multiple
different
implementations
of
services.
Here,
I'm
going
to
call
this
route
pet
store
and
I'm
going
to
leave
everything
by
default.
You
can
obviously
specify
a
specific
host
or
specific
protocols
that
you
want
to
have
support
for
specific
http
methods
that
you
want
to
support
or
disallow,
and
then
you
specify
path.
B
B
B
I
have
a
spec
associated
with
this,
so
I
can
go
ahead
and
update
my
my
service
entry
here
using
the
pet
store,
yaml
file
that
came
out
of
insomnia,
and
so
what
that
will
do
was
give
me
the
documentation.
I
need
when
it
comes
time
to
expose
this
api
out
through
a
developer
portal,
and
so
that's
going
to
be
our
next
step.
We'll
see
that
we
have
one
service
in
here
we
see
some
analytics
starting
to
come
through.
B
So
this
is
an
api
that
I
want
to
want
to
make
discoverable.
So
I
want
to
put
this
into
a
developer
portal
so
that
consumers
of
it
can
come
and
search
for
it,
discover
this
api
understand
how
it
works,
test
it
and
play
with
it
inside
the
developer
portal,
read
the
documentation
and
then
hopefully
want
to
use
it,
and
so
they'll
want
to
be
able
to
get
registered
access
for
that
api
and
issue
some
api
keys
and
we'll
walk
through
how
that
would
work
today
as
well
and
so
up
in
here.
B
I
would
go
back
to
my
api,
let's
run
into
my
services,
so
we
can
see
here
in
service
hub.
We
now
show
that
one
pet
store
api
and
I
can
click
on
here
and
say.
The
two
bits
I'm
interested
in
is
enable
application
registration
so
that,
when
developers
come
and
request
access
to
this
api,
I
can
specify
how
we
will
and
how
what
sort
of
authentication
we'll
expect
when
people
are
registering
for
access.
B
If
I
was
to
head
over
here
to
the
developer
portal
pane,
you
could
see
that
there's
one
api
which
is
being
exposed
as
through
the
develop
portal,
and
we
can
go
ahead
and
view
that
in
the
portal,
so
here
I
need
to
register
to
get
access
to
the
portal
because,
as
I
mentioned
earlier,
we
provided
this
is
a
private
developer
portal.
There
is
an
option
down
here
in
the
shared
config,
where
sorry
in
the
settings
where
you
can
specify,
where
are.
C
B
So
I'll
pick
the
right
page
you
can
specify
here.
If
you
want
to
enable
a
public
portal
or
not,
I
haven't
quite
often
you'll
find
organizations
will
have
a
mix
of
both
they'll
have
private
portals
for
for
apis
that
they're
just
exposing
internally
to
their
own
developers
or
to
a
community
of
managed
developers,
and
you
may
also
have
a
public
portal
where
your
public
facing
apis
are
exposed,
and
people
may
not
need
to
register
to
use
that.
So
I'm
going
to
go
ahead
and
sign
up
as
a
developer,
really
quick.
B
B
You
know
you
can
go
ahead
and
try
this
out
search
for
all
available
pets
and
you'll,
see
that
we'll
go
ahead
and
call
that
api
to
be
able
to
test
it
out
and
return.
This
response
coming
back
so
once
you've
understood
a
little
bit
about
this
this
api,
perhaps
you
know
the
next
step-
is
to
go
ahead
and
register
for
it
before
I
show
you
that
you
can
see
here
that
you
know,
there's
lots
of
in
the
developer
portal
settings.
There's
lots
of
different
things.
B
You
can
you
can
set
up
so
this
we
mentioned
you
can
set
up
a
public
portal.
You
can
auto
approve
developers.
Auto
approve
those
applications,
as
we
saw
before
you
can
configure
the
appearance
of
the
portal,
and
so
you
can
change
themes.
You
can
add
logos,
you
can
change
text,
colors
and
buttons
and
font
sizes.
If
you
want
inside
of
every
particular
service
that
you
have
registered,
you
can
actually
also
upload
markdown.
B
There's
also
support
here
for
a
portal
cli,
so
you
don't
have
to
do
all
of
this
through
through
this
connect
ui.
You
can
obviously
automate
that,
through
the
back
end
using
that
that
portal
command
line
tool
as
well,
okay,
so
as
a
potential
user
of
this
api,
I
want
to
get
access
to
it.
So
I
click
register
and
I
create
an
application
I'll
call
it
my
app
reference.
B
I
want
to
request
access
to
this
particular
api
because,
as
we
saw
before,
we
had
the
the
requirement
to
register
for
access
to
this
api
enabled
and
so
that
pops
up
that
button
there
and
then,
when
I
go
back
as
a
administrator
of
this
api,
you'll
see
that
there's
a
request
to
approve
that
particular
application
for
access
to
that
api.
So
I'll
go
ahead
and
approve
that.
B
Now,
when
I
go
into
the
list
of
my
apps
as
a
developer,
so
I
look
at
the
catalog
of
applications
that
I
have
you'll
see.
My
app
was
the
one
that
I
got
access
I
registered
for
access
to
that
store
service.
At
the
back
end,
I
can
go
ahead
and
create
a
credential,
and
so
I
specified
a
standard
api
key,
and
so
I've
got
an
api
key.
That's
been
registered
here
for
calling
that
particular
backend.
B
So
when
we
registered
or
published
that
api
through
developer
portal,
we
enabled
that
key
authentication
for
people
who
want
to
consume
it.
So
now,
if
I
was
to
try
and
call
that
api
you'll
see
that
we
get
the
401
unauthorized
coming
back,
because
there
is
a
policy,
that's
been
enabled
on
this
particular
service
and
I'll
show
it
to
you.
B
You'll
see
that
there's
a
key
authentication
plugin
that's
been
enabled
right
and
so
plugins.
We
provide
a
whole
library
of
plugins
that
you
can
configure
and
apply
to
your
your
services
inside
of
kong
at
the
global
level,
so
that
all
services
inherit
a
particular
a
particular
policy.
You
can
do
it
at
the
service
service
or
the
service
version
level,
and
you
can
also
do
it
at
the
route
level
as
well.
B
B
We
obviously
need
to
apply
an
api
key
here
to
be
able
to
call
it.
So
if
I
copy
that
key
you'll
see
now
as
a
registered
user
of
that
api,
I
can
now
call
that
api
using
that
key
and
that
will
start
to
capture
my
identity
through
the
analytics
and
the
platform
so
that
it
knows
who
that
it's
me
that's
being
called
that
you're
able
to
capture
that
analytics
and
that
consumption
data
and
perhaps
monetize
it
if
you're
charging
people
for
use
or
calling
of
those
apis.
B
You
see
if
I
change
something
in
the
api
key.
It's
not
going
to
work
so
that
is
actually
validating
me
using
a
valid
api
key
you
could
be
using
oauth,
you
could
be
using
open
id
connect.
You
could
be
using
saml,
you
could
be
using
all
sorts
of
various
different
authentication
and
authorization
mechanisms
to
protect
and
and
secure
those
sorts
of
apis.
I
just
used
a
simple
api
key
here,
because
it
was
really
easy
for
me
to
test
some
of
the
other
policies
that
we
provide,
that
I
can
show
you
here.
B
So
if
I
was
to
add
a
new
plugin
it'll,
take
you
to
our
plugin
hub.
You
can
see
there's
a
whole
bunch
of
plugins
around
authentication,
whether
it's
you
know
authenticating
against
back-end
identity
management
systems
or
using
various
different
protocols
like
oauth,
open
id
connect.
On
top
of
that
to
be
able
to
do
some
sort
of
more
complex
authentication
jot
and
jot
signing
security.
So
there's
bot
protection,
there's
cross-origin
resource
sharing,
there's
restriction
of
ips,
there's
all
sorts
of
capabilities
inside
of
there
there's
a
whole
bunch
of
traffic
control
proxies.
B
So
you
can
see
when
we
called
this.
You
saw
that
the
the
latency
inside
the
gateway
was
very
small,
and
so
it
was
sort
of
you
know
three
two
or
three
milliseconds,
but
the
the
latency
to
the
back
end
was
much
larger
and
that's
because
I'm
calling
some
sort
of
service
which
is
out
across
my
broad
abandon
up
in
the
cloud
somewhere,
and
so
one
of
the
policies
here
we
have
is
caching,
there's
standard
cache
in
advanced
caching
and
there's
some
caching
that
we
provide
and
rate
limiting.
B
We
provide
around
various
different
protocols
like
graphql,
so
I'll
just
enable
proxy
caching
there's
lots
of
different
things
you
can
configure
here.
I'm
just
gonna
specify
it
for
the
purpose
of
the
demo
that
we're
using
an
in-memory
caching
strategy
and
click
create
and
now
you'll
see
that
plug-in
appear
down
here.
B
If
I
was
to
call
that
service
again,
it
would
go
all
the
way
through
to
the
back
end
and
cache
the
response,
and
you
can
see
that
the
the
round-trip
latency
is
about
the
same,
because
the
first
cache
was
a
miss
the
second
time
I
call
it.
We
got
a
cache
hit,
and
so
there
was
no
latency
to
the
upstream
service
because
we're
caching
that
directly
inside
of
the
gateway
and
the
response
you
see
here-
is
basically
either
one
millisecond
or
even
zero.
It's
negligible
altogether.
B
Some
of
the
other
policies
available,
there's
rate
limiting,
for
example,
so
you
might
choose
to
say
this
particular
api.
In
this
example,
I'm
using
can
only
accept,
say
you
know
three
requests
per
minute
and
you
might
choose
to
apply
that
on
a
route
or
a
particular
part
of
the
api
or
a
particular
version
of
the
api,
or
on
on
base
to
brown
particular
styles
of
consumers
or
groups
of
consumers
that
are
using
it.
B
B
B
And
now,
when
I
call
it,
we
should
get
a
response
and
you'll
see,
there's
some
additional
headers
that
have
been
provided
here,
which
is
around
the
rate
limiting.
So
I
can
call
that
three
times
before
I'll
get
a
four
to
nine
too
many
requests
so
there's
an
example
of
a
rate
limiting
plugin,
that's
in
use
there,
and
so
you
know,
there's
a
whole
library
of
plugins
available
here,
most
of
them
provided
by
kong.
B
So
we
provide
a
plug-in
development
kit
where
you
can
build
plugins
using
go
or
lua
to
do
any
sort
of
custom
work
that
you
typically
want
to
do
with
inside
of
an
api
gateway,
and
you
can
have
your
plugins
provided
in
this
list
here
as
well,
and
so,
if
we
go
back
to
the
very
high
level
of
service
hub
you'll
see
that
our
pet
store
now
starts
to
have
some
analytics
associated
with
it
based
around
the
consumption
of
it.
So
we
can
see
the
requests.
We
have
any
failed
requests.
B
The
number
of
versions
that
are
provided
the
analytics
can
also
be
broken
down
on
a
particular
version
basis
as
well.
So
these
analytics,
depending
on
your
your
tier
of
usage
of
connect,
can
be
the
data
here
can
be
held
within
the
platform
for
up
to
a
year.
But
obviously
you
have
the
ability
to
export
that
data.
We
also
through
those
plugins.
B
You
saw
before
have
the
ability
to
feed
in
real
time
the
consumption
data
and
the
logs
out
to
tools
like
prometheus
or
zipkin,
or
out
to
telemetry
tools
or
things
like
splunk
or
datadog
or
sumo
logic.
For
you
to
integrate
with
those
various
other
tools
and
environments,
you
have
as
well
just
to
give
you
a
bit
of
an
overview
very
quickly
on
some
of
the
automation
aspects
associated
with
this.
We
mentioned
that
the
insomnia
tool
has
an
inso
cli.
B
There's
a
portal
cli
there's
also
a
tool
which
we
call
deck
and
deck
is
our
command
line
tool
to
be
able
to
automate
all
of
the
aspects
of
connect
and
the
api
gateway
in
a
programmatic
approach.
We
also
provide
platform
level
apis
and
so
there's
apis
within
connect
that
you
can
call
with
different
personas.
In
order
to
do
all
the
things
you
saw
me
do
in
that
ui
and
to
interact
with
the
platform
in
a
programmatic
fashion
and
dec
is
really
a
cli.
That's
just
leveraging
those
apis.
B
So
the
first
thing
I
might
want
to
do
here
is
use
deck,
to
connect
to
my
connect
instance
and
to
dump
the
configuration,
and
so
what
that
command
there
is
actually
doing
is
producing
a
yaml
file,
which
is
a
declarative
representation
of
the
configuration
of
connect.
And
so
you
see
in
here
I
have
my
pet
store
api.
There's
one
version:
there's
all
of
the
attributes
and
configuration
aspects
associated
with
that.
B
So
using
my
automation
tools
and
my
ci
cd
pipelines
that
I
might
have
in
place,
I
would
potentially
be
able
to
create
a
new
api
using
the
tools
which
edits
this
particular
file,
which
I
would
then
install
in
github
for
for
versioning
purposes.
So
I
might
call
this
my
new
api
and
I
might
give
it
a
description
of
this-
is
a
new
api.
B
I
won't
give
it
any
other
configurational
aspects,
other
than
just
registering
that
as
a
new
service
with
with
that
as
the
minimum,
and
so
when
a
new
api
gets
developed
by
a
team,
the
configuration
of
it
inside
of
connect
can
be
injected
into
this
yaml
file.
You
pull
it
out
of
github
inject
this
into
there
version
control.
You
get
diff
management,
you
get
support
for
all
the
different
versions
of
it,
and
then
you
can
sync
that
back
with
your
connect
instance.
B
B
B
I
can
put
you
know
just
to
make
another
quick
change
for
you
here.
I
can,
you
know,
put
hello
world
there
and
you'll
see
when
I
sync
that
will
dynamically
change
that
information
up
here
inside
of
server
sub,
and
so
this
deck
tool
will
show
you
what's
going
on
and
the
changes
that
have
been
made
and
here
you'll
see
that
this
work.
This
number
changed
here
from
pixel
api
to
pitstop,
api,
hello,
world.
B
So
lots
of
tools,
lots
of
apis
lots
of
support
for
for
automation,
of
everything
that
we've
seen
here.
If,
if
the
api
gateway
that
I'm
running
here
in
docker
happened
to
be
a
sidecar
that
was
running
as
a
data
planning
site
of
a
mesh
to
manage
your
microservice,
for
example,
and
if
that
was
in
kubernetes,
then
I
would
use
cubecuttle
to
do
a
similar
sort
of
thing.
I
I
would
configure
kong
and
the
services
that
were
being
provided
in
the
configuration
of
the
services
that
were
being
provided
for
that
particular
micro
service.
B
I
would
use
git
ops
practices
and
I
would
use
a
declarative
approach
to
how
I
would
configure
and
manage
those
using
cube
cuttle
rather
than
using
the
deck
tool
here
as
well.
So,
as
I
mentioned
before,
the
way
you
automate
the
way
we
interact
with
those
technologies
are
native
to
those
tools,
as
well
as
very
natural
for
the
operators
of
those
tools.
B
C
Yeah,
please
feel
free
and
enter
your
questions
in
the
chat
or
speak
up
if
you'd
like
to
as
well.
We
have
a
smaller
group,
so
you're
welcome
to
speak
up,
yeah
brad,
there's
no
any
specific
question,
because
from
my
side
it
was
my
first
you
know
means
I
have.
I
have
touched
api
gateways
earlier,
like
api
gateway
and
aws
or
api,
that
that
was
majorly
I
test
and
but
I
was
curious
to
see.
B
B
B
You'll
see
the
these
ones
who
support
us
the
stuff
we
build.
We
obviously
provide
enterprise
support
for,
but
if
I
scroll
down
to
some
of
these
other
ones
here,
you
can
see
that
the
octa
one
is
supported
by
octa,
so
we'll
support
them.
The
ones
in
here
we
do
provide
support
for,
but
they
may
be
plugins
that
have
been
built
by
a
partner
or
third
party.
B
B
C
You
could
imagine
if
you,
if
you
yeah,
yeah
sorry
to
interrupt
so
there
are
two.
There
are
multiple
cases,
a
few
cases
which
coming
in
my
mind,
one
case
like
in
in
any
full
stack
product
engineering
where,
where
there
are
tens
of
micro
services-
and
they
are
talking
with
the
web
web
uis
or
mobile
uis.
But
that
is
closely
because
the
expose
the
thing
exposed
to
the
public.
It
is
the
ui.
C
And
yes,
that
is
good
what
you
have
shown
it
is
if
we
can
have
better
management
of
apis,
better
management
of
you
know
course,
or
multi
course,
or
anything
which,
which
can
which
can
give
trouble
to
api.
If
you
don't
manage
it,
whether
it
is
internal
as
internal
impleme,
internal
management
or
whether
is
any
threat
which
is
coming
from
out
outside.
C
So
one
part,
which
I
mean
one
thing
you
have
explained
well
that
how
it
can
be
automated
managed
well
when,
when
we
have
sets
of
micro
services,
but
this
automation
can
help
to
have
better
better,
better
development
within
a
team
and
all.
But
another
question
is,
I
would
like
to
know
you
know.
Let's
say
we
don't.
Let's
say
we
can
manage
by
ourselves.
Management
is
going
point
how
it
can
help
us
on
those
application
where
we
have
exposed
ui
to
the
public.
B
Yeah
and
it
because
the
nature
of
the
audience
and
the
sort
of
problem
space
you
you're
looking
at,
would
differ
from
apis
that
you're
exposing
publicly
and
have
developer.
You
know
sort
of
human
level
developers
that
are
building
applications
to
access
it,
and
so
you
care
a
lot
about
there
about
identification
and
authorization,
authentication
you
care
about
the
quality
of
traffic,
so
if
they're
paying
customers,
they
might
get
a
higher
tier
of
quality
of
service
than
free
customers
and
protecting
against
the
sorts
of
threats.
B
Oh,
you
might
be
nailing
down
microservices
with
with
you
know,
access
control,
lists
or
and
saying
you
know,
this
microservice
will
only
ever
talk
to
these
other
three
microservices
rather
than
the
whole
estate.
So
I
understand
what
you're
saying
and
the
I'm
just
saying
that
there's
policies
here
that
will
work
and
be
more
useful
for
for
one
audience
versus
the
other
it
and
and
some
for
both
it
really
depends
on
what
the
problem
space
is.
C
But
it
is
so
it
it
it's
like
do
you
have
some
data
or
some
kind
of
anything
where
you
can,
because,
let's
say
let's
say
in
some
project,
we
do
not
have
that
experience
once
we
deploy
into
the
into
you
know
to
the
public
domain,
so
api
may
get
threats
or
api
may
get
in
trouble,
which
we
have
not
thought
of.
So
because
of
that
that
trouble
means
which
which,
which
is
just
theoretically
we
know,
but
and
or
we
will
know
once
that
you
know
any
threat,
actually
will
come.
B
B
So
saying
look
I'm
going
to
let
you
use
it,
but
I
know
you're
only
playing
with
it
for
the
time
being,
it's
a
sandbox,
so
you
get
a
limited
amount
of
capacity
or
limit
amount
of
throughput
or
calls
to
protect
you
against
an
overload
of
someone
just
suddenly
performance,
testing
or
benchmark
in
your
api.
When
they're
not,
you
know
a
valid
customer
for
you,
as
well
as
allowing
paying
customers
a
higher
level
of
customers
to
be
able
to
interact
with
it,
and
then
you
will
see
there's
some
threat
protection
policies
here.
B
B
What
I
showed
you
here
was
just
a
piece
of
the
broader
connect
platform.
We
saw
service
hub
runtime
and
we
talked
with
just
the
gateway
component.
Obviously,
there's
also
this
the
support
for
mesh
and
support
for
kubernetes
ingress
controllers,
and
I
did
this
all
on
my
laptop.
I
didn't
show
you
any
of
this
running
with
the
terraform
script
or
any
of
this
running
on
any
of
these
cloud
providers,
but
that's
part
of
one
of
the
future
meetups
that
we
will
have.
B
We
talked
a
little
bit
about
automation,
and
so,
as
you
know,
we
saw
before
there's
apis
and
command
line
tools
to
give
you
full
end
to
end
automation
in
a
native
and
natural
fashion,
and
we
call
that
api
ops
and
we
we
have
a
really
strong
opinion
and
differentiation
around
how
you
should
use
ap
ops,
to
enable
automation,
and
so
with
that,
go
and
fill
your
boots.
You
can
sign
up
for
connect
for
free
and
get
using
it
and
playing
with
it
and
do
much
of
what
I've
shown
you
today.
B
We
have
a
new
tier
called
plus,
which
we've
just
released,
where
you
can,
if
you
want
to
start
using
it
for
real
inside
of
your
organization,
swipe
your
credit
card
and
get
busy
with
it
and
pay
as
you
go,
and
then
for
organizations
that
want
to
know
more
and
thinking
about
using
some
of
this
sort
of
technology,
more
strategically
and
broadly
across
your
organization
come
and
have
a
chat
to
us,
because
there's
lots
to
talk
about
so
unless
there's
any
other
questions,
I
know
we're
at
the
time.
Thank
you
very
much.