►
From YouTube: AppDynamics’ Journey to Microservices [Keynote Part 9]
Description
Kong CTO and Co-Founder Marco Palladino is joined by AppDynamics Senior Director of Architecture Ty Amell as they discuss AppDynamics’ journey from monolith to microservices and the cultural change that had to occur to support it.
Learn more about Kong: https://konghq.com/
A
B
So
epidemics
is
an
application
performance
monitoring
company
we
got
bought
by
Cisco
roughly
two
and
half
years
ago.
I
think
so
we're
all
about
storing
metrics
in
status
of
your
apps,
giving
you
a
flow
of
how
data
is
going
around
and
we've
got
a
unique
setup
where
we
have
a
SAS
environment,
but
also
on
Prem.
So
we
work
with
some
of
the
largest
banks
in
the
world
and
they're,
obviously
more
on
Prem.
B
They
have
an
adoptive
cloud
as
much,
but
then
we've
got
some
of
the
biggest
entertainment
companies
that
are
all
cloud,
and
so,
as
we've
gone
from
a
monolith
to
micro
services
and
this
transition
from
on
Prem
to
cloud,
we
had
a
lot
of
challenges
around.
How
do
we
do
that?
How
do
we
have
discoverable
api's?
How
do
we
manage
the
traffic
inbound?
How
do
we
secure
all
that
data
because
we're
storing
data
from
you
know
hundreds
of
thousands
of
customers?
B
So
epidemics
has
a
concept
of
agents
right,
so
we
install
agents
within
applications
and
servers.
We've
got
millions
of
those
coming
into
our
SAS
environments
and
they're
talking
every
15
to
30
seconds
I'm,
getting
config
information
and
other
things,
and
then
we
have
metrics
coming
in
and
metrics
coming
in
is
roughly
I
believe
around
300
to
350
100
million
metrics
in
minutes.
So
it's
pretty
large
volume
not
to
mention
the
UI
traffic
and
everything
else.
B
B
So,
like
I
said,
we
we're
a
monolith,
that's
kind
of
how
it
was
done.
Ten
years
ago
when,
when,
after
you
started
and
as
we
broke
broke
apart
the
monolith
into
microservices,
we
really
needed
a
way
to
understand
the
services
that
we
were
building
and
understand
where
traffic
was
coming
in
and
how
to
route
it.
Ssl
termination
all
that
kind
of
stuff,
I've
previously
done
that
at
PayPal
and
I
knew
coming
in
that
we
didn't
want
to
build
something
like
that.
B
Try
to
do
it
on
our
own
and
you
and
I
have
known
each
other
for
a
long
time
and
so
I
reached
out
to
you
guys
and
wanted
you
to
come
in
and
kind
of
educate
the
team
on
what
your
ability
and
why
it
was
such
a
powerful
tool
for
us-
and
you
know
we
could
talk
about
all
the
kind
of
technical
reasons
for
doing
it.
But
the
discoverability
alone
is
important
for
us
and
we
can
use
the
same
solution,
whether
it's
on
premiere,
whether
it's
on
says
yeah,.
A
A
So
so
you
know
we
talk
a
lot
about
transitioning
technology,
the
monolith
to
micro
services,
but
how
does
the
organization
have
to
change
to
support
this
new
way
of
building
software,
because
it's
not
just
technology
is
also
the
people
the
operations?
And
how
does
that
change,
especially
in
the
in
a
specific
case
about
dynamics,
yeah.
B
You
know
it's
it's
ok
to
fail,
but
fail
fast,
and
so
let's
get
things
out
faster.
Let's
not
do
these
six
months
releases,
let's
get
down
to
six
weeks,
and
then
you
know
three
weeks
and
then
a
week
and
then
you
know
the
goal
is
to
get
every
day.
But
it
is
a
it's
a
massive
cultural
shift.
People
are
so
used
to
doing
it
one
way
or
another
and
to
kind
of
switch
that
and
turn
on
their
head.
It
takes
a
little
while
to
get
used
to
that.
A
B
I
would
say
we're
not
done
yet.
First
of
all,
I've
seen
it
done
at
really
large
companies.
We
did
it
at
PayPal,
we're
doing
it
at
a
dynamic
and
honestly,
all
of
our
customers
are
even
the
banks
are
starting
to
do
that,
which
tells
me
we've
finally
hit
that
escape
velocity
with
that,
because
the
banks
are
usually
the
slowest
to
move.
B
So
it
takes
our
customers
a
really
long
time
to
upgrade
to
new
versions,
again:
banks,
telcos
they're,
deploying
maybe
two
or
three
times
a
year.
So
we
have
to
support
our
different
versions
of
our
software
for
18
to
24
months.
So
it's
a
very
long
transition
to
this
and
I
would
say
we're
probably
about
halfway
there,
but
it's
it's
a
fun,
exciting
way
to
kind
of
change
and
disrupt
a
company.
What.
B
I'm
gonna
say
the
same
answer
for
both
questions
and
now
and
I'll.
Explain
why
one
I
think
in
the
in
the
very
beginning,
we
sprinted
and
didn't
talk
about
that
foundation,
changes
that
need
to
happen
within
your
services
and
foundational
services
like
Kong
and
service
meshes
and
things
like
that.
We
kind
of
sprinted
to
just
doing
services,
and
then
we
said,
oh
crap.
How
do
we
do
logging?
How
do
we
do
API
gateways?
How
do
we
do
secrets
in
this
new
world?
B
So
we
made
that
mistake
very
early
on,
but
I
think
one
of
the
best
things
that
we
did
was.
We
realized
it
and
we
kind
of
pushed
pause
and
we
went
back
and
we
said
okay,
we're
making
a
mistake.
Let's
omit
that
we're
making
a
mistake,
let's
learn
from
it.
Let's
iterate
quickly
and
now
we've
gone
back
and
said:
okay,
here's
how
we're
gonna
do
secrets,
here's,
how
we're
doing
our
IP
gateway
we're
partnering
with
Kong.
B
You
know:
here's
how
we're
doing
our
service
messages,
here's
how
we
were
doing
our
deployments
and
here's,
how
we're
doing
our
deployments
on
pram
and
in
SAS,
because
on
Prem
we
can't
run
anything
as
so
kubernetes
and
all
the
fun
stuff
that
we
get
to
do
in
the
cloud,
isn't
an
option
for
us
on
Prem.
So
it's
finding
these
platforms
in
these
situations
where
we
can
adapt
and
abstract
away
from
these
platforms,
and
so
our
developers
don't
have
as
big
of
a
problem
developing
for
on
Prem
versus
versus
ass
in
the
micro
services
world.
What.
B
B
B
I
think
to
talk
to
these
huge
banks
that
have
scale
that
you
know
most
of
us
don't
actually
comprehend
in
terms
of
the
money
that's
flowing
through
the
systems
and
talk
to
them
about
how
are
they
updating
and
how
are
they
becoming
more
relevant
and
moving
away
from
the
mainframes,
the
bare
metal
machines
and
how
are
they
moving
to
kubernetes
or
OpenShift
or
whatever
they're
moving
to
internally?
And
how
can
we
help
them?
And
so
app
dynamics
is
in
a
really
interesting
position.
B
I
think
because
we're
there
we've
got
their
applications
instrumented,
but
then
they're
coming
to
us
and
being
like.
How
do
we
do
this?
How
do
we
transform
to
a
digital
company?
How
do
we
transform
and
act
more
like
a
startup,
so
as
one
of
the
main
architects
I
get
to
have
all
these
fun
conversations
with
them?
There
are
massive
systems
and
and
how
they're
transforming
to
be
more
cloud
and
be
more
agile.
B
So
that's
that's
pretty
fun
for
me
and
I
think
it's
gonna
unlock
and
we're
starting
to
see
it
with
some
banks,
but
a
lot
more
user
friendly
and
better
interactions
with
some
of
these
legacy.
Companies
that
honestly,
their
customer
service
and
their
interaction
with
our
end,
customers
has
has
been
lagging.
A
B
It's
all
across
the
board.
You
know
you
do
see
some
bigger
companies
where
the
executives
are
more
trusting
the
technology
leaders,
but
a
lot
of
times.
We
do
find
that
you
know
there's
technologists
that
are
trying
to
do
something,
and
one
of
the
questions
is,
is
how
do
I
convince
my
so
and
so
boss
or
how
do
I
convince
the
person
above
me?
But
this
is
the
right
thing
to
do
and
a
lot
of
times
we
can
use
our
our
metrics
and
our
flow
map
and
kind
of
explain
to
them.
B
Things
that
are
that
are
easy
and
safe
to
move,
and
you
know,
would
be
a
greater
performance
and
talk
about
honestly,
our
dream
and
how
it's
made.
You
know
their
lives
easier
and
better,
and
you
know,
we've
helped
their
careers
in
all
honesty
and
once
they
kind
of
understand
that
impact
there
have
a
better
story
for
their
execs
and
a
better
alignment.
It's.
B
Because
I
mean
I
had
a
large
enterprise
like
that,
like
you're
putting
your
neck
on
the
line
a
little
bit
right,
you
are
you're
basing
your
career
on
this
transition
and
if
you
succeed,
you're
going
to
be
rewarded
and
you're
gonna
be
promoted
and
you're
going
to
get
your
dream
job.
If
you
fail,
that's
not
so
much
fun,
so
they
are
definitely
put
in
putting
their
necks
on
the
line
a
little
bit,
but
we've
seen
you
know
a
lot
of
our
customers
really
grow
within
there
within
their
companies
and
being
that
agent
of
transformation.