►
From YouTube: ArgoCon '21 - Scaling Kubernetes across BlackRock’s Aladdin Platform (Ryan Umstead & Mike Bowen)
Description
BlackRock has always pushed technical boundaries and innovation to deliver asset management capabilities to clients through the investment and risk management platform, Aladdin. As part of this journey, we realized the power of container-native development and landed on Kubernetes as a way to abstract infrastructure concerns from the application tier and future-proof the platform for the cloud. In this talk, Ryan and Mike will discuss BlackRock’s journey of releasing Kubernetes, partnering with external GitOps practitioners, and building BlackRock’s Aladdin Kubernetes Platform.
A
Ryan
is
a
senior
engineer
in
the
aladdin
product
group
at
blackrock.
Blackrock
is
the
world's
largest
investment
management.
Company
and
aladdin
is
blackrock's
end-to-end
investment
and
risk
management
technology
that
they
use
internally
to
run
investment
businesses
and
deliver
externally
to
other
financial
institutions
in
ryan's
role.
A
Ryan
graduated
kumari
from
drexel
university
in
2015,
with
a
bs
in
computer
science
and
a
minor
in
business
administration
mike,
is
a
senior
principal
engineer
on
the
aladdin
project
and
the
director
of
open
source
at
blackrock
mike
has
led
the
blackrock
open
source
program
office
where
he
partners
with
other
members
of
the
tech
ecosystem
and
representatives
from
the
blackrock
legal
and
compliance
team
to
deliver
on
those
open
source
goals.
The
open
source
project
maintains
open
source,
license
compliance
through
governance
and
oversight.
A
A
B
Thanks
sarah,
I
appreciate
the
intro.
I
don't
have
a
tremendous
amount
to
say
anymore,
hello,
cute,
argo
khan,
very
happy
to
be
here
and,
as
sarah
just
went
over,
who
both
ryan
and
I
are
and
what
we
focus
on.
We
really
want
to
jump
right
to
our
cloud
native
journey,
because
that
is
far
more
exciting.
As
a
firm.
B
As
sarah
noted,
we
work
at
blackrock,
we
work
within
the
aladdin
product
group
and
we
work
on
a
platform
called
aladdin
and
just
to
reiterate,
aladdin
is
a
is
delivered
as
a
platform
to
our
clients.
It's
a
centralized
source
of
data
helps
them
make
better
and
more
informed
investment
decisions,
effective
risk
management,
and
we
do
it
at
very,
very
high
operational
scale
and
efficiency.
So
as
we
look
to
evolve
that
to
the
next
iteration,
how
do
we
increase
our
scale?
How
do
we
go
faster?
How
do
we
do
better?
B
How
can
we
have
higher
velocity
and
more
operational
scale
for
our
clients
as
a
fiduciary
and
to
deliver
for
them
the
business
value
that
they
need
to
make
those
better
investment
decisions
every
day?
So
we
just
talked
about
aladdin
and
where
I
want
to
go
now
is
like
our
journey.
Every
journey
and
story
begins
somewhere,
and
I
think
all
these
pictures
on
this
slide
are
going
to
be
very,
very
familiar.
B
We
wanted
to
standardize
and
adopt
on
containers
we
we
knew.
That
was
a
good
idea.
We
knew
it
was
an
amazing
innovation
and
we
started
leaning
into
it
pretty
hard
around
development
environments,
repeatable
environments.
How
can
we
test
locally
and
in
controlled
environments
just
like
we're
testing
in
production,
so
we
have
fewer
bugs.
We
have
higher
efficiency,
higher
velocity
fewer
dev
cycles
as
we
go
through
it.
B
The
these
are
the
trail
map
is
kind
of
interesting
because
it
has
dragons
on
it
right
and,
as
you
start
defining
this
trail
map
for
yourself
and
where
you
pick
up
what
you
need.
Containers
cicd
things
like
argo
cd
and
you
start
moving
and
matriculating
through
both
the
landscape
and
the
trail
map.
We
kind
of
knew
that
we
had
to
define
our
own.
We
knew
that
cloud
native,
compute
foundation
and
cloud-native
patterns,
practices
and
conventions
could
help
us,
but
maps
are
only
so
helpful
right.
B
There
are
a
lot
of
options,
so
we
knew
that
we
needed
to
lean
into
things
like
open
source
and
communities
and
other
like-minded
people
in
the
industry
and,
specifically
the
cloud
native
compute
foundation.
This
this
linux
foundation,
collection
of
just
great
human
beings
that
helped
us
move
from
what
got
us
here
to
how
we're
gonna
get
there.
B
Now
again,
I
said:
there's
dragons:
this
is
an
oregon
trail,
it's
a
it's
a
it's
a
tough
hill
to
to
climb,
but
but
we're
climbing
it
and
this
this
arc
kind
of
shows
where
we
started
so
back
in
2014.
2015
containers
are
great.
We
love
them.
We
want
to
lean
into
it
a
whole
lot
more.
We
started
experimenting
with
things
like
mesos
and
pre-version,
one
of
kubernetes
and
even
had
both
of
those
deployed
in
production
context,
slash
lab
context
for
us
to
get
better,
smarter,
faster
with
them.
B
We
started
understanding
more
about
it
and
needing
to
get
more
enterprise
grade
with
it,
because
we
are
a
highly
regulated
company
and
we
are
audited
regularly
by
those
regulations,
our
clients
and
and
so
on.
So
we
decided
to
land
on
openshift
as
our
first,
our
next
logical
step
as
a
as
a
partner
in
red
hat,
who
had
a
kubernetes
distribution,
had
support,
and
we
could
continue
to
learn
more
with
partners
and
we
segued
directly
into
a
case
study
with
the
cncf.
B
How
do
we
deploy
kubernetes
leveraging
openshift
in
100
days
for
just-in-time
machine
learning
and
data
science,
environments
for
our
quants
and
desk
side,
traders
and
investors
that
all
went
really
really
well?
We
found
out
quickly,
we
needed
oh,
a
venting
framework.
B
C
Like
mike
said
since
blackrock
is
a
highly
regulated
financial
firm,
it
was
easier
to
build
than
go
through
the
process
of
onboarding
external
products.
We
also
had
many
unique
use
cases
which
didn't
always
align
with
roadmaps,
so
blackrock
created,
homegrown,
proprietary
software.
We
built
release
systems,
messaging
cues
tool
chains
and
even
even
an
orchestration
system.
C
Some
of
you
might
get
a
quick
laugh,
but
we
call
our
orchestration
system
mi6,
server
and
sam,
where
sam
can
stand
for
two
things:
secret
agent,
man
or
server
agent
manager,
depending
on
who
you
ask,
and
then
you
have
mi6
server,
managing
the
secret
agent
men
or
the
double
o's
believe
it
or
not.
The
orchestration
system
has
many
similarities
with
kubernetes,
though
nowhere
near
as
rich
feature
set
or
as
robust.
C
C
C
C
In
comes
akp,
the
aladdin
kubernetes
platform,
though
sometimes
it
is
called
a
kubernetes
platform
because
it
actually
started
supporting
more
than
just
aladdin
and
took
on
some
of
other
blackrock's
products
to
quote
phil
carlton.
There
are
two
hard
things
in
computer
science,
cash
invalidation
and
aiming
things
and
we're
a
struggle,
we're
still
struggling
to
name
things
to
this
day,
as
we
were,
building
out
the
platform,
we
formed
a
laundry
list
of
requirements.
C
We
needed
a
delivery
system,
a
messaging
system
that
worked
in
and
outside
of
kubernetes,
telemetry
top
to
bottom
and
everything
in
between
disaster
recovery
in
a
cloud-native
way
that
cut
against
how
we
do
it
are
for
aladdin.
Currently,
a
highly
available
registry
doesn't
matter
how
many
clusters
we
have
if
we
can't
pull
an
image
policy
management,
everything
from
information,
security
to
blackrock's
policies
that
will
be
audited
against
all
these
requirements.
C
C
C
C
It
also
allows
developers
to
ship
code
and
not
containers
and
then
finally,
the
coupe,
the
the
core
platform,
has
an
extendable
interface
that
developers
can
build
upon
and
then
the
highest
row
are
segments.
This
makes
up
our
business
applications,
some
of
which
make
up
aladdin
and
others
that
just
make
up
other
blackrock
products.
C
C
Desired
state
make
it
so
so
this
is
where
we
got
to
in
about
a
year
and
a
half
174
clusters
a
little
bit
under
1800
applications
and
100
kit
repositories,
and
if
I
were
to
refresh
that
graffana
dashboard
right
now,
I
guarantee
those
numbers
have
changed.
The
clusters
and
applications
deployed
via
argo
cd
are
all
green
field.
This
doesn't
even
touch
our
bare
metal
estate,
where
I
was
talking
about
mi6
server
and
sam.
C
And
then,
finally,
what's
still
in
the
works,
we
still
have
a
lot
to
do
some
of
the
other
auto
product
projects
that
we
are
looking
to.
Leverage
are
argo
cd,
autopilot
to
help
onboard
our
bare
metal
estate
and
have
an
opinionated
onboarding
process
and
help
us
get
to
a
quicker,
easier
button
application
sets.
I
already
showed
you
1800
argo
cd
applications
and,
as
we
onboard
the
aladdin
estate,
we're
going
to
get
to
even
more.
We
need
that
factory
of
argo
cd
applications.