►
From YouTube: Moving from Legacy Infrastructure to the Cloud in a Government Organization - Chris Carty
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
Moving from Legacy Infrastructure to the Cloud in a Government Organization - Chris Carty, City Of Ottawa
Cloud native tech isn’t just for start-ups. But, if you’re in a government organization looking to go cloud native, you can expect to face extra challenges. How can you select the best tools that will work with the processes you already have? What new skills are needed? How do you train staff? How to get anyone to actually use the framework once it’s in place? How to even start? The City of Ottawa (yes the capital of Canada) was an organization that started applying DevOps practices just a few years ago. It now has a Kubernetes platform with fully automated CICD pipelines being used by multiple teams and growing. Using The City as a case study, we will examine the common issues faced by government organizations and how The City developed workable solutions on its cloud native journey.
https://sched.co/UabP
A
B
All
right,
so,
thank
you,
everybody
for
coming
out
to
see
my
I
talk
today.
It's
first
time,
speaking
at
cube
Con,
so
bear
with
me
as
I
work
out
some
the
jitters
and
just
want
to
kind
of
also
get
a
bit
of
a
pull
the
audience
who
here
also
works
for
a
government
organization
of
some
kind
awesome,
there's
more
of
you
than
I
was
expecting
how
about
so.
Who
here
is
just
starting
out
on
their
journey
to
kubernetes,
whether
your
government
or
private
awesome.
B
So
this
is
kind
of
structured
to
be
that
presentation
to
kind
of
give
that
helping
hand
and
hopefully
give
you
some
tools
on
your
journey
and
selecting
the
tools
and
dealing
with
how
to
you
know,
manage
all
those
vendors
at
the
vendor
hall
with
what
to
pick
and
where
to
put
them
what
things
so
without
further
ado,
we'll
get
on
with
the
presentation.
Actually,
one
thing
I
want
to
do.
Do
you
mind
if
I
take
a
quick
photo
just
to
prove
that
people
actually
showed
up
to
this?
B
Oh
great
I'll.
Take
that
as
a
yes
thank
you,
I
get
something
to
show
off
back
home.
So
that
being
said
just
to
give
a
little
context
for
the
city.
As
most
of
you
probably,
you
know,
government
Canada,
that's
their
home
base,
we're
wedged
over
here
in
between
Montreal
and
Toronto.
So
too
much
cooler.
Cities
normally
go
there
for
fun,
but
we're
getting
better.
It's
good.
B
B
So
there's
that
the
full
spread
and
to
give
you
again
to
kind
of
give
you
the
scale,
the
city
we're
big.
We
can
fit
most
in
major
cities
and
in
Canada
within
our
city
limits.
Despite
that,
we
only
have
a
population
of
well.
We
just
hit
a
million
people,
so
the
overall
Square
area
you're
looking
at
about
1,800
square
miles.
B
We
did
this
by
taking
over
a
bunch
of
cities
in
about
2001,
so
I
was
at
11
cities
and
townships
that
amalgamated
to
form
this,
this
heart-shaped
City,
which
is
kind
of
great
it's
also
delightfully
representative
of
Canada.
We
have
a
large
rural
area
and
a
small
urban
area,
that's
clustered
along
the
Quebec
border,
so
much
like
we
cluster
along
the
American
border.
B
So
with
that
brings
a
lot
of
complexity,
so
we
have
to
support
all
the
federal
workers
that
live
in
the
area,
cross-border
travel.
We
have
a
lot
of
urban
and
rural
concerns
that
we
also
have
to
address
and,
of
course,
maintaining
value
infrastructure
for
all
this.
So
we
have
libraries,
paramedics,
accountants,
a
ton
ton
of
different
services
that
we
need
to
provide
to
our
constituents
so
pull
this
up.
So
we
have
17,000
employees
about
300
of
which
are
an
IT
and
about
a
hundred
of
those
are
developers.
The
rest
are
either
infrastructure
business
business
support.
B
B
The
staff
to
application
ratio
is
a
little
off,
so
we
have
one
team
that
they
have
six
people,
but
they
have
thirty
applications
to
support
across
the
number
of
different
business
lines
so
a
day
to
day
and
the
life
of
one
of
our
developers
can
be
quite
quite
unique
and
challenging.
So
so
here
we
go.
B
We
have
about
400
applications
that
we
need
a
support
of
plus
and
it's
you
know
it's
off
the
shelf,
maybe
big
things
like
ArcGIS
or
custom
homes,
groaned
Java
applications
are
dotnet
and
we
also
have
a
large
amount
of
Perl
applications,
which
is
also
interesting.
So
where
did
we?
Where
discriminate
is
kind
of
fit
into
this?
So
we
wanted
to
more
us
we're
coming
to
the
end
of
a
life
of
a
lot
of
our
old
UNIX
servers
and
we
can
migrate.
Those
and,
of
course,
cloud
is
a
big
part
of
that.
B
So
kubernetes
fits
into
our
landscape
is
nice.
You
know
transition.
We
can
move
things
on
Prem
to
containers
and
then
eventually
port
over
to
a
zoo,
which
is
our
cloud
provider
of
choice.
It
gives
us
a
nice
safe
area
to
experiment
adopt
and
then
we
can
move
up.
So
that's
that's
why
we
chose
kubernetes
so
I'm,
assuming
most
of
you,
have
decided
you're
already
going
to
go
to
kubernetes
every
kind
of
working
get
going
so
that
how
do
you
get
going?
First,
spot
I
looked
it's
the
sign,
I've
seen
SEF
landscape.
B
Now,
first
time,
I
looked
at
that
I
was
kind
of
scared.
Anyone
else
had
the
same
kind
of
like
just
turned
close
to
my
browser
go
away
yeah,
so
we
had
decided.
We
wanted
to
go
down
that
red.
We've
done
a
bunch
of
small
demos.
You
know
spin
up
the
cube
a
DM
server,
but
we
needed
to
be
a
little
bit
more
formal.
We
knew
we
didn't
have
the
staff
to
support.
You
know
growing
her
own,
so
we
had
to
go
with
something
that
was
enterprise
grade.
B
So
look
at
that
the
guys
scared
went
back
to
the
drawing
board.
We
created
a
project
team
of
creative,
but
with
a
bunch
informed,
so
the
cross-functional
team,
so
we
had
security
at
the
table.
Infrastructure
applications
some
architects
and
we
can
worked
breaking
things
down
into
smaller
chunks.
So
this
is
what
we
did.
You
don't
have
to
necessarily
follow
that
it's
just
a
way
to
four.
It
was
a
way
for
us
to
kind
of
logically
break
things
down
to
stages,
so
out
of
the
gates,
the
first,
the
big
thing
for
us
was
finding
kubernetes
distribution.
B
So
this
was
about
year
year
and
a
half
ago,
maybe
a
little
bit
more
since
then,
the
CN
CF
has
put
out
there
Trail
Minh
Trail
Road,
so
it
has
a
very
similar
purpose.
This
categories
are
a
little
different.
It's
focused
on
the
CNC
F
tools,
not
necessarily
greater
landscape,
but
it's
another
nice
tool.
If
you
want
to
kind
of
compartmentalize
things
and
build
out
from
there.
So
as
I'm
saying
the
last
slide,
our
first
point
of
order
was
think
of
distribution.
B
At
the
time
a
lot
of
so
I've
been
thinking
about
the
difference
between
this
coupon
and
past
cubed
con
just
in
terms
of
themes.
So
last
Cape
Cod
I
think
have
every
major
vendor
was
going
on
their
conformant.
You
know
there
there's
the
kubernetes
conformant
test,
so
that
was
everyone
was
moving
that
way,
which
is
awesome.
So
all
these
here
are
now
kubernetes,
compliant
or
kubernetes
conformant
distributions.
So
you
can
take
your
kubernetes
code
roll
off
through
which
is
great
for
not
getting
locked
in,
but
year
and
a
half
ago
it
wasn't
as
as
prevalent.
B
So
we
had
to
be
a
little
bit
more
selective
in
terms
of
how
we
did
that.
So
we
broke
things
down
a
little
bit
more
in
terms
of
our
project
objectives,
so
we
wanted
vendor
agnostic,
we've
been
burned
in
the
past.
By
going
with
a
specific
technology
or
vendor,
then
the
technology
dies
off
and
we're
stuck
holding
holding
on
to
that.
So
we
wanted
the
ability
to
be
able
to
move
effectively.
You
know,
have
more
ownership
and
control
over
our
destiny
instead
of
waiting.
B
For
you
know
someone
else
to
release
a
patch,
so
kubernetes
gave
us
that,
so
we
focus
why
we
need
open
source
kubernetes.
We
knew
we
wanted
to
roll
on
Prem,
as
well
as
in
the
cloud
reason
for
that.
Some
of
our
workloads
in
laboum
were
sensitive
from
a
security
point
of
view,
so
we'd
want
to
keep
those
on
house
as
well
as
extent
to
the
cloud
for
a
more
public
facing
non-sensitive
for
that
workloads.
B
We
also
want
to
focus
on
multiple
clusters,
so
instead
of
having
a
large
monolithic
cluster,
having
big
acity,
to
make
small
little
clusters
to
reduce
the
blast,
radius
being
able
design
our
special
snowflake
cluster
for,
say,
pci
applications,
things
like
that,
where
security
isn't
necessarily
mainline
and
the
big
of
the
other
big
one
is.
We
need
an
enterprise
level
support
because
we
didn't
have
the
knowledge
in-house.
We
were
working
on
it,
we
still
are,
but
we
just
couldn't
dedicate
a
full
team
to
know
Indians
and
outs
of
everything
kubernetes.
So
this
would
be
something.
B
This
was
something
that
would
be
able
to
help
us
and
of
course
he
would
get
help
get
management
on-site.
Knowing
that
you
know,
if
you
know,
someone
like
me
were
to
win
the
lottery,
they
would
be
able
to
fall
back
on
something
and
the
other
big
one
at
that
point
in
time
was
our
base.
Our
back
wasn't
as
prevalent
as
it
is
today.
So
there's
a
couple
distributions
that
didn't
support
that.
So
that
was
one
of
our
requirements.
B
B
So
we're
still
just
grasping
you
know
kubernetes
or
all
these
other
things
on
top
that
added
a
little
bit
more
complexity,
so
started
asking
around.
Well,
what
are
we
using
for
internal
storage,
provisioning?
It
turned
out,
it
was
NetApp.
Netapp
has
a
level
you'll
plug
in
or
a
tool
called
trident,
which
allows
a
hook
in
tune
it
up
and
use
that
as
your
persistent
storage
provider,
so
that
let
us
take
advantage
of
technology
that
rehab
we
had
already
invested
in
in
a
rehab
the
knowledge
and
house.
B
C
B
A
lot
of
tools
that
a
lot
organizations
may
already
have
in
place,
so
you
know
it's
really
good
idea
to
try
to
leverage
what
you
already
have
also
trials,
they're,
really
easy
to
come
by.
As
most
of
you
have
already
found
out
in
the
vendor
Hall,
you
know
you
just
ask
and
you'll
be
able
to
get
a
key
and
be
able
to
try
something
see
how
it
fits
in
your
environment,
and
this
also
from
a
government
point
of
view
who
here's
how
to
go
through
an
RFQ
or
request
from
proposal
requests
for
CRO
code.
B
It's
not
fun
right.
It
takes
a
really
long
time
apologize
to
any
vendor
that
I've
dealt
with
in
here.
We've
done
a
couple,
multi-year,
RFPs
or
engagements.
So
you
know
sometimes
you
talk
to
something
like
oh
come.
You
have
this
wrapped
up
by
the
end
of
the
month.
Yeah
not
gonna
happen
so
being
able
to
try
things
out
it
lets
you
we
found
to
let
you
or
let
let
us
write
a
more
detailed
proposal,
so
it
doesn't
just
go
to
so.
You
get
a
better
idea
of
what
you
actually
want.
B
So
many
things
go
to
procurement:
you
get
a
better
product
as
you've
narrowed
it
down
to
what
needs
you
actually
have.
Then
there's
also
they're
focusing
on
processes.
We've
heard
a
lot
of
this
and
DevOps.
You
know
people
over
process
process
over
tools.
I
tend
to
think
of
this.
Can
kubernetes
DevOps
landscape,
its
giant
build-it-yourself
puzzle.
All
these
pieces
are
swappable
which
is
really
fun.
So
when
you
start
focusing
on
patterns
instead
of
over
what
a
specific
tool
does,
you
can
really
do
some
fun
things.
B
This
thinking,
this
thinking
lettuce
towards
get
ops,
which
is
a
lot
of
funds
who
anyone
here
here,
flux
or
agro
or
Jenkins,
X.
Okay,
a
few
check
them
out,
they're,
really
really
really
good.
So
we
have
a
few
teams
that
you're
using
different.
You
know
get
sources,
so
github
get
lab
and
is
there
DevOps
and
we
needed
something
kind
of
fit
in
between
all
of
those
the
gaps
model.
You
store
your
configuration
and
get
and
with
flux
or
our
go,
something
sits
in
your
cluster
and
then
just
pulls
down
and
syncs.
So
it's
rather
neutral.
B
It's
not
a
push
action,
so
it
doesn't
need
our
access
to
your
cluster
on
the
instant
and
the
case
of
azure
DevOps.
You
know
you
see
if
your
code
up
there,
it
doesn't
necessarily
it
can't
go
sorry.
They
can't
necessarily
push
to
you
on-prem
cluster.
So
that's
where
something
like
flux
or
Jenkins
or
Argo
comes
in,
because
it's
a
pull,
pull
based
action,
which
is
really
really
awesome
and
of
course
you
don't
have
to
give
people
cube
CTL,
which
also
a
bonus
try
to
avoid
that
for
production
stuff.
B
So
with
all
that,
so
this
is
just
a
glimpse
of
our
toolset.
So
far,
these
are
some
of
my
favorite
favorite
tools
that
I
have
in
place.
Flux,
open
policy
agent.
If
you
haven't
caught
any
of
the
opah
session
so
far
this
week,
I
think
there's
a
bunch
more
do
check
them
out
for
a
firmer
security
point
of
view.
Any
time
we
bring
up
being
able
to
write
policy
as
code.
For
those
of
you
don't
know,
open
policy
agent
is
a
generic
policy
engine,
so
you
can
write.
You
know
big
example
would
be.
B
You
can
only
pull
in
a
container.
That
comes
from
a
trusted
registry,
so
it
gives
you
a
lot
of
control
over
that
and
you
know
really
helps
get
your
security
team
on
board
and
it
opens
up
a
lot
of
a
lot
of
potential
for
locking
down
and
controlling
what
goes
into
your
cluster.
Of
course,
flux
is
really
good.
It
works
great
with
open
policy
agent.
Linker
D
is
an
awesome
service.
Mesh
sto
is
another
good
one.
That's
you've,
probably
also
heard
of
so
all
that
said
this
is
the
easy
part
assembling
your
your
own
puzzle.
B
Great
hard
part
is:
how
do
you
get
people
to
use
it?
So
we
did.
We
took
the
Builder
build
at
first
and
let
them
come
approach.
We
found
that's,
you
know
kind
of
dangled
the
carrot.
People
will
eventually
come
instead
of
doing
that.
Forced
adoption
find
that
normally
gets
people's
back
up
I'm
using
a
you
know,
working
closely
with
a
few
teams,
letting
them
have
some
success
kind
of
working
out
and
using
kind
of
setting
up
like
champions.
B
B
So
we've
taken
the
approach.
You
know:
we've
done
a
lot
of
bunch
presentations
and
presentations,
senior
management
just
together
the
word
out
get
people
comfortable
with
the
concept
of
moving
from
servers
to
containers,
to
kubernetes
to
DevOps
and
all
that
fun
stuff
to
one-on-one
sessions.
We've
also
found
15
friends,
everyone
loves
them.
If
you
can
get
the
book
from
Microsoft
do
so.
It
breaks
things
down
into
nice.
Little
bites
on
earth
kubernetes
concepts
into
a
nice
bite-sized
format.
B
We've
also
people
are
just
before
I
left
had
a
email
from
my
manager
asking
me
to
set
up
a
calendar,
invite
to
watch
the
kubernetes
office
hours,
something
just
a
little
self-promotion,
I'm
a
panelist
on
that,
and
we
do
a
live
stream.
Every
Wednesday,
a
third
Wednesday
of
the
month,
where
we
take
questions
from
the
crowd
or
Internet
and
try
to
help
hope.
These
are
sorry
help
people
use
kubernetes,
so
acting
kind
of
helping
people.
So
we
try
to
work,
as
my
team
is
focused
on
supporting
that
core
platform.
B
This,
hopefully
solves
that
and
then
from
a
policy
point
of
view,
open
policy
agent
gets
a
lot
of
people
excited
I've,
never
seen
so.
Many
managers
leave
a
meeting
about
kubernetes
with
smiles
on
their
face.
After
seeing
what
open,
open
policy
agent
can
do
and,
of
course,
automating
all
the
things,
so
we
try
it
more
as
when
you
asking
people
to
learn,
kubernetes
and
containers,
it's
bit
of
a
cognitive
load,
you're
dropping
on
them.
It's
like
okay!
Well,
first,
you
have
to
learn.
B
Containers
any
have
to
learn
kubernetes,
they
have
to
learn
distributed
computing,
they
need
an
alert
race.
It's
just
kind
of
eyes
will
glaze
over
eventually
so
to
help
people
adopt
that
we've
tried
to
hide
a
lot
of
the
the
scary
stuff,
but
behind
layers
of
automation.
So
the
onboarding
process
is
pretty
quick,
but
as
people
want
to
peel
back
the
layers
of
the
onion,
they
can
so
everything
is
captain
captain
get
its
open,
so
people
can
see
see
what
doing
so.
B
We're
using
starter
templates
have
summer
security
policies
already
baked
in
we're
looking
to
migrate,
that
to
helm.
Now
that
humphrey
is
out
until
there
is
gone
as
this.
This
helps
and
first
folks
is
also
a
key
part
of
that.
So
one
of
the
nice
features
of
flux
is,
it
will
also
monitor
your
container
registry
and
if
it
sees
a
new
image,
it
pulls
it
down
and
deploys
it
into
that
deployment
and
then
it'll
write
the
change
back
to
get
so
you
have
this
kind
of
full,
automated
cycle,
which
is
really
awesome.
B
So
we've
been
using
this
process
to
migrate
all
of
our
legacy
Perl
and
Java
apps.
Thankfully,
they're
mostly
you
know,
there
are
pre
simple
web
websites,
they're,
not
micro
services,
so
the
pattern
that
we
use
to
build
and
deploy
is
pretty
pretty
similar
and
there's
not
a
lot
of
modifications
so
to
get
people
on
board
it.
It's
a
good
path
to
kind
of
have,
though
this
is
easy
and
kind
of
go
from.
There
build
some
confidence,
and
then
there
you
know
before
long
they're,
starting
to
build
their
own.
B
Things
are
asking
questions
about
what
else
can
I
do,
which
is
awesome?
I,
love
those
questions,
and
so
just
to
give
you
an
example
of
one
of
the
process
and
installation
processes
we
have.
So
this
is
our
when
a
project
github
get
lab
repo,
so
we
have
a
get
Bob
a
CI
em
all
in
there
and
docker
file
in
a
war
file,
so
some
teams
are
still
migrating
from
TFS
to
get
so
that's
another
thing
that
we
have
to
working
on
teaching
everybody.
So
in
this
case
the
dev
can
still
work
with
TFS.
B
They
get
the
war
file
out.
They
drag-and-drop
that
war
file
into
get
lab.
It
kicks
off
a
build
process.
The
poise
the
container
updates
the
ammo
file
for
the
appointment
and
the
second
repo,
so
with
just
that
click-and-drag
they've
deployed
to
kubernetes
and
they
get
an
automated
build
process
which
is
awesome
and
then
also
out
of
that.
You
know
we
got
metrics
from
Prometheus.
You
get
reasoning
with
your
DS
so
to
get
traffic
traffic
metrics,
and
then
we
also
get
per
web
metrics.
So
they
get
this
out
of
the
box.
B
They
don't
have
to
do
anything.
They
have
to
reinvent
those
fifteen-year-old
applications.
It's
just
there.
This
is
really
exciting
for
people
okay.
Well,
now
what
can
I
do?
Okay?
How
do
I
take
advantage
of
this?
So
we
try
to
follow
that
up.
A
lot
of
training
try
to
get
people
interested,
so
we
have
an
access
to
Pearl
sight.
Anyone
else
have
organizations
with
Pearl
sight.
B
Awesome.
The
Nigel
Putin
kubernetes
courses
are
fantastic,
give
them
and
seen
them
already.
They
also
recently
signed
up
for
kubernetes
fundamentals,
so
we're
sending
a
I
think
about
six
people
on
that
to
get
them
skilled
up
and
also
based
on
so
I
I
mastered
by
cuban
agent
min.
So,
based
on
my
experience,
I
recommend
anyone
wants
to
go
further.
The
tasks
on
kubernetes
io
and
if
you
want
to
really
peel
back
the
layers
of
the
onion
Kelsey
Hightower's
kubernetes,
the
hard
way
is
second
to
none
and
that
those
regards
so
kind
of
wrap
it
up.
B
B
We
did
it
on
a
Friday
and
no
no
servers
died
or
pots
for
that
matter,
which
is
awesome,
and
you
know
we
still
have
some
teams
who
are
active,
who
are
we're
a
little
ahead
of
the
curve
in
terms
of
it,
often
in
the
cloud.
So
we
have
some
people
using
app
services
as
well
as
people
are
still
wanting
to
use
VM,
so
we
still
have
a
little
bit
of
evangelizing
in
sales
as
sales
pitching
to
do,
but
a
long
story
short,
not
everything
belongs
in
kubernetes
and
using
app
service
for
a
cloud.
B
Migration
is
a
valid
choice
and
it
can
be
proved
to
be
quite
useful
so
to
look
ahead
for
where
we're
going,
we
should
have
a
bunch
more
Java
applications
served
up
to
the
public,
so
we'll
be
taken
down.
The
Tomcat
apps
are
being
served
on
old,
UNIX
servers,
putting
them
into
containers
all
of
our
Perl
web
forms
are
gonna,
be
hosted
internally.
We're
also
coming
up
to
the
final
stages
of
writing
our
corporate
security
standards.
So
I
was
one
of
the
fun
things
voting
engaging
security
really
early
on.
B
We
started
this
process
of
writing
the
standards.
So
sometimes
security
is
viewed
as
kind
of
like
that.
To
that
roadblock,
you
build
your
application.
They
come
swooping
in
at
the
end
like
no,
it's
not
secure
enough,
so
we've
been
engaged
with
them
to
bake
a
lot
of
that
in
so
there
aren't
those
surprises,
and
it's
been
very
fruitful
and
they've
been
a
very
strong
supporter
of
what
we've
been
doing,
and
it's
been
a
very
it's
been
a
very
beneficial
relationship
in
terms
of
cloud
adoption.
B
We've
been
working
on
just
sorting
out
the
governance
when
you
moving
to
Azur
setting
up
your
service
accounts
or
service
principles,
how
do
you
dole
them
out?
How
do
you
do
automation
around
them?
It's
something
that
is
kind
of
a
broader
organizational
question
that
we're
we're
working
through
right
now
and
we're
also
working
on
increasing
the
automation
tooling,
on
behind
our
platforms.
So
one
of
our
mission,
sorry,
one
of
our
big
business
plans
for
next
year
is
to
automate
everything
in
our
kubernetes
stack
from
the
server's
through
to
firewalls
and
networking.
B
So
it's
a
big
project,
but
we're
hoping
to
have
that
done
sometime
the
next
calendar
year.
So
right
now,
it's
lightly.
Automated,
but
we
want
to
go
full
automation,
which
would
be
awesome,
so
some
key
takeaways
we
found
setting
objectives
was
really
helpful.
Working
through
that
pathway
that
you
developed,
based
on
what
your
requirements
are
either
developed
or
just
focus
on
deployment
patterns
so
patiently
spread
awareness
use,
find
those
carrots
that
you
can
dangle
to
get
people
to
come
over
and
automate
as
much
as
you
can.
B
It
is
awesome,
just
be
able
to
not
have
downtime
or
copy
and
pasting
files
over
from
one
server
to
the
other,
for
different
production
regions
and,
of
course,
keep
come
back
to
cube
con.
It's
I
learned
a
lot
every
time,
I
come
here
and
I
hope.
Everyone
here
has
to
kind
of
help
out.
I
put
I
made
a
github
repo
with
a
bunch
of
kubernetes
resources.
Some
deployment
files
sum-sum
get
up
style
stuff,
as
well
as
a
list
of
educational
resources
and
other
tooling.
B
D
B
B
D
B
Harder
that's
been
taking
some
time.
We
have
a
couple
teams
who
are
very
so
the
team
that
I
mentioned
earlier.
If
we're
they
have
like
30,
apps,
they're,
very
operationally
driven.
So
we
have
a
member
of
our
team
who's
a
little
bit
more
programming
oriented,
so
he's
been
working
more
or
less
embedded
with
them
to
build
out
or
migrate
their
tool
to
containers
to
give
them
a
help
that
helping
hands.
So
when
they're
ready
this,
which.
D
B
E
I'm
sure
you
found
in
your
experience
and
as
you
touched
on
a
little
bit
here,
one
of
the
hardest
things
to
change
in
all
of
this
is
actually
the
culture
government's
are
notoriously
change.
Averse,
yes,
I
was
just
wondering
if
you
could
go
a
little
bit
deeper
into
like
any
specific
examples.
You
have
or
things
that
you
you
and
your
team
did
to
help
for
lack
of
a
better
term,
the
frozen
middle,
and
you
know
the
bureaucracy
kind
of
overcome
their
fear
to
make
it
as
changes
in
transition
to
do
to
do
this.
Yeah.
B
So
maybe
was
working
more
for
my
grassroots
approach,
so
the
core
team
that
I've
been
working
with
has
been
made
up
of
infrastructure
and
some
networking
and
security,
so
working
with
them,
showing
what's
possible
and
having
their
input.
I
gots,
the
big
one
like
yeah
people,
are
risk
averse,
but
it's
a
fine
that
really
comes
out
if
they're
being
told
to
do
something
instead
of
helping
hand
and
okay.
Well,
we
have
these
issues
that
we
need
to
solve.
B
E
Terms
of
like
helping
middle
management
and
everything
figure
it
cuz
like
we
found
that
developers,
they
love
the
idea
they
want
to
do
it
yeah
and
we've
even
found
that,
like
the
highest
levels
of
the
organization,
they
also
love
the
idea
and
they
want
to
do
it
because
they,
but
it's
it's
just
the
the
middle
management
are
really
because
they've
got
a
job
to
get
done
as
well
and
they're
trying
to
divide
their
time
up
and
that's
kind
of
where
it's
hard
to
help
them.
So.
B
I
think
we
kind
of
got
lucky
there
being
that
all
of
our
servers
are
kind
of
end-of-life.
We
had
to
migrate
somewhere
and
we
kind
of
caught
that
and
the
pharmacy
is
CTO,
had
the
directive
of
moving
a
lot
of
stuff
to
the
cloud,
so
that
was
already
part
of
the
project
roadmap
for
a
lot
of
a
lot
of
management,
so
we've
been
able
to
ride
that
wave
more
or
less
so
rather
advantageous.
Everyone,
yeah
Pippi,
is
also
a
great
way
to
kind
of
break
down
some
of
those
concepts.
You
find
that
that
helps.
B
F
B
Just
because
of
those
issues
where
you
know,
we
can't
have
multiple
replicas
without
doing
a
lot
of
extra
work
like
you
can
put
it
in
a
stateful
set,
there's
something
like
that,
but
it
may
not
be
worth
the
operational
way
to
try
to
fit
it
into
that
regard.
That
system
it
might
just
be
easier
just
to
take
it
as
it
is
and
just
drop
it
into
it.
App
service
versus
a
Windows
environment.
H
B
Most
of
it
is
open
source,
so
we're
not
paying
any
major
major
fees
for
for
a
lot
of
it.
The
team
is
pretty
small,
so
we're
not
like
hiring
tons
of
extra
people
to
support
it.
We're
hoping
that
by
consolidating
a
lot
around
certain
platforms
like
kubernetes,
we
don't
have
so
those
for
answered
applications.
A
lot
of
that
is
technical
debt
from
the
those
mergers
with
the
other
townships.
B
G
B
H
B
I'm,
the
the
big
reason
was
just
in
terms
of
support
ability
for
a
small
team,
currently
I'm
right
now,
the
only
full-time
staff
that's
supporting
this,
so
adding
is
still
on
top
of
that
well
really
awesome
and
adds
that
extra
layer
of
operational
complexity,
which
we
couldn't
quite
afford.
Linker
D,
gave
us
the
tools
that
we
needed,
namely
server,
smash,
sorry,
the
mutual
TLS
and
symmetric
insights.
B
So
it
gets
us
on
our
way
to
moving
towards
the
zero
trust
model
and
it
gives
it
makes
our
security
or
it
gives
us
a
better
security
stance
without
having
to
read
you
know,
retool
things
or
have
a
team
dedicated
to
supporting
that
same
reason.
We
didn't
go
with
spinnaker
like
really
awesome,
but
you
need
a
lot
of
people
to
know
know
how
it
works.
Thank
you.
I.
C
B
It
definitely
does
so
a
big
example:
we're
not
allowed
keeping
public
information
in
our
DMZ,
so
the
application
we
just
launched
it
has
a
file
upload
component,
so
we
were
gonna,
do
everything
in
cluster
out
in
the
DMZ,
but
being
that
there's
a
security
restriction
where
we
could
store
the
data.
We
had
to
set
up
a
instance
on
the
internal
network
and
have
a
connection
that
way
where
the
data
is
stored
internally,
but
the
application
is
external.
That
was
one
instance.
B
I
B
Gonna
be
hybrid,
we
have
some
applications
that
deal
more
that
deal
with
a
lot
of
scale,
but
mostly
seasonal.
Those
would
probably
remain
on
Prem
right
now,
while
we
migrate
and
get
a
feel
for
things,
but,
like
I
said
earlier,
like
we're
going
based
on
security
concerns,
we're
gonna
have
stuff
on
Prem,
as
well
as
in
the
cloud
so
for
the,
for
the
future
will
be
a
hybrid
model.
Okay,.