►
Description
en Español from OpenShift Commons Gathering Buenos Aires August 2018
Maximiliano Rinaldi - Security Media Analytics @ Dirección Nacional de Migraciones
A
Hello,
good
morning,
me
and
Maximiliano
Rinaldi
from
the
National
Directorate
of
Migrations
and
accompanying
the
project
from
its
beginnings,.
More
than
anything
on
the
infrastructure
side,
to
be
able
to
implement
sham,,
we
had
to
endure
a
large
number
of
changes,
because
we
always
work
with
everything
that
it
is
infrastructure
on
the
side
of
hyper,
vice,
microsoft,
etc.
and
to
be
able
to
get
to
this.
We
need
to
implement
redhat
openshift
virtualization,
which
was
a
significant
change
to
be
able
to
get
to
that.
At
the
beginning.
A
A
Keep
in
mind
that
migrations
It
is
an
organization
whose
main
function
is,
on
the
one
hand,
to
register
and
document
all
the
people
who
want
to
stay
in
the
country,.
But
it
is
also
necessary
to
register
the
income
and
expenses
of
the
country,
controlling
them
previously,
because
we
cannot
let
anyone
enter
or
leave
the
country..
There
is
an
order
of
200,000
people
entering
or
leaving
the
country
at
our
more
than
230
land
and
sea
border
/,
air
crossings,,
of
which
there
are
around
a
million
that
are
biometric
transits
at
our
airports,.
A
Keep
in
mind
that
this
takes
some
time
both
to
validate
it
previously
and
to
record
so
much
information
today,
and
we
are
also
working
with
the
geometric
doors.
I,
don't
know
if
some
of
them
are
old
recently,
but
they
go
to
the
airport.
They
don't
even
have
to
talk
to
anyone.
They
pass,
they
put
their
fingerprint
passport
photos
and
continue
on.
They
will
be
about
20
30
seconds,,
which
makes
it
very
fast
to
be
able
to
optimize
times,
because
we
are
not
really
that
many
in
migrations.
A
Well,,
the
question
is
that,
among
so
many
applications
that
we
have,,
we
have
data.
That
is
not
so
normalized
because
the
applications
come
from
different
sources,
and
there
are
many
that
have
a
large
amount.
For
years
and
analysts,
before
being
able
to
decide
whether
or
not
someone
can
enter,
had
to
control
between
4
and
5
systems,
among
other
things,.
They
had
to
control
Excel,.
They
had
to
request
it
from
the
information
office
to
the
systems.
Area.
A
Some
time
ago,.
We
looked
for
a
way
to
be
prepared
in
advance
for
who
may
or
may
not
arrive,
and
that
is
why
we
began
to
work
with
what
What
is
xavi?
It
is
the
advance
passenger
information
that
is
implemented
in
many
countries
in
this
way,
some
time
before
the
person
comes
to
the
country.,
We
already
have
a
message,
letting
us
know
who
can
be
and
who
cannot
be,
so
that
they
give
us
a
window
to
find
out.
If
Does,
the
person
have
a
problematic
record?
If,
they
have
a
criminal
record,.
A
A
Will
realize
that,
in
order
to
process
all
this,,
an
analyst
cannot
wish
how
many
people
they
think
there
may
be
in
an
address
like
this
to
filter
all
this
kind
of
content.
200
thousand
people
per
day
are
not
really
that
much.
We
will
have
of
the
order
of
40
people,
today,
analyzing,
all
this,,
for
that
we
had
to
implement
a
unified
control
system,,
which
is
the
migration
analysis
system,,
which
participated
a
few
months
ago
in
the
san
francisco
innovation
awards
and
received
an
award.
Now
I
am
going
to
leave
you
a
video
of
this.
B
C
D
E
D
F
A
Well,,
what
was
the
implementation
of
the
immigration
analysis
system
took
a
long
time
for
what
it
was
to
know
the
business,
because
it
was
He
talked
to
all
the
analysts
looking
one
by
one
to
see
how
what
kind
of
searches
they
did,,
how
they
did.
Them,
like
the
documentation
he
had,,
which
were
huge
books
to
be
able
to
learn
and
implement
all
this
within
our
system..
Now,
Victoria
is
going
to
tell
you
a
little.
more
about
it.
G
One
of
the
great
challenges
we
had
in
conjunction
with
migrations
was
the
powerful
andrew
these
needs,
and
this
became
the
system
that
was
built
under
the
name
of
what
sam
migration
analysis
system
does.
Is
that
in
a
single
access
point
it
allows
reconciling
different
sources
that
can
be
open,
can
be
closed,.
The
origin
does
not
matter,.
The
format
does
not
matter,
as
max
also
mentioned,,
and
it
allows
the
analyst's
expressive
capacity
to
be
enhanced
because,
as
mentioned,,
we
incorporated
the
heuristics
that
are
typical
of
migrations,.
G
G
Come
sources
from
the
Ministry
of
Justice
from
what
is
the
Ministry
of
Security
from
what
is
anses
of
what
is
afip
of
what
I
mentioned.
Also
of
the
advance
passenger
information,
which
is
what
is
called
api
and
the
information
and
internal
sources
that
migrations
have,,
which
is
where
the
different
eradication
requests
are
registered,.
The
different
steps
with
which
land,
river,,
sea
and
air
immigration
control
points
and
all
the
requests
and
restrictions
that
are
happening
that
are
being
reconciled.
At
this
point.
G
All
this
information
is
pre-processed
and
homogenized
to
pass
these
heuristics
and
based
on
that,
we
transform
it
into
a
model
that
is
intelligible
to
the
user,,
which
means
that
what
we
do
is
unify
and
consolidate
the
profile
of
a
user,.
That
information
goes
from
sambi
huando,
because
every
time
someone
passes
through
an
immigration
checkpoint
sometimes
and
goes
through
the
they
will
have
experienced
it
in
a
terrestrial
point.
G
It
reduces
the
technological
reluctance
that
there
is
about
how
to
access
that
data
and
that
he
can
directly
manage
his
knowledge,
then,
in
this
way,
It
facilitates
the
work
of
the
analyst,
and
it
also
makes
it
easier
for
you
to
work
from
the
systems
area
because,
as
max
mentioned
before,
and
the
different
different
search
formats
were
requested,,
the
systems
people
should
go.
Away,
remember
that
this
is
an
area
that
works
24/7.
7.
G
G
One
of
the
main
points
that
we
mention
is
the
issue
of
searches
in
searches,.
One
of
the
particularities
is
that
sometimes
one
receives
information
as
if
it
were
a
puzzle,.
What
does
this
mean
that
sometimes
I
know
that
a
person's
name
is
accurate,
but
then
I
have
other
attributes
that
can
be
fuzzy,.
So
different
types
of
attributes
are
combined
that
make
me
some
non-precise
searches
because
I
don't
know
exactly
that:
person,,
the
type
of
nationality
that
they
can
have
or
the
last
name,.
How
can
it
be
written
exactly
semantically,?
G
Then
they
are
made
these
searches
but
approximations.
Another
important
point
is
that
one
can
do
these
precise
searches
or
you
can
do
some
more
segmentation
searches?
What
does
this
mean
that
one
can
be
focused
on
studying
or
seeing
how
migration
patterns
change
over
time
and
based
on
this,?
Certain
anomalies
are
detected.
G
G
G
What
does
this
mean?
This
gives
life
to
what
we
call
a
network
of
relationships
that
allows
me
to
look
at
this
information
in
graph
format,,
which
is
the
slightly
easier
way
to
visualize
it
and
thus
understand
how
it
is
All.
These
relationships
are
made
up
of
how
they
are
traced,
as
they
enter.
The
platform.
Another.
Very
important
point
is
that
if
you
have
a
consolidation
of
data
in
a
common
repository,,
you
leave
an
audit
trail
of
the
people
who
observe
that
data
who
access
it.
G
It
can
also
issue
alerts
about
that,,
something
that
we
are
going
to
show
you
now,
of
course,.
It
is
with
synthetic
data,.
That
means
that
it
is
simulated
data
for
a
matter
of
confidentiality,,
also
with
migrations
and
the
data,.
This
is
what
I
was
telling
you
that
one
has
as
if
they
were
Google
internally
from
their
own
sources,
where
one
can
do
a
search
that
can
ask
for
a
segment
of
saying.
Well,
all
the
female
genders
that
have
passed
between
such
and
such
a
date.
G
G
G
In
this
case,
I
am
on
the
profile,,
which
is
the
one
we
saw
previously,
the
transit
one,.
If
you
look
at
this
map,
I
can
see
that
this
person,,
this
possible
profile
or
set
of
profiles,,
has
a
large
influx
of
steps
through
ground
controls,.
So
in
that
way,
a
quick
way
one
can
see
what
is
the
frequency
of
the
transits?
G
What
are
the
transits
that
predominate
the
frequent
sources
in
the
case
of
admissions
also,
what
are
the
requests
for
eradication
is
then
where
they
are
located
and
in
this
way
to
be
able
to
work
on
everything
that
It
is
also
the
traffic
of
people,
because
sometimes
it
happens
that
certain
addresses
coincide
with
some,
some
managers,.
So
in
this
way,
one
can
begin
to
detect
this
type
of
pattern,,
the
frequency
of
transits,
in
this
way,
to
understand
a
bit
of
the
information,,
not
only
from
the
search,
but
with
the
characterization
by
certain
metrics,.
G
G
Frequent
words
in
sentiment.
Analysis
have
to
do
with
whether
it
is
positive
or
negative,
neutral,
to
know
and
have
a
thermometer
of
what
was
being
talked
about,
and
how
often,
and
at
what
time,
positive
or
negative
is
being
spoken..
This
also
allows
us
to
improve
what
It
is
the
quality
of
services,,
the
different
migratory
control,
points,.
G
One
can,
for
example,
in
this
understand
there
is
something
we
call
detection
of
topics
that
allows
me
to
understand
what
is
being
talked
about:
here.
For
example,.
What
we
did
was
download
something
nice
related
to
today
that
it's
good
thursday,
and
what
are
you
talking
about
the
words
related
to
what
is
good
thursday?
Well,
there
are
some
that
we
already
know
are
part
of
the
daily
news
in
argentina
like
vida,
al
macri,
mañana
sylvestre,
part
of
the
news
that
is
happening.
People
also
allows
us
be
attentive
and
understand
what
is
happening.
G
G
Men
then
about
that
one,
then,
then,
from
a
very
easy
and
very
agile
way,
you
can
start
generating
this
type
of
user
interface.
Rules
start
combining,
it
is
see
the
hierarchy,
and
that
is
very
agile,
because
it
is
designed
precisely
for
a
user
who
is
not
technical,,
so
he
has
knowledge
management
and
knows
what
he
is
looking
for.
Who
knows
the
business,
and
in
this
way
one
can
be
dynamically
changing
these
rules
or
patterns
that,.
G
G
I
can
see
about
this,
for
example,
in
this
profile
or
talk
about
what
I
had
mentioned
to
you,
before,,
which
was
the
relationship
network
and
understand
how
they
are
related,,
which
are
the
directly
or
indirectly
related
profiles,.
One
can
also
start
playing
with
this
type
of
network,,
which
one
It
is
the
relationship
that
links
these
profiles
in
this
way,.
It
allows
me
to
understand
how
the
information
we
already
have
is
made
up
of
different
perspectives.
G
That
is
why
it
is
something
that
we
sometimes
call
that
there
is
gold
in
your
garbage,
because
sometimes
you
do
not
know
all
the
information
you
have.
Because.
It
is
not,.
It
is
not
analyzing
what
patterns
it
can
find
in
real
time
and
about
that
to
be
able
to
be
alert,,
that
is
in
broad
strokes
and
a
bit
so
that
you
can
understand
all
this
that
max
was
explaining
and
how
it
was
transformed
into
the
application
and
One
of
the
things
that
made
it
possible
to
implement.
G
It
was
obviously
the
good
synergy
between
the
different
teams
in
the
technical
team,,
the
analyst
team,
and
obviously
the
fact
that
it
has
been
one
of
the
first
projects
started
with
open-source
tools
to
do
advanced,
analytics
and
that..
It
is
something
important,
and
now
we
are
going
to
see
from
the
back
how
this
could
be
done..
There
are
many
canned,,
sometimes
known
as,
analytical
can
be
carried
out
in
a
simple
way
and
precisely
with
what
is
the
platform
of
the
open.
If
now,
Martin
is
going
to
explain
a
little
how
it
was
possible,.
H
The
particular
challenge
was
that
in
this
migration
installation
we
have
around
70
nodes
distributed
in
three
clusters,
also,
which
was
what
allowed
us
to
achieve
this
final
result.
But
in
order
to
have
that
point,
we
started
at
the
beginning
is
fine
and
we
had
to
make
the
most
of
the
advantages
of
openshift,
which
are
these
three
that
are
here
the
deployment
time,
the
architectural
complexity
and
also
the
monitoring
service
that
we
can
install
in
the
first
place.
H
Not
only
that,
but
we
could
create
new
ones
in
openshift
in
a
matter
of
minutes,
which
was
a
great
advantage
and
working
with
data
and
Working
with
some
distributed
databases,,
something
that
generated
a
lot
of
value,
was
the
point
of
being
able
to
deploy
distributed
technologies.
In
a
few
minutes,
for
example,,
we
were
able
to
deploy,,
which
I
am
now
going
to
show
you
below,
a
very
deploy,
an
elastic
search,,
an
astic
search
cluster
of
18
12
in
a
matter
of
an
hour
two
hours
in.
H
That
was
thanks
to
the
fact
that
we
didn't
need
to
do
a
fine
configuration
at
the
beginning,
but
we
simply
needed
a
cluster.
We
needed
to
be
able
to
index
the
information
we
had
to
be
able
to
start
working.
Secondly,
the
complexity.
The
architecture
implied
through
of
these
three
clusters,
together,
openshift
with
gel,
we
were
able
to
generate
how
to
do
the
job
using
them
with
we
had
a
back
of
in
the
different
versions
of
all
the
microservices
that
we
had.
H
If
there
was
a
problem,
we
could
take
a
specific
photo
and
be
able
to
trace
with
the
component
that
had
a
problem.
That
was
very
useful
after
the
consolidation
of
the
different
data
sources
when
doing
so,
the
agents
that
generate
the
import
sessions,,
which
are
the
ones
that
allow
you
to
grab
the
migration
information
to
be
able
to
index
them.
We
also
need
scalability,
automatic
scaling
to
be
able
to
keep
up.
H
With
the
volume
of
data
that
migrations
have
and
finally,
for
this
part,,
the
deployment
of
microservices,
machine
learning,
not
only
generating
the
models,
are
important,
but
also
being
able
to
use
them,,
but
the
part
of
generating
them
was
very
essential,
being
able
to
give
access
to
the
data,.
You
know,
says
the
analysts
to
notebooks
that
we
deploy
now
within
the
same
cluster,
that
is
in
the
same
environment,
can
make
use
of
the
entire
platform
to
be
able
to
generate
the
models
and
validate
them
in
the
same
place.
H
There
is
no
separate
analytics
process
for
the
soft
model,
but
it
was
in
the
same
application,
which
was
what
allowed
us
a
lot
of
speed
to
be
able
to
exchange
with
the
data
to
generate
models
that
are
finally
useful
throughout
this
process,.
The
monitoring
part
was
very
important
because
we
had
a
lot
of
hypotheses
and
we
are
validating
and
in
order
to
validate
those
hypotheses,
we
need
to
be
able
to
understand
what
was
happening
to
understand
what
was
happening.
H
We
used
prometheus
to
count
metrics
of
all
the
posts,,
how
they
interact
with
each
other
and
how
the
different
applications
were
reacting,,
which
was
what
allowed
us,
for
example,
to
go
a
long
way
to
eliminate
many
bottlenecks.
Later
this
promised
put
together
the
metrics
graf
anna
allowed
you
to
visualize
them
now,
I'm
going
to
show
you
how
to
access
fun.
H
All
all
developers
have
access.
Therefore,
there
was
no
separate
development
monitoring
team,
but
the
same
development
team
could
look
up
their
own
metrics
to
see
what
they
wanted
monitor
and
from
there
continue
its
development
and,
finally,
more
than
a
preventive
issue,
we
could
generate
proactive
alerts
when
we
saw
that
something
was
becoming
saturated
with
a
not
falling.
We
can
raise
an
alert
and
be
able
to
react
with
those
time.
H
What
I
am
going
to
show
you
now
too?
This
is
one
of
the
projects
of
one
of
the
clusters
that
we
have
in
migrations,
how
we
were
to
deploy
all
these
microservices
in
a
simple
way,.
But
the
most
important
thing
here
are
two
things,
in
the
first
place,
that
we
fuse
in
part
to
make
them
the
connectors
with
the
social
networks,.
H
We
can
deploy
it
as
a
container
inside
fuse
for
the
version
they
have
for
xbox,
which
saves
you
a
lot
of
time
because,
from
the
same
code
that
we
were
generating
in
a
matter
of
minutes,
we
could
have
an
image
running
the
same
in
the
same
environment,
which
greatly
speeds
up
the
build
process.
development
and
also
the
cluster
of
tic's.
This
is
what
I
tell
you
that
we
had
a
cluster
of
18
nodes
that
was
simply
to
scale
the
posts.
It
was
not
necessary
to
configure
more
than
that.
H
It's
fine,
that
's
all
I
was
telling
you,.
It
was
extremely
advantageous
because
we
didn't
have
to
be
there
at
an
early
stage,.
We
didn't
have
to
do
search
optimizations,,
which
allowed
us
to
start
working,.
Then
we
got
into
the
part
about
how
to
get
more
forms
of
the
application,
but
we
had
that
advantage
in
the
development
and
later
on,
the
monitoring
part.
We
had
all
these
metrics
prometheus.
Allow
us
to
get
many
metrics
at
different
levels
of
different
components,
and
for
that
it
allows
us
to
generate
different
alarms.
H
We
can
generate
different
alarms
according
to
the
need
of
the
moment,
to
see
what
is
being
done.
What
we
are
doing,
we
have
a
bottleneck
that
if
there
is
a
connectivity
problem
and
finally
all
this
is
quickly
visualizable
and
by
the
developers
in
the
data
that
we
can
create
with
graf
anna
here,
we
can
see
all
the
performance
of
the
cluster.
We
can
the
performance
of
the
pods.
A
Thank
you
very
much
Martin
among
the
next
steps,
in
addition
to
working
with
api
that
we
I
had
commented
that
in
a
few
months,
we
will
be
working
with
PNL,,
which
is
the
record
paso
and
sernam,,
and
which
includes
all
the
information
of
a
passenger
that
is
registered
in
conjunction
with
a
flight
reservation,.
If
the
person
registered
a
hotel,,
a
car,
etc.,,
we
are
also
working
a
lot
with
the
united
states
government
to
exchange
a
number
of
restrictions
to
know
the
status
of
visas.
A
Now
we
are
working
a
lot
with
what
is
it
you
know,
or
electronic
travel
authorizations,
where
a
person
who
already
holds
a
us
visa
simply
by
This
can
come
to
agree
to
enter
the
country
without
needing
an
Argentine
consular
visa.
After
all,
the
controls
that
everyone
has
in
other
cases,
any
questions
or
something
you
want
to
know.