►
Description
Ortelius hoards an amazing amount of data. As we move the project forward, Data Science will allow us to leverage this data to pursue machine learning and predictive behavior of microservices shared across teams. Join Arvind, a data science enthusiast and brilliant University student as he shared is thoughts on the road forward to applying data science to the Ortelius project.
A
B
Hi,
steve,
hello,
everyone,
I
am
irvin
singhar.
I
am
a
college
student
from
india
and
I
am
in
my
current.
Currently
second
year
of
my
car
is
live,
and
I
am
doing
my
beta
degree
in
computers.
Today,
I
am
exploring
the
topic
of
how
we
use
data
in
the
audience
open
source
product
autelius
is
a
microsoft
catalog
that
version
and
track
microservice
means
it
take
all
of
your
micro
services.
B
B
A
You're
welcome
welcome
aboard
we're
glad
you're
able
to
take
up
this
topic
and
give
us
some
some
insight
into
what
we
can
do
with
machine
learning
and
the
data
we
have
in
artelias.
B
First,
let
us
try
to
understand
what
is
below
material
report
and
the
blast
radius
report.
In
the
first
diagram.
The
orange
boxes
are
all
the
different
micro
services.
Take
it
as
a
part
of
the
program
or
part
of
the
application
that
you
are
using.
It
may
be
any
front,
and
it
may
be
an
impact.
It
may
be.
B
A
B
The
same
is
it's
a
large
radius
report.
It
is
little
bit
slightly
different,
but
not
that
much,
but
in
this
diagram
the
orange
one
is
the
same.
The
microservices
part
and
pink
one
is
the
same
application
that
we
use,
and
this
diagram
is
showing
that
a
single
micro
services
is
used
by
three
different
programs.
B
B
The
pink
one
which
is
connected,
which
is
using
macro
services,
is
a
application
and,
if
just
store
lesser,
which
is
the
dark
blue
one,
it
is
a
type
of
cluster
that
is
using
types
of
history
store
or
many
history
stores.
A
B
One
in
the
localhost,
where
we
are
storing
our
android
now
now,
let's
talk
about
the
main
data
points
that
we
want
to
cover
in
this
presentation.
The
first
one
is
license
consumed
by
a
component
version
or
micro
services.
It's
basically
tell
about
the
license
or
the
it
is
telling
about.
The
version
of
the
micro
services
that
we
are
using
like
every
micro
service
has
an
updated
version.
It
has
many
version
and
it
has
many
dependencies.
B
B
B
Our
third
part
is
service
to
service
mapping.
In
this
section
we
are
talking
about
the
api
endpoint
for
is
and
rest
api
calls.
Basically,
it
is
showing
that
it
is
telling
us
about
the
last
api
that
we
are
using
and
what
are
the
results?
It
is
showing
at
different
conditions
and
last
one
we
are
talking
about
transport
and
trust,
microsoft.
B
B
We
are
data
holders.
Basically,
we
are
the
reducer.
We
are
the
producers.
We
are
looking
to
grab
data
from
the
low
level
enumeration
packages,
python
module
operating
system
packages
and
associates
into
microservices
further.
We
are
doing
this
to
takes
units
over
time
and,
basically
to
mean
trade
changes,
which
means
we
are
trading.
We
are
keep
recording
of
our
version
of
our
micro
services.
B
This
sometimes
creates
risk,
and
for
this
I
mean
the
dependency
changes.
We
have
to
manage
all
the
dependency
that
we
are
using.
For
example,
I
have
application.
Av
is
just
using
two
dependency,
two
different
software,
b1
and
b2,
and
in
the
current
vendor
we
I
am
using
the
updated
version
of
b1,
but
if
I
roll
back
or
if
I
go
back
to
the
previous
version,
maybe
it
is
working
on
the
not
the
updated.
It's
maybe
working
on
the
lesser
button
of
the
of
my
dependency.
B
A
B
B
For
that
we
need
some
arguments
or
we
need
some
type
of
data
about
how
we
can
do
that
on
what
basis
we
can
give
stars
to
values
or
favorites,
so
for
as
a
single
shopper,
microsoft.
These
are
some
of
the
points.
First,
one
had
the
micro
service
fin
deployed
successfully
and
constantly,
if
matter,
if
we
can
deploy
our
microservice
easily
without
any
problem
that
definitely
it
has
a
low
risk
value
because
we
are
not.
It
is
not
getting
that
much
problem
and
it
is
worked.
Fine
on
the
deployment.
A
B
10
is
at
high
risk
value.
What
does
it
mean
that,
if
my
application
is
using
more
micro
services,
then
we
have
to
keep
track
of
definitely
so
that
number
of
micro
services,
and
for
this
we
have
to
keep
dragging
off
that
part.
For
example,
if
you
have
over
10
micro
services,
we
have
to
keep
updating
or
we
have
to
keep
managing
is
updated
button
or
its
latest
version
of
10
parts,
so
it
is
creating
much.
B
What
domain
is
is
in,
for
example,
if
I
am
facing
any
problem,
little
less
high
value,
if
I
end
facing
the
problem
in
the
database
is
maybe
it
is
slightly
high
value
is
at
this
point,
but
if
I,
if
I
am
facing
the
problem
in
the
security
world,
then
it
is
seriously
a
major
concern.
It
will
affect
my
whole
application
and
maybe
my
application
also
shut
down
due
to
this.
So
we
have
to
keep
tracking
of
this
type
of
cns.
B
B
A
B
B
B
A
B
B
One
has
leave
here
three
years
and
two
has
and
the
second
one
has
a
leave
around
four
year.
Then
our
total
leave
here
is
three
for
four
equal
to
seven
years
and
if
our
application
has
leave
here
more
than
ten
years,
then
it
has
high
risk
value.
We
are
doing
it
keep
back.
We
are
during
the
last
stream
algorithm
as
compared
to
the
current
one.
B
B
A
B
B
Historical
is
during
version
1.0
of
car
services
and
if
social
labor
day
is
using
version
2.0
of
the
car
services,
they
both
are
using
the
same
microservices,
but
they
both
are
using
the
different
folder,
and
so
we
have
to
keep
tracking
office
version.
Maybe
they
are
using
the
same
functionality
or
maybe
they
are
using
different
functionality.
B
A
When
we
start
assigning
all
of
these
risk
values
to
the
different
microservices,
what
does
that
allow
us
to
do?
B
Our
end
user
can
trust
our
application
more
like
we
are.
They
are
using
the
microservices,
but
no
one
guarantees
you
that
it
is
the
best
or
it
does
not
create
any
problem
because
we
are
using
a
vast
amount
of
the
our
application
is
so
big,
and
so
this
micro
services
parts
this
this
score
will
give
them
more
assurance
and,
second
part
like
we
are.
B
B
B
Let's
say
we
have
100
developers,
making
two
micro
services
updates
a
day
and
we
end
up
getting
200
centers
per
day
that
need
a
risk
score
calculated.
But
why
do
we
need
a
discord
support
out
of
200
changes?
True
changes
can
destroy
our
application,
so
we
have
to
prepare
before
that's
why
the
that
this
strangest
can
affect
our
application
or
not,
because
it's
interesting
for.
B
We
can
be
a
for,
but
but
what
about?
We
have
a
massive
big
company
and
we
are
putting
thousands
of
changes
per
day.
Then,
in
this
case
it
is
hard
for
us
to
determine
that
fit
path,
creates
any
problem
and
we
can't
roll
back
thousands
of
changes
into
the
previous
further.
It
will
take
so
much
manpower
and
time
to
follow
this.
We
need
a
criteria
or
we
provide
them
a
risk
code,
so
it
decide
that
the
changes
that
we
are
using
is
suitable
for
our
application
or
not.
B
And
the
next
question
is
yeah:
let's
we
give
the
give
the
changes,
the
risk
code
and
what
we
do
next
and
this
part
is
covered
in
the
pipeline
part
we
can
make
in
the
ci
cd
pipeline.
Suppose
we
make
a
change
and
it's
at
very
low
risk,
maybe
0.1
or
0.2.
A
B
B
A
A
B
A
A
And
do
you
think
the
as
we
apply
these
models
to
the
data
that
the
models
will
be
able
to
start
learning
about
the
success
or
failure
of
the
changes.
B
B
When
we
make
a
model
and
it's
it
will
improve
every
time,
it's
track
any
changes,
because
every
time
you
take
any
vendor
averages
or
you
take
any
failure,
it
added
into
its
data,
it's
data,
and
it's
it's
learned
from
it
so
next
time
that
that's
the
type
of
error
occurs,
it
will
automatically
tell
us
that
it
is
not
good.
It's
just
gates
problem.
So
in
this
way
we
are
building
a
more
secure
or
more
powerful
module
to
do.
Desktops.
A
So,
based
on
what
you
just
told
me,
it
sounds
like
when
these
models
start
learning
that
the
stability
and
the
security
of
our
applications
is
gonna
get
better
and
that
we're
gonna
have
fewer
failures
around
the.
What
we're
releasing
because
of
our
ml
model.
Being
able
to
learn
is.
Is
that
correct.
B
Yeah,
for
example,
last
day,
let's
take
a
simple
example:
I
have
a
front-end
application
and
there
are
three
people
contributing
to
my
application
and
I
am
using
maybe
let's
say
a
virtual
provider
automatic
checking
of
the
application.
B
Let's
take
an
example
that
I
am
the
main
contributor
and
or
someone
else
did
a
bit
of
pr
and
this
create
a
forum
in
the
application
it
has,
and
if
I
have
this
this
score
and
if
the
risk
code
is
slow,
then
I
can
easily
move
my
xpi
into
my
application,
but
if
it
has
higher
score,
I
can
definitely
tell
him
to
cancel
that
pr
and
make
another
pr
so
that
in
this
way
I
am
it
is
safe.
We
are
using,
we,
I
don't
have
to.
B
A
Got
it
that's
going
to
be,
this
is
going
to
really
help
with
being
able
to
get
developers,
changes
to
production
much
faster
in
a
in
a
safer
way.
B
B
B
A
So,
thank
you
again,
arvin
for
doing
this
and,
like
I
said,
if
you're
interested
in
more
about
data
sciences
and
where
we
can
take
the
information
that
we're
grabbing
from
the
different
parts
of
the
ortelius
project
and
what
we
can
do
with
it
or
if
you
have
ideas
about
what
we're
missing
other
ways,
we
can
view
the
world
please
reach
out
to
us
or
just
reach
out
to
arvin
on
linkedin
or
twitter.