►
From YouTube: Cloud: .NET Community Standup - February 28, 2019
Description
Join members from the .NET teams for our community standup covering great community contributions for Azure, AI, ML, Containers, and more.
A
A
The
program
manager
lead
for
tooling
out
of
the
ships
out
of
the
Dutton
end
organization,
which
includes
web
and
apps
service,
to
lean,
as
well
as
the
core
stuff
like
project
systems,
refactorings
all
the
stuff
that
makes
your
life
great
and
pleasant
as
developers
in
visual
studio
and
Phil
spends
most
of
his
time
working
on
our
apps
service
to
lean
in
visual
studio.
So
today
we're
have
a
couple
things
queued
up
for
you.
A
A
Traditional
sequel
and
then
we'll
spend
probably
the
majority
time
at
the
end
talking
about
docker
containers.
What
are
they?
Why
are
they
interesting
and
how
would
you
get
started
with
them
in
some
of
our
tools?
So
with
that,
let's
flip
over
to
his
blog
post,
that
I
found
and
so
a
blog
post
talking
about
the
new
cosmos,
DB
SDK
and
as
a
starting
point,
Phil
tell
me
talking
a
little
about
what
is
cosmos
DB,
so
cosmos
DB
is.
B
Depending
upon
what
level
you're
looking
at
it,
it's
a
storage
technology
that
exists
in
Azure
and
it's
best
known
for
its
no
sequel
approach
to
doing
things.
So
this
is
non-relational
data.
There's
a
few
other
providers
out
there.
You
can
do
similar
things
in
the
traditional
storage
accounts
on
Azure
too,
but
this
is
a
more
scalable,
more
automatically
indexable
way
of
doing
things.
Okay,.
A
A
Right:
Thanks,
yes,
so
they
kind
of
scroll
down
I!
Think
one
of
the
interesting
things,
one
of
the
other
really
interesting
things
that
he
walks
through
in
this
blog
post.
Talking
about
cosmos
DB
is
when
you
look
at
using
the
things
so
in
traditional
relational
data.
There's
a
very
strong
schema,
that's
stored
with
the
table
in
write.
So
when
I
lay.
B
A
Table
and
then
which
doesn't
always
map
perfectly
to
the
way
I'd
want
to
use
objects
in
my
code
and
c-sharp,
and
so
that's
why
we've
created
called
object,
relational
mapping,
Zoar
ORM
systems
and
in
a
framework,
of
course,
being
the
most
popular
one
in
dotnet
land.
But
there
are
many
others
as
well
right.
So
there's
sort
of
this
like
abstraction
layer
that
we
have
to
put
in
between
our
code
and
the
database
to
get
things
to
work
together
correctly
as
a
programmer.
I.
A
Think
one
of
the
more
interesting
things
about
no
sequel
tables
is
because
their
schema
and
then
so
right.
If,
for
example,
if
I
decide
that
I
want
to
add
a
new
column
to
my
database,
because
my
thing,
you
know
my
application
needs
more
data
on
an
object.
I
then
have
to
you
know,
go
update
those
tables
and
get
the
database
where
that's
going
to
be
stored,
updated
in
order
to
read
and
write
right.
So
if
I
try
to
write
up
to
a
table
or
a
column
that
doesn't
exist,
I'm
gonna
get
an
error
right.
A
A
B
You
can
do
some,
you
can
do
effectively
a
lazy
upgrade
as
opposed
to
having
some
amount
of
downtime,
while
you're
updating
the
schema
or
read-only
state
or
there's
a
number
of
ways.
You
could
do
it
with
relational
with
non-relational.
You
just
add
the
property.
You
default
the
property,
oh
the
property's,
not
there.
It
acts
like
X
and
therefore
you
can
adapt
from
there
and.
A
So
in
this
blog
post
I
think
it
does
a
good
job
right.
So
this
person
has
created
this
object
called
test
entity,
and
so
we
just
add
some
stuff,
and
then
he
writes
it
and
then
I
said
in
the
future.
It
would
be
just
as
easy
as
I
know
the
properties
only
one
there
is
one
schema
requirement.
I
should
mention
in
cosmos.
Do
you
know,
and
that
is
every
property
must
have
an
ID.
So
every
item
written
into
the
database
has
to
have
an
ID
attribute
yep,
and
that's
just
so.
A
You
can
look
it
up
later,
which
make
sense.
So
that's
the
only
schema
actually
so
this
person
talked
about
and
the
new
version
of
the
SDK
that
they
found
there
used
to
be
an
option
in
the
older
version
to
auto,
generate
the
SDK
or
Sarah
auto
generate
the
ID
field,
and
it
looks
like
they've
actually
taken
that
away.
So
now
it's
your
responsibility
to
actually
provide
a
ID
implementation
in
the
object
that
you
write
into
it.
So
they
kind
of
interesting
blog
post.
A
They
talk
about
just
evaluating
the
new
SDK
and
the
one
thing
that
got
them
was
they
started
off
by
just
saying
our
item
new
test
entity.
They
didn't
provide
an
ID
and
they
got
an
error
message
back
and
then
the
fix
is
actually
to
provide
their
own.
Do
that
cuz
now
it
doesn't
have
the
ability
to
auto-generate
it
once
you
actually
know
the
ID
that
you
wrote
in
so
you
can
look.
It
up
was.
A
Fully
so
the
error
message
says:
response
code
does
not
indicate
success.
Reason.
The
input
name
null
is
invalid
and
sure
to
provide
a
unique
non,
empty
string,
less
than
255
characters,
so,
okay,
probably
unique
enough
to
Google,
but
probably
not
unique
enough
to
infer
what
you
needed
to
do.
Implicitly
from
the.
B
B
B
It
really
depends
upon
your
call
pattern
and
your
ability
to
predict
the
queries
you're
going
to
need
to
do
in
the
future.
If
you
have
a
an
application
that
has
a
very
well-defined
flow
through
it
and
that,
therefore,
you
can
predict
the
queries
and
you
can
do
what
you
need
to
do
in
order
to
make
them
run
quickly,
so
that
you
can
scan
off
of
the
more
easily
indexable
fields.
You
can
do
some
amount
of
denormalization
like
have
the
same
data
in
multiple
rows
with
slightly
different
names.
So
you
can
query
them
different
ways.
B
A
No
sequel
database.
Is
that
a
fair
that
that's
totally
fair
and
it's
all
about
measuring
and
understanding
the
scale
of
the
data
and
the
types
of
queries.
So
we
don't
want
to
fall
into
hammer
and
nail
trap
right.
We
don't
meaning
when
the
only
tool
you
have
is
a
hammer.
Everything
looks
like
a
nail,
so
it's
not
that.
No,
that
sequel
and
relational
databases
are
dead.
A
It's
just
that
no
sequel
opens
up
new
use
cases
for
things
that
sequel
databases
or
relational
databases
may
not
have
been
as
great
at
before,
and
I
had
to
work
hard
to
change
the
shape
of
my
data
to
fit
it
into
my
database.
Absolutely
great
one
of
any
other
benefits
of
Cosmos
that
you
can
think
of.
B
B
A
So
there's
actually
thing
I,
always
thinking
of
when
I
ask
that
question
is
cosmos
to
be
automatically
solves
geo
replication
for
you.
So
if
you're
trying
to
work
with
an
application
around
the
that's
going
to
run
in
multiple
regions,
it
will
automatically
be
instance
in
each
of
the
region
that
goes
with
their
website
and
it
will
handle
making
that
data
consistent
between
all
of
the
regions,
and
it
will
do
it
very
very
quickly
on
the
order
of
milliseconds.
So
there's
no,
let's
run
a
job
every
night
at
3:00
a.m.
to
try
to
get
our
databases.
B
A
All
of
our
regions,
or
whatever,
like
cosmos,
handles
it
in
milliseconds.
There
have
been
some
really
interesting
as
your
Friday
videos,
where
Scott
Hanselman
is
had
people
from
the
cosmos
TV
team
on
and
they
tried
to
force
an
inconsistency
error.
I
mean
they
had
a
script
written
to
try
to
force
an
error
to
get
the
data
out
of
sync,
and
sometimes
even
the
script
can't
get
it
out
of
sync,
because
it
gets
in
sync
that
fast
across
regions
cool
because
it
means
there
are
milliseconds
like.
A
Obviously
you
can't
move
data
from
one
location
to
the
to
the
other
instantly,
but
it
is
really
really
fast
like,
even
if
you're
trying
to
break
it
and
get
it
out
of
sync,
it's
actually
kind
of
hard
good
good
good
and
they
had
an
interest.
There's
another
interesting
blog
post,
like
I
found
as
I
was
doing.
A
This
I
probably
should
have
had
a
queued
up,
but
it
was
actually
the
person
was
talking
about,
let's
see
if
we
can
find
it,
but
the
person
was
talking
about
like
actually
using
cosmos
DB
as
a
cache
instead
of
Redis,
and
the
reason
for
that
is
because
cosmos
DB
solves
all
of
the
eventual
consistency
issues
so
quickly.
They
were
saying
that
they
had
so,
let's
see
if
we
find
this
say
you
know,
cosmos,
DB,
Redis,
cache
and
I
actually
had
talked
about
it.
A
Yeah
using
Azure
cosmos
DB,
as
your
persistent
geo
replicated
distributed,
cache
for
asp.net
core
walk
through
the
blog
post,
necessarily
right
now,
I
thought
it
was
interesting.
They
kind
of
talked
about
like
Redis,
is
very
much
designed
for
this.
Then
they
talked
about
when
they
were
trying
to
work
across
multiple
regions.
A
I
mean
it's
interesting
technology,
that's
worth
playing
with
and
exploring
it's
something
that
whenever
I've
used
cosmos,
it's
really
is
really
pleasant
to
me.
If
there
is
now
an
entity
framework
or
did
create
an
EF
core
model
over
cosmos,
if
you
prefer
to
work
with
entity
framework
that
came
new
and
core
to,
but
you
don't
need
to,
unlike
sequel,
where
you
really
needed
to.
A
They
start
with.
What's
asp.net
core
well
gloss
over
that
in
this,
let
me
talk
about
some
of
the
problems
that
containers
solve,
and
so
can
you
just
tell
me
a
little
bit
about
kind
of
what
is
it?
How
do
I
think
about
a
container?
Is
it
sort
of
this
magical
thing
that
solves
all
of
my
problems
as
a
developer?
A
B
Only
those
existed
but
containers
give
you
advantages
in
isolation,
so
that
you
can
run
a
number
of
applications
on
the
same
hardware,
but
that
they
don't
interact
with
each
other.
The
way
that
they
would,
if
you
boot
it
up
a
VM
and
install
three
or
four
applications,
they
could
have
contention
over
various
system,
resources
and
so
I
picture
containers,
as
micro
VMs,
with
the
ability
to
have
isolation
for
the
different
apps.
The
apps
can
still
use
the
overall
system.
B
Resources
and
they'll
always
compete
for
the
same
network
card
in
the
same
physical
CPUs,
but
they
won't
can
contend
over
all
those
kernel
objects
in
their
own
space
and
that
they
won't
contend
with
the
same
file
system
and
overwrite
version
of
one
library
with
another
version
of
another
library,
with
a
subtle
inconsistency
that
only
occurs
if
both
apps
are
installed
on
the
same
machine.
Okay,.
A
That
was
a
lot
so
I'm
gonna
unpack.
What
you
just
said
probably
could
have
written
a
master's
thesis
on
about
one
minute,
so
I
think
there's
a
starting
point.
A
docker
container
really
is
a
unit
of
deployment
from
my
application
is
that
is
that
a
right
way
to
think
about
it?
So
I'm
gonna
use
an
analogy.
A
A
I
think
this
is
where
you
talk
about
isolation,
and
things
like
that,
so
one
of
them,
in
you
know,
I,
think
that
will
be
familiar
to
any
dotnet
developer
has
been
what
do
I
do
if
I
have
an
old
application
that
nobody
wants
to
fix,
because
nobody
understands
anymore
that
doesn't
work
on
a
newer
version.
Of.Net
framework,
mm-hmm
dotnet
framework
is
a
singleton
on
the
entire
server
and
so
I
have
worked
and
talked
with
customers
who
are
still
stuck
on
very
old
versions.
A
Of.Net
framework,
because
versions
of
some
application
that
somebody
wrote
eight
years
ago
requires
net
framework
for
Oh,
for
example,
and
then
they've
left
the
company
and
they
tried
to
upgrade
the
server
to
a
newer
version,
of.net
framework
and
the
application
didn't
work
instead
of
debugging
it
now
everybody's
just
stuck
continuing
to
target
dotnet
framework
for
dot
or
dot.
Oh
because
it's
hard
to
upgrade
the
server,
and
so
when,
anyway,
you
talked
about
isolation.
A
I
think
the
reason
that's
key
I
think
this
picture
does
actually
a
really
good
job
as
I
go
down
the
blog
post
now
I
could
have
solved
that
before
with
a
virtual
machine
mm-hmm,
but
the
problem
with
the
virtual
machine
from
a
size
perspective
is
I'm
bringing
an
entire
copy
of
the
operating
system
that
everything
needs
to
running
right.
So
every
time
I
have
a
VM
or
a
VHD
or
whatever
your.
A
Technology
is
right,
I'm
in
this
place.
Where
and
this
this
actually
picture
does
a
really
good
job.
I
should
sue
min
here.
Nothing
illustrates
it
really
really.
Well,
that's
a
lot
of
wasted
space
and
computing
power
and
resources
on
the
actual
metal
to
run
multiple
copies
of
the
same
operating
system
when
the
problem
I'm
simply
trying
to
solve,
is
runtime
dependencies
or
things
like
that.
So
the
idea
behind
a
container
right-
and
you
describe
this
really
well,
so
we're
still
unpacking
minute.
A
What
you
said
in
this
picture
does
a
really
good
up
illustrating
is
you're
actually
sharing
the
host
OS,
so
the
container
gets
to
be
much
smaller
than
a
virtual
machine,
because
I
don't
have
an
entire
copy
of
the
guest
OS
in
it.
I
just
have
the
application
and
the
dependencies
that
my
application
have.
A
Have
you
know
random
third-party
software
dependencies
mm-hm,
and
so,
if
I,
think
about
going
to
a
cloud
like
asher
generally,
where
I
want
to
be
is
I
want
to
be
in
a
managed
service,
something
like
as
your
app
service,
where
they're
taking
care
of
managing
the
infrastructure?
For
me,
upgrading
my
OS
things
like
that,
but
those
managed
services
generally
don't
give
you
the
ability
to
customize
the
environment
right.
That's
why
it's
a
managed
service.
A
So
if
my
old
asp
net
application
depends
on
Crystal,
Reports
or
needs
office
installed
on
the
machine
to
do
some
level
of
processing,
I
can't
do
that
in
a
managed
environment
in
as
you're
going
to
a
peer
managed
environment.
But
if
I
put
package
that
in
a
Windows
container,
I
can
install
all
of
that
software
into
that
container
and
I
can
deploy
that
container
onto
the
managed
service.
B
Absolutely,
and
with
that,
you
can
also
use
it
kind
of
like
a
a
virtual
machine
image
in
that
you
have
a
guarantee
that
all
those
machines
have
exactly
the
same
copies
of
all
those
assets,
all
those
upstream
requirements
that
are
necessary.
So
if
you
want
to
deploy
to
like
kubernetes,
you
give
it
a
docker
image
and
therefore
all
the
machines
and
whatever
group
that
is
get
exactly
the
same
image.
A
B
A
We're
gonna
switch
over
to
your
machine
I.
Think
I'm
gonna.
Ask
you
the
question
of
how
would
I
work
if
I
wanted
to
get
started
with
containers
if
I
want
to
play
with
them?
If
I
were
interested
in
them,
what
would
I
do
or
how
would
I
get
started
with
that?
This
blog
post
actually
does
a
great
job
so
URLs
up
here
for
anybody
watching
on
the
stream.
I
would
probably
recommend
searching
for
the
title,
but
they
actually
are
doing
it
with
vias
code.
I.
A
Think
you're
gonna
show
us
how
to
do
it
in
Visual
Studio,
but
the
one
requirement
that
I
want
to
call
out
first,
because
they
have
this
link
here
under
the
requirements
install
docker
is
this
does
depend
on
a
tool
or
something
being
installed
on
your
operating
system,
it's
available
for
Mac,
Windows
and
Linux,
but
chrome
docker
desktop,
and
so
it
isn't
just
natively
on
the
operating
system.
This
is
coming
from
docker
incorporated,
and
so
you
do
have
to
install
there
set
of
tools
that
understand
how
to
build,
hosts
and
run
containers
great
with
that.
A
B
B
So
this
is
the
cloud
tool
stand-up,
so
cloud
tools
stand
up
and
we're
gonna
go
with
an
asp.net
core
web
app.
Okay,
we're
gonna
hit
OK.
Now
this
brings
us
to
a
little
piece
of
UI.
That's
going
to
give
us
some
choices
in
different
templates
to
use
normal
web
application
and
I'm
going
to
intentionally
leave
this
enable
docker
support
checkbox
that
I'm
hovering
over
unchecked,
because
I
want
to
show
that
experience
of
I
have
an
existing
application.
I've
been
using
it
I
already
know
that
it
runs
locally.
I
want
to
use
this
inside
of
containers.
B
A
While
you're,
there
sure
I
think
you
just
already
answer
the
question,
that's
what
I
actually
said,
let's
assume
I
have
an
existing
application.
I
want
to
try
to
containerize.
So
if
I'm
starting
brand
new,
yes,
is
easy
and
Visual
Studio,
2017
and
then
vs.
P
of
2019
also
works
the
same
way
as
checking
a
checkbox
when
I
create
the
project
that
says
docker
eyes
this.
For
me,
absolutely.
A
B
Hope
not
so
what
you
need
to
do
is,
after
you
have
the
solution
open
you
right.
Click
on
the
project,
node
and
under
add
there's
a
couple
of
options,
ones
called
container
orchestration
support
and
the
other
one's
called
docker
support.
Container
orchestration
support
gives
you
kubernetes,
which
I
kind
of
touched
on
a
little
bit
as
an
option,
but
also
docker
compose
and
then
to
get
the
same
effect
you
come
over
here.
A
You
there
one
question
right.
Second,
so
I
think
you
said
you
because
this
you
got
to
choose
between
Windows
and
Linux,
because
this
is
an
asp.net
core
application,
because
obviously
dotnet
core
is
capable
of
running
either
Windows
or
Linux.
And
if
this
was
a
asp.net,
dotnet
framework
application
or
another
dotnet
framework
application,
it
would
have
just
picked
Windows
for
you,
because
don''t
framework
obviously
only
supports
Windows.
So
we're
cleaning.
A
B
B
A
Again,
so
no
sorry
I'm,
just
gonna
pause.
You
int
and
talk
about
I
think
a
little
bit
of
the
stuff
that
we
were
talking
about
before
which
I
thought
might
be
interesting
to
do
within
the
context
of
this.
So
what
you
have
now
is
you
have
a
docker
file.
It's
gonna
tell
it
how
to
build
that
single
docker
image
right
and
so
Inc.
One
of
the
things
misconceptions
a
lot
of
times
that
people
have.
Is
they
stride
to
work
with
docker
as
part
of
what
we
would
call
the
inner
loop
for
regular?
A
You
know
edit
compile
debug
kind
of
that
tight
I'm,
trying
to
get
functionality,
working,
mm-hmm
and
I.
Think
most
of
you
know
we
talk
to
a
lot
of
our
bigger
customers.
One
of
the
patterns
we
hear
is.
We
talked
about
most
of
the
benefits
of
docker
or
around
the
idea
of
deployment
right.
It's
a
single
package.
It
enables
isolation
on
the
server
and
so
I
think
there
shouldn't
be
a
feeling
of
there's
absolutely
nothing
wrong
right
with
me
as
a
developer.
A
If
I
say
you
know
what
I'm
gonna
do
my
inner
loop
outside
the
context
of
a
container,
because
it
is
a
little
slower
to
build
and
and
launch
and
start
containers
for
part
of
that
inner
loop
and
then
say
you
know
what
obviously
before
this
goes
to
production.
One
of
the
great
things
about
docker
is
because
everything
is
compiled
into
a
single
image.
A
The
image
that
I
would
build
on
my
local
desktop
machine
is
the
same
image
in
theory
that
would
get
built
in
the
production
server,
and
so
it's
you
would
be
useful
for
me
to
test
and
make
sure
one
if
the
my
operations
team
or
we're
choosing
to
deploy
them
as
a
container
into
production.
That's
gonna
be
the
environment
that
we're
gonna
use
to
host
it.
I
will
obviously
probably
want
to
test
it
before
it
actually
goes
to
production
right
and
so
I
think
that's
what
you're
gonna
show
here
now
is
hey.
A
How
would
I
actually
test
publishing
this
thing
in
a
way
that
I
could
test
it
in
a
sort
of
broader
doctrine
if
I
have
a
test
environment
or
something
like
that?
Without
necessarily
the
need
to
have
every
f5b
in
a
container
now
by
default,
that
will
be
I'm.
Gonna
say
if
you
just
click
the
little
selection
option
up
there
right
I
can
choose
to
not
launch
into
the
docker
container.
If
I
think
that
slows
me
down,
but
then
I
would
still
be
in
a
position.
B
B
A
Actually,
take
a
question
that
we
just
come
in
while,
while
we're
you're
doing
this
and
the
question
is,
why
is
it
not
possible
to
add
docker
support
for
both
Windows
and
Linux
if
you're
listening,
I
love
love
for,
if
you
can
kind
of
type
I'm
gonna
ask
a
couple
follow-up
questions
you
can
clarify
what
you
meant
is
the
question.
Why
can
why
do
you
have
to
choose
either
or
or
is
the
question
that
you're
not
seeing
Lenox?
A
A
The
answer
is
actually,
if
we
look
at
Phil's
screen,
if
you
highlight
this
base
image
right
up
here,
yep
perfect,
so
there's
two
reasons:
one
the
container
so
you're
gonna
have
a
base
image
that
is
gonna
have
all
of
the
like.
The
copies
of
dotnet
framework.
Things
like
that
that
are
designed
for
the
host
operating
system
and
the
reason
either/or
is
you
will
actually
in
some
cases
have
to
have
it.
A
A
You
are
gonna,
be
able
to
pick
a
base
image
that
you
inherit
from
just
like
inheritance
and
programming
that
is
going
to
have
the
copies
of
dotnet
framework,
that's
designed
to
run
on
the
host
OS,
and
so
yes,
so
the
base
image
is
going
to
be
either
Windows
or
Linux,
because
dotnet
framework
or
the.net
core
runtime
that
is
actually
going
to
run.
We
is
Nate
has
a
lot
of
native
code
in
it
and
it's
compiled
either
for
Windows
or
Linux,
and
so
your
base
image,
you
need
to
know,
am
I
going
to
run
like
do.
A
I
need
the
copies
that
are
gonna
run
on
a
linux
host
OS
or
a
Windows
host
OS.
So
switching
between
Windows
and
Linux
can
be
done,
but
you're
gonna
have
to
pick
a
different
base.
Image
and
you're
gonna
have
to
maybe
flip
a
couple.
Different
settings
we've
actually
made
this
really
really
easy.
In
the
case
of
dotnet
core,
you
can
go
into
project
properties
and
you
can
flip
the
target
OS.
A
We
don't
need
to
necessarily
do
that
right
now,
but
you
can
pivot
back
and
forth,
but
the
answer
is
the
reason
you
can't
do
both
at
the
same
time
is
you
actually
have
to
know
which
one
you're
going
to
compile
and
use
and
package
into
that
container,
depending
on
the
host
OS
that
you're
gonna
run
on
and
on
Windows?
You
can
target
either
Windows
or
Linux,
because
there
is
a
Lennox
via
that's,
going
to
run
inside
hyper-v
that
will
host
your
Linux
container.
If
you
want
to
f5
locally.
A
A
B
A
container
registry
is
an
online
store
of
images.
Now,
each
of
those
images
you
can
think
of
as
the
actual
packed
up
file
from
the
the
docker
tooling.
That
is
the
one
file
that
is
going
to
be
the
the
application
that
is
run
with
all
of
its
dependencies,
and
the
container
registry
is
a
place
to
store
them
all.
B
This
is
the
place
we're
going
to
post
all
of
our
images,
or
you
could
say,
I
want
to
create
a
new
one
depends
upon
how
you
want
to
run
things
we're
going
to
go
through
the
create
experience,
because
you
have
to
have
created
one
in
order
to
select
one
later
so
I'm
gonna
click
create
profile
there.
It
brings
up
this
UI.
That's
going
to
enable
us
to
create
this
container
registry.
This
kind
of
online
folder
we
could
call
it
it
has
to
have
a
name.
I
will
actually
remove
some
disambiguation
text.
B
A
B
A
B
A
I
think
we
talked
about
a
little
bit
about
maybe
what
you're,
creating
that
it
create
things
are
gonna
create
off,
and
you
talk
to
do.
You
talked
about
the
idea
of
a
container
registry
being
a
sort
of
glorified
folder
in
the
cloud
yep.
You
talked
to
me
a
little
bit
more
about
what
that's
going
to
enable
for
me.
So
you.
B
Can
then
refer
to
the
images
in
that
container?
So
again
it's
a
folder.
You
can
think
of
the
images
as
files.
Each
of
those
images
can
have
kind
of
sub
versions
which
we
refer
to
as
tags
and
a
really
common
tag.
People
use
is
latest
to
say,
I
want
the
latest
version
of
such-and-such
an
application,
but
when
your
application
is
going
to
run,
if
it
doesn't
already
have
the
image
or
has
whatever
the
semantics
are
around
refreshing,
it
will
then
say:
oh
this
is
the
container
registry
I
need
to
talk
to.
B
A
Yeah
and
I
think
a
couple
things
worth
teasing
out
there
again
kind
of
that.
You
talked
about
so
ain't
the
point:
it's
a
way
of
sharing
images
that
have
a
release
well-defined
protocol
over
it
that
optimizes
for
getting
a
specific
copy
of
an
image
and
like
writing
and
reading
updates
from
it
just
like
git,
is
a
protocol
and
then
obviously
github
is
probably
the
most
popular
implementation.
That
actually,
you
know,
has
a
cloud
provided
a
service
that
supports
that
protocol,
but
much
like
source
control
or
get
specifically.
Everything
operates
on
differentials
in
container
registry.
A
So,
unlike
if
I
think
about
sort
of
a
standard
share
where
I
might
publish
my
application,
if
I
just
zipped
it,
let's
call
me
puts
a
zip
file
for
now.
I
would
probably
have
a
specific
cop,
like
I
would
have
a
one
zip
file
per
copy
of
my
application.
Even
if
98%
of
that
was
exactly
the
same
over
time
and.
A
And
one
denied
really
nice
benefits
that
I
contain
the
register.
Gives
me
is
the
first
time
I
push
an
image
to
it.
It's
gonna
be
big,
because
I
have
to
push
the
entire
copy
of
the
image.
So
if
it's
a
hundred
twelve
megabytes
I
have
to
push
one
hundred
twelve
megabytes,
but
then
all
subsequent
up
updates
to
that
image.
A
Yeah
and
latest
would
be
equivalent
to
head
when
the
nice
things
is,
then
the
next
time
I
push
an
update
to
that
container.
If
it
only
changed
by
one
megabyte,
it's
only
gonna
push
and
pull
one
megabyte
to
and
from
the
service,
as
opposed
to
the
need
to
actually
push
and
pull
the.
What
did
I
say.
A
hundred
and
twelve
megabytes
of
the
image
which
so
container
registries
are
yeah
a
really
nice
way
to
share
things
in
a
way
that,
after
that
initial
commit
makes
it
really
lightweight.
Cuz
you're,
just
dealing
with
deltas.
B
Very
cool,
so
here
we
are
back
in
the
we
usually
refer
to
this
as
a
published
profile
inside
a
visual
studio.
This
is
the
container
registry
we
want
to
publish
to
I
briefly
showed
for
people
who
are
watching
carefully
the
Edit
image
tag.
We
can
actually
generate
it,
one
automatically
based
upon
the
date
and
time
or
you
can
use
a
custom
tag
again
latest
is
really
common,
so
that
is
our
default.
B
I
come
in
here
we
kick
publish
it's
going
to
actually
compile
the
application
for
release,
and
then
it's
going
to
start
running
that
set
of
commands
that
were
in
the
docker
file.
This'll
this'll
take
a
minute
or
two,
but
I
want
to
show
you
the
usage
on
the
other
side
of
things,
which
is
then,
if
we
come
into
the
azure
portal
and
we
want
to
create
a
new
app
service-
and
this
would
be
a
typical
way
in
Azure
to
make
use
of
a
docker
image,
as
we
say,
add
a
new
app
service.
B
We
grabbed
the
the
web
app
template
I'm
going
to
go
through
these
couple
of
steps
really
quickly,
because
the
goal
is
to
show
the
container
registry
integration
come
in
here
and
I
mark
the
same
subscription,
so
I
can
find
it.
I
go
and
I
say:
I
want
this
publish
to
come
from
a
doctor,
docker
image.
There's
a
configure
container
option
down
here
and
I
can
actually
make
this
a
little
bit
bigger
for
people
a
little
too
big.
B
Configure
container
and
then
it
will
come
over
here
and
I,
say
I
want
as
your
container
registry
and
if
you
remember,
we
called
that
the
clout
to
will
stand
up
registry
and
there
it
is,
and
that
was
created
just
as
we
were
doing
it
and
then
for
the
image,
and
the
tag
will
actually
have
to
wait
for
that
publish
to
finish.
But
that
is
how
you
would
be
able
to
list
them
and
set
them
up.
So
so.
A
I
think
what
that
means,
if
I
pin
myself
to
latest,
for
example,
anytime
I,
then
want
to
update
that
tag.
If
I
push
an
update
to
the
latest
tag,
the
service
will
automatically
then
pull
that
image
tag
correct,
so
anytime
latest
gets
updated.
It's
gonna
know
that
and
it's
gonna
automatically
pull
that
latest
updated.
So
I
don't
have
to
come
back
in
correct.
B
A
And
I
think
so
in
this
particular
case
is
worth
pointing
out
that
you
used
app
service
as
an
example,
because
app
service
is
probably
generally
the
easiest
place
to
get
started
with
an
azure
for
all
of
this
stuff,
but
you
could
do
it
in
any
service
that
supports
containers.
The
great
thing
about
containers
is
their
standard
unit
of
delivery.
Docker
doesn't
care.
A
What's
inside
the
container
you
can,
it
could
be
node,
it
could
be
go,
it
could
be
Python,
it
can
be
dotnet
and
it
just
knows
that
it's
gonna
run
a
container,
and
so
you
do
anything
once
you
get
it
up
into
that
container
image
or
container
registry
that
you
should
publishing
from
Visual,
Studio
I
could
say:
I
want
to
have
a
crew,
Brad
ease
cluster
and
I
could
drive
it
from
the
azure.
Cli
and
I
can
just
point
at
in
the
manifest
file.
A
A
Once
you
get
into
a
container
registry,
you
can
put
it
on
any
hosting
environment,
app
service,
obviously
being
the
easiest
example
to
get
started
with
perfect,
so
I
think
the
last
thing
that
would
be
worth
briefly
just
talking
about
a
little
bit
because
it
comes
into
play
when
we
talk
about
the
differences,
the
services-
and
you
hinted
at
it
a
little
bit
when
you
were
talking
about
how
to
add
container
support,
is
the
idea
of
container
orchestration
mm-hmm.
So
what
does
an
Orchestrator
so
we're?
A
Obviously
in
Visual
Studio
we
have
the
option,
for
your
just
add,
docker
support,
which
does
this
like
just
docker
eyes
as
my
project,
but
there's
another
option
that
says
orchestration
support.
So
what
does
it
work?
Is
the
difference
in
dock
rising
a
single
project
and
working
with
an
Orchestrator,
so
the
orchestrator
will
let.
B
You
so
there's
there's
two
of
them:
one
is
the
kubernetes
helm
and
the
other
one
is
the
docker
compose.
So
docker
compose
will
actually
create
another
project
in
your
solution
and
let
you
gather
up
the
contents
from
multiple
projects
and
put
them
together,
even
if
they
don't
have
a
direct
dependency
where
1x1
has
a
dll
and
another
one
has
a
dll
and
there's
a
project
reference
between
them.
A
B
That's
that's
the
that's
from
my
particular
experience.
More
the
other
end
of
things,
which
is
I,
have
a
set
of
machines
that
have
different
roles.
I
want
those
different
roles
to
have
different
images.
I
want
the
different
roles
to
be
mapped
to
effectively
different
docker
images
and
tags,
and
the
orchestrator
is
the
thing
that
says:
I
need
three
machines
of
this
role
with
this
image.
With
this
tag,
I
need
five
of
this
and
I
want
to
make
sure
I
monitor
the
health
of
them
and
here's
how
I
check
the
health
yeah.
A
Absolutely
yeah
your
crabs
li
correct,
if
that's
all
built
in
I,
think
so.
Yes,
you,
it's
an
and
not
not
not
a
not
an
or
the
the
two
aren't
contradictory,
I.
Think
from
a
sort
of
a
basic
benefits
perspective
like
that's
a
very
operational
you're,
absolutely
correct.
I
was
thinking
one
of
the
more
sort
of
tangible
benefits
for
why
I
might
want
to
use
container
orchestration
even
locally
on
my
local
machine
is
it
enables
multiple
containers
to
start
and
then
enables
them
to
discover
each
other
via
URLs.
A
So
if
I
won't
have
a
you
know,
web
front-end
and
a
Web
API
back-end
I
can
just
call
it
Web,
API
back-end
and
when
I
call
that,
if
that's
how
I
try
to
call
if
my
web
front-end,
the
orchestrator
will
do
the
dns
resolution
for
me
and
let
me
discover
that
and
it'll
to
your
point.
If
I
have
multiple
instances
set
up,
you
know
it'll
automatically,
handle
the
load,
balancing
and
routing,
and
things
like
that.
A
But
the
nice
thing
from
a
developer
perspective
instead
of
having
to
hard
code
or
add
URLs
and
to
config
somewhere,
it's
just
relative.
So
if
I'm
developing
locally
and
doing
name
based
resolution
behind
doesn't
matter
if
it's
docker
compose
or
mini
coop
for
local
kubernetes
home
when
I
deploy
that
to
production,
nothing
about
my
code
or
configuration
needs
to
change,
assuming
that
the
relative
names
of
the
containers
stay
the
same
so
the
services,
so
that.
A
Is
a
really
nice
benefit
that
orchestration
and
then,
as
you
said,
all
the
operational
benefits
so
I
think
operationally.
Yes
to
everything
you
said
locally,
if
I'm
developing
the
reason
I
may
be
interested
in
instead
of
going
to
the
single
image
thing
doing,
the
compose
flavor
is
if
I'm
trying
to
have
a
couple
services
talk
to
each
other.
I
can
do
that
in
exactly
the
same
way.
Locally
I'd
be
doing
that
in
production.
A
So
I
think
I
don't
see
any
more
questions
online
right
now,
we'll
give
people,
maybe
about
30
seconds,
to
type
one
in
if
they're
they
have
anything
else,
they'd
like
to
ask
but
other
than
that
I
think
it's
we're
kind
of
approaching
the
in
it
near
the
end
of
the
hour.
So,
let's
plan
on
wrapping
it
up
I'd
like
to
thank
Phil
for
joining
me
today
for
spending
the
time
to
to
chat
with
me.
Thank
you
for
the
opportunity.
I.
A
Think
one
of
the
opportunity,
one
of
the
great
things
about
stand-up
is,
you
know,
kind
of
an
informal
thing.
You
know
tend
to
be
Microsoft
conferences
or
more
formal
things.
We
tend
to
send
oftentimes
the
same
people
all
the
time.
We're
stand-up
gives
us
the
opportunity
to
get
some
people
who
are
maybe
more
behind
the
actual
bits.
Phil
does
all
of
the
hard
work.
I
just
get
to
stand
on
stage
and
take
credit
for
it
more
regularly
haven't
seen
any
more
questions
come
in,
so
with
that
we
will
call
it
a
wrap.
Thank
you.
Everybody.