►
From YouTube: Where's My Container? Visualizing the GitOps Container Journey at Mi... Andre Briggs & Samiya Akhtar
Description
Don’t miss out! Join us at our upcoming events: EnvoyCon Virtual on October 15 and KubeCon + CloudNativeCon North America 2020 Virtual from November 17-20. Learn more at https://kubecon.io. The conferences feature presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.
Where's My Container? Visualizing the GitOps Container Journey at Microsoft - Andre Briggs & Samiya Akhtar, Microsoft
GitOps has become the de facto approach for enabling simple, secure, and auditable deployments. When you deploy to multiple Kubernetes clusters, knowing where your services are in the GitOps journey can be challenging. It can be almost impossible to track permutations of service versions across clusters with current tools available. In the session you will: • See a comprehensive dashboard view of a container’s journey from code check-in to deployment on Kubernetes clusters • Learn concrete patterns for testing in production with GitOps and deployment rings • See a new automation tool that can be used to deploy a seamless end-to-end GitOps workflow.
https://sched.co/Zes5
A
A
So
a
little
more
about
us,
my
name
is
samia.
After
I'm
a
software
engineer
at
microsoft
based
out
of
san
francisco.
I
work
in
the
commercial
software
engineering
team
here
at
microsoft,
and
this
is
where
my
journey
with
kubernetes
began.
I
am
passionate
about
open
source
and
visualizing
workloads
at
scale.
My
co-presenter
is
andre
briggs.
He
is
an
engineering
manager
in
the
commercial
software
engineering
team
here
at
microsoft
as
well.
He
spends
most
of
his
days
helping
partners,
develop
kubernetes-based
solutions
and
sharing
results
as
open
source
projects.
A
A
It
remains
the
single
source
of
truth
and
kubernetes
being
declarative.
It
tries
to
match
the
state
in
the
git
repository
at
all
times,
developers
are
familiar
with
git
and
they're
able
to
use
their
knowledge
to
modify
infrastructure
easily.
So
a
very
simple
github
process
would
look
like
this
here
in
this
diagram.
The
developer
pushes
code
to
their
git
repository,
which
kicks
off
a
ci,
build
that
pushes
the
new
image
to
the
container
registry
for
this
code
change
and
then
the
cluster
is
synced
to
a
config.
A
So
what
is
the
advantage
of
using
git
ops?
Githubs
is
a
simple
mental
model.
It
has
a
very
easy
learning
curve
because
developers
are
making
change.
Git
commits
and
pull
requests
to
make
configuration
changes
just
like
how
they
would
make
code
changes.
It
allows
for
a
consistent
and
standardized
end-to-end
workflows
with
strong
security.
B
B
B
We're
really
concerned
with
the
left
side
of
this
all
about
your
acting
and
your
releasing
being
able
to
know
how
long
something
takes
to
get
to
your
cluster
being
able
to
see
how
it's
going
towards
your
cluster
are
all
very
important.
This
is
the
area
that
we
want
to
focus
on
in
the
industry.
We've
seen
a
lot
of
challenges
around
this
particular
area.
B
B
Let's
say
you
have
lots
of
teams
working
on
one
cluster.
All
their
microservices
are
going
to
end
up
on
one
cluster.
In
the
real
world
teams,
work
in
branches
oftentimes.
You
will
go
and
deploy
your
your
applications
from
branches.
Before
you
go
to
your
main
branch,
if
you're
really
embracing
a
microservice
architecture,
you
most
likely
have
many
services
across
many
branches.
B
You
can
go
one
step
further
and
if
you're
attempting
a
multi-tenancy
solution
now
you
have
even
more
complexity.
For
instance,
some
teams
might
try
something
like
ringed
deployments,
or
they
might
do
some
sort
of
flighting
methodologies.
That
will
also
increase
the
complexity
of
how
you
release
your
software.
B
B
A
Fabricam
trivia
is
a
small
team
building
a
trivia
website.
They
would
like
to
increase
their
usage
by
integrating
several
apis
into
their
website,
which
are
being
built
by
separate
teams.
So
here
you
can
see
a
screenshot
of
what
their
trivia
app
currently
looks
like
they
have.
The
following
team
members
we
will
be
talking
about
their
developer
is
danny.
Their
product
owner
is
paul
and
their
operations
engineer
oliver.
So,
let's
see
what
their
scenario
looks
like
and
how
gitoffs
can
help
them.
A
So
until
now
they
had
a
monolith
application
and
now
they
would
like
to
modernize
their
setup
to
include
include
the
deployment
of
microservices.
So
the
main
idea
behind
a
microservice
architecture
is
that
applications
are
simpler
to
build
and
maintain
when
broken
down
into
smaller
pieces
that
work
seamlessly
together.
So
these
are
different
apis.
They
will
be
integrating
into
their
infrastructure.
A
A
So
fabricam
trivia
engineers
learn
about
git
ops,
and
now
they
are
starting
to
store
helm
charts
for
these
environments.
In
a
in
a
specific
folder
in
their
git
repositories,
the
cd
pipelines
will
generate
these
charts
into
kubernetes
manifests
and
flux
is
running
on
their
clusters,
which
will
sync
the
generated
manifest
to
their
cluster.
A
So,
as
you
can
see
in
this
diagram,
there's
quite
a
few
developer
teams
pushing
code
into
their
repos,
followed
by
ci
bills,
which
are
triggered
to
build
images,
and
then
they
also
have
a
group
of
approvers
who
approve
changes
to
be
deployed
into
the
cluster
using
the
cd
pipelines
which
generate
the
helm
charts.
So
this
architecture
has
finally
allowed
fabricam
engineers
to
incorporate
work
from
several
teams
together.
A
Now,
danny
and
oliver
run
into
a
few
problems
they're
finding
that
managing
multiple
micro
services,
pushing
into
the
same
cluster
is
starting
to
become
complex.
There
are
so
many
cicd
pipelines
and
tracking
is
difficult.
For
example,
each
microservice
has
a
pipeline
for
their
ci
another
for
their
cd,
resulting
in
2n
pipelines.
Already
developers
are
also
having
trouble
debugging
problems,
because
each
service
has
its
own
set
of
logs
oliver.
The
operator
is
lacking
observability
into
what's
getting
deployed,
versions,
audit
logs
and
has
trouble
fixing
issues
if
bugs
make
it
into
production.
A
So
fabric
and
trivia
engineers
would
also
like
to
test
their
features
in
a
single
cluster
to
cut
cost
because
they're,
a
small
company
they
learn
about
the
concept
of
deployment
rings
that
allow
multiple
environments
to
exist
in
a
single
cluster
with
the
help
of
a
service
mesh.
This
would
allow
them
to
test
their
features
in
production
within
the
same
cluster
by
setting
a
header
on
their
ingress
routes,
and
each
individual
developer
would
be
able
to
test
their
features
in
the
cluster
without
disturbing
the
default
production
ring.
A
A
A
A
A
B
B
It
allows
you
to
see
all
your
microservices
in
your
ci,
going
from
your
build
all
the
way
to
the
cluster
in
a
good
ops
fashion
with
spectate,
you
can
go
and
look
at
multiple
applications
at
once
across
multiple
teams,
even
across
multiple
branches
and
rings,
and
be
able
to
find
out
if
your
pr
is
being
blocked
where
it
is
and
so
on.
Let
me
quickly
show
a.
B
B
B
A
B
B
B
B
B
If
we're
doing
this
correctly,
we
can
actually
go
and
automate
a
pr
that
will
happen
after
we
go
and
make
a
code
change
for
our
docker
image.
This
will
be
great.
We
have
a
concept
of
something
called
a
high
level
repository
one
of
the
concepts
of
bedrocks
in
the
bedrock
family.
We
have
a
tool
called
fabricate
which
is
built
on
top
of
helm.
Fabricate
really
allows
you
to
deal
with
higher
level
abstractions
of
yaml
and
is
a
great
way
to
be
able
to
deal
with
multi-cluster
multi-region
setups.
B
This
graph
shows
what
what
bedrock
does
in
terms
of
helping
you
set
up,
get
ops
pipelines
allowing
you
to
see
more
observability
in
your
get
ups
process.
By
calling
bedrock
deployment
on
dashboard
also
show
some
commands
for
fabricate
another
tool
that
we
developed
in
the
effort
to
have
better
get
ops,
observability.
B
B
So
I
think
we
are
answering
some
questions.
I
see
one
from
hermann
banken
does
bedrock
also
work
without
azure.
Yes,
some
parts
of
it
work.
The
bedrock
cli
right
now
is
targeted
to
azure
devops,
but
we
you
know,
we
recently
saw
the
good
ops
toolkit
by
weaveworks
at
target's,
github
actions
and
get
lab,
and
I
think
that
we
can
make
some
contributions
there,
but
red
block
patterns
in
general.
You
don't
have
to
use
azure.
We
have
it
running
in
gitlab
and
jenkins
and
github
actions
in
terms
of
doing
it.
A
B
A
B
I'll
take
another
one:
how
does
how
is
this
different
from
release
pipeline
than
azure
devops
release
pipeline?
Well
in
azure
devops
in
particular,
there
is
a
you
know.
They
introduced
multi,
multi-stage
pipelines,
I
think
last
year
or
a
year
before,
and
so
I
think
that's
a
very
general
question.
The
difference
is
that
you
don't
have
to
go
to
a
separate
pipeline.
You
can
have
everything
in
one
so.
B
A
A
B
Cool
I'm
going
to
go
just
quickly,
answer
number
17
from
burkan.
Can
we
integrate
spectate
with
azure
devops?
I'm
not
sure
if
you
mean
inside
azure,
devops
currently
spectate
works
really
well
with
azure
devops.
It
just
runs
separately,
and
the
reason
for
this
is
that
sometimes
you
might
want
certain
people
to
have
access
and
if
it
runs,
for
instance,
as
a
docker
image
in
the
kubernetes
cluster
or
outside
you
can
control.
You
can
easily
control
who
sees
it,
but
you
could
probably
technically
create
a
web
view
inside
of
azure
devops.
A
B
I'll
take
number
19.,
so
how
configurable
is
the
spectate
dashboard
for
integrating
with
in-house
pipelines
I'll
say
it's
pretty
extensible
one
of
the
one
of
the
things
that
we
really
want
to
go
and
do
is
make
sure
that
you
know
folks.
You
know
different
projects
might
have
different
needs
for
what
appears
on
the
dashboard
and
so
within.
B
B
B
B
Some
of
the
things
we've
thought
about
is
some
some
folks
like
to
go
and
have
their
infrastructure
their
terraform
and
also
their
what
we
call
high
level
definition
of
bedrock
in
one
repository
this
way,
if
you
want
to
make
a
change
in
a
gaps
fashion,
that's
across
both
your
infrastructure
and
your
applications,
you
can
do
that
all
in
one
pr,
which
is
pretty
great,
so
I'll
say
that
you
know
we
support
things
like
that.
You
want
to
take
number
22
something.
A
Yeah
sure
number
22:
do
you
spin
up
ephemeral
clusters
for
integration
tests
is
flux
to
use
there
too,
I
would
say:
yes,
we
are
using
flux
in
the
integration
tests,
but
that
really
depends
on
how
you
set
it
up
and
that's
andre.
If
you
would
like
to
add
to
this,
I'm
not
fully
sure.
If,
if
I
understand
the
clusters.
B
B
We
haven't
had
to
go
all
the
way
to
that
that
level
of
testing
of
having
to
to
do
that.
I've
heard
the
people
doing
that,
but
I
think
mostly
we
push
things
to
the
cluster
and
then
we
we
have
basically
a
callback
to
make
sure
that
things
have
reached
the
cluster,
and
then
we
can
run
our
integration
test.
So
it's.
A
A
Yes,
so
actually
bedrock,
sorry
spectate
offers
you
to
integrate
multiple
services
and
when
you're
using
multiple
services,
you
could
have
multiple
repositories
and
you
are
having
the
option
to
filter
as
much
as
you
like
in
the
spectic
dashboard
itself.
So
the
status
is
clear,
depending
on
how
many
rows
you're
seeing
and
that
you
can
adjust
in
the
in
the
url.
B
Itself,
great:
let's
go
back
up
number
23..
Your
solution
does
not
take
care
of
infrared
as
code
right,
yeah
I'll
say
that
when
it
comes
to
infrastructures
code,
you
can
treat
that
as
good
ops
as
well.
I
think
that
yeah
we're
more
focused
on
on
the
application
level,
rather
than
your
first
code,
because
really
in
the
future.
I
think
that
you
know,
as
long
like
there
are
many
different
ways
to
set
up
your
infrastructure
in
azure
we
have
arm,
I
mean,
there's
infrastructure's
code
is
terraform.
B
As
long
as
you
have
your
infrastructure
and
we
think
that
arm
or
terraform
is
a
great
way
or
if
you
have
another
way,
then
you
know
we
want
to
sort
of
get
people
going
from
that
stage
because
oftentimes
in
many
organizations
the
people
who
control
the
infrastructure
are
different
than
the
developers,
because
there's
a
cost
factor
with
the
infrastructure.
And
so
you
know.
A
B
It
number
26,
how
do
you
know
when
flux
applied
the
new
manifest
to
the
kubernetes
cluster?
Oh,
so
I
mentioned
a
callback.
If
you
look
at
flux
there
actually
is
a
some
extensions.
You
can
do
to
flux.
I
think
that
for
weave
cloud
or
something
they
provide
a
web
hook,
and
so
you
can-
and
there
are
events
now
using
prometheus,
where
you
can
tell
whether
or
not
stuff
was
applied
correctly.
B
B
B
B
A
B
B
B
A
B
Repository
the
difference
between
spectate
and
bedrock
number
36
is
spectate
is,
is
a
standalone
docker
image
that
goes
is
a
standalone
docker
image
bedrock
is
more
of
a
pattern
than
in
cap,
and
we
have
a
bedrock
cli,
which
is
a
concrete
implementation
of
that
pattern.
In
terms
of
creating
standing
up,
get
ops
pipelines
and
so
spectate
is
its
own
thing
for
visualization.
A
B
B
Yeah,
I
think
we'll
probably
just
have
one
minute
left
so
is
get
off's
the
best
way
to
deploy.
That's
that's.
That's
an
opinion.
I
think
that
get
ops
is
really
something
that's
designed
for
operators,
and
so,
if
you
are,
if
you
have
people
that
are
operating
something
and
you
want
to
make
it
really
simple
yeah,
I
think
that
we,
you
know
we
can
continue
this
discussion,
but
I
think
that
get
ops
has
a
lot
of
advantages
compared
to
other
other
ways.