►
From YouTube: Introduction and Deep Dive Into Containerd - Kohei Tokunaga & Akihiro Suda, NTT Corporation
Description
Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon North America 2021 in Los Angeles, CA from October 12-15. Learn more at https://kubecon.io The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.
Introduction and Deep Dive Into Containerd - Kohei Tokunaga & Akihiro Suda, NTT Corporation
Join containerd maintainers and reviewers in a combined introduction and deep dive session. They will discuss the overview and the recent updates of containerd as well as how it is being used by Kubernetes, Docker and other container-based systems. The brief introduction about its architecture and service design will be included. The talk will also deep dive into how to leverage contained by extending and customizing it for your use case with low-level plugins like remote snapshotters, as well as by implementing your own containerd client. Upcoming features and recent discussion in containerd community will also be covered.
A
A
Quantity
is
a
consumer
runtime
project.
It
is
the
fifth
cncf
graduated
project
containing
manages
resources
about
containers,
including
container
processes,
images
file,
system,
metadata
and
resource
dependencies.
Then
it
provides
clean
interface
to
upper
tools.
I
will
introduce
internal
architecture
of
continuity.
Data
as
well.
A
A
Quantity
is
now
widely
used
by
container-based
tools
in
community,
including
kubernetes
and
docker.
Let's
now
focus
on
the
adoption
status
in
community
from
the
next
slide
container
is
widely
used
in
community,
including
docker's
use
of
continuity.
Continuity
is
83
percent
of
container
usage
according
to
the
survey
in
2021
container
is
used
by
managed
services
as
well
as
open
source
projects,
including
managed
services
like
gke
aws,
fargate,
aks,
ats
development
tools
like
docker
mobile,
build
kit,
humanities
distributions,
like
k3f
kind,
mini,
cube,
cube
spray
micro,
ks
and
k3s,
and
functional
service
like
fasti.
A
A
A
A
A
Continuity
is
also
used
as
general
control
management
tool.
Several
applications
are
developed
based
on
continuity.
They
include
build
git,
fasti,
nod,
color
and
porch
container
continuity.
Api
is
commonly
used
by
these
tools
containing
provides,
go
client,
library
and
utilities
for
continuity-based
content
management.
A
Applications
can
use
them
for
integrating
to
quantity.
Quantity
can
also
be
extended
using
plugins
without
recompilation,
as
discussed
later,
as
shown
in
the
previous
slides
containing
manages
images,
containers
and
parts.
So
how
is
it
done?
Let's
take
a
look
at
quantities,
internal
components,
contradi
has
client
server
architecture,
it
provides
apis
and
content
management
functionalities
to
the
client
continuity
project
provides
core
client
library
for
easily
integrated
to
content.
A
A
It
also
supports
non-oci
runtimes,
including
firecracker.
Continuity
is
highly
extensible,
so
users
can
customize
its
functionality
using
various
low-level,
plugins
custom
services
or
custom
client
library.
We
will
discuss
about
extensibility
in
more
detail.
Later.
Quantity
project
provides
quantity,
api,
client
library.
A
It
is
called
smart
client
and
contains
rich
container
related
utilities,
including
quantity,
api
bindings
container
registry
client
pulling
and
pushing
images,
image
unpacked
and
creating
osa
config
for
oci
runtimes
using
the
client
library.
There
are
several
container
declined.
Implementations
in
community
ctl
is
a
cli
client
for
quantity.
This
is
developed
in
contently
project.
It's
like
a
content.
Api
wrapper.
Instead
of
control
management
utilities,
it's
mainly
used
for
developing
of
content
or
training.
New
features.
A
We
will
discuss
about
narrow
cutter
in
more
detail
later
with
akihiro,
as
mentioned
in
previous
slides.
Various
quantity-based
tools
are
developed
as
contrary,
client
in
community
leveraging,
client
library
and
continuity,
api
content,
details
or
core
provides
content
management
functionality
to
the
client
via
glpc
api.
A
The
content
decor
is
implemented
as
a
set
of
microservices,
and
the
set
of
apis
of
these
services
are
provided
to
the
client
as
content.
The
api,
for
example,
as
shown
in
the
figure
there
are
several
services
like
container
service
for
contaminated
management,
image
service
for
image,
metadata
management
task
service
for
performing
container
execution
and
cri
service
that
implements
cri
among
these
services.
A
There
is
a
shared
metadata
underneath
it's
bible-based
key
values,
tool,
it
stores
metadata
of
contents,
images,
contents,
file,
system,
snapshots,
etc.
It
also
manages
reference
graph
of
these
resources
for
performing
garbage
collection.
Cri
for
kubernetes
is
also
implemented
as
a
micro,
servicing
quantity.
A
A
Cni
plugin
is
used
for
network
interface,
preparation
for
pods
and
an
arrive
plugin
is
used
for
managing
node
resources
like
cgroup,
because
cri
service
relies
on
other
services.
It
contains
contaminated
client,
calls
for
talking
to
other
contender
services
and
plugins
based
on
these
services.
Sara
implements
parts,
contents
and
images
management
in
content
decor.
There
are
several
services
that
provide
low
level
role
of
control
management.
A
Content
store
is
a
store
for
image
content,
including
manifest
and
layers.
It
stores
these
contents
as
its
bases,
so
without
the
compression
neural
extraction.
These
contents
are
content,
addressable
and
killed
by
the
content.
Digest
snapshotter
also
manages
image
contents,
but
it
focuses
on
the
file
system.
Snapshots
snapshots
are
extracted
in
stacked
view
of
continuous
root
file
system
layers.
A
Snapshots
are
created
by
snapshot
and
passed
to
ocl
runtimes
as
root
file
system
of
contents.
There
are
several
snapshot:
implementation
per
packing
file
system
overlay,
fx
snapshot
is
one
of
the
most
used
one,
but
there
are
also
other
implementations
like
butterfa
snapshooter,
a
uefa,
snapshotter
and
fused-based
snapshot,
etc.
A
Runtime
service
executes
low
level
runtimes
via
sim
shimmy
is
a
wrapper
demo
for
ocr
runtime
and
manages
the
container's
lifecycle
and
logging
different
from
one
short
oci
runtime.
Cls
sim
is
long
learning
process
that
has
the
same
lifetime
with
the
container,
so
it
will
also
do
runtimes
that
need
to
manage
stateful
resources
like
virtual
machine
of
cattle
containers.
A
A
Registry
related
utilities
are
implemented
in
the
contender's
smart
client
library,
once
the
client
pulls
image
contents.
It
stores
the
contents
to
contain
the
course
content
store
as
its
basis.
So
without
decompression,
your
extraction
image
layer
contents
stored
in
the
content
store,
are
decompressed
and
extracted
by
this
service
and
the
stacked
boot
file
system
view
is
managed
by
snapshot.
A
Client
library
provides
unpacker
utility
for
this
process,
then
snapshots
are
used
by
task
and
runtime
service
and
passed
to
oci
runtimes,
and
they
are
used
as
continuous
root
file
system
task
service
can
be
used
via
quantity.
Api
bindings
by
the
client,
as
discussed
in
previous
slides
continuity,
is
highly
extensible.
A
A
Another
is
a
plugin
as
an
executable,
binary
string,
processor
and
sim
are
disabled,
plugins
quantity.
Api
is
also
extensible
by
implementing
their
own
custom
service.
For
example,
file
clock
quantity
has
new
apis
called
control
api
by
their
own
control
service.
From
next
slides.
Let
me
introduce
some
examples
to
extend
constantly
using
these
plugins
using
plugins.
You
can
enable
lazy,
blink
on
quantity,
remote
snapshot.
Plugin
is
a
plugin
enables
this
feature.
A
Remote
snapshot
allows
quantity
to
perform
lazy
playing
of
images
from
arbitrary
remote
store.
Lazy
playing
here
means
quantity
can
start
up
contrast
without
waiting
for
the
entire
image
contents
being
locally
available.
Instead
necessary
chunks
of
image
content
are
put
on
demand
as
soon
in
the
following
figure.
Remote
snapshot.
Plugin
has
a
responsibility
to
fetch
image
contents
from
the
back-end
store
and
provide
continuous
root
file
system
snapshots
to
quantity,
as
mentioned
in
the
previous
size,
snapshotter
can
plug
into
quantity
via
unique
socket,
so
no
recompilation
is
required.
A
There
are
several
remote
snapshot.
Implementations
in
community,
for
example,
strategy
snapshotter
is
a
remote
snapshot
developed
as
a
non-course
project
of
quantity.
This
enables
content
to
lazily
pull
images
from
the
standard
registry
content
handles
image
layers
for
creating
containers,
root
file
system.
The
image
layer
formats
are
defined
by
ocr
image,
spec
and
gzip.
That
standard
and
blender
layers
are
currently
supported.
However,
not
limited
to
ocl
layers
contradict
can
handle
actually
layer
formats
like
encrypted
layers,
even
if
they
are
not
supported
by
the
oci
image
spec.
A
A
This
service
converts
image
layers
downloaded
from
the
registry
into
root
file
system
snapshots
with
chaining
these
during
processors,
each
stream
processor
converts
the
media
type
of
the
layer
into
another.
Diff
service
can
handle
actually
media
type
of
layers.
As
long
as
the
corresponding
stream
processor
is
plugged
into
this
service.
A
A
Quantity
can
integrate
to
arbitrary
low
level
runtimes
low
level.
Runtimes
are
not
limited
to
oci
runtime,
but
no
oci
runtime,
like
firecracker,
can
also
integrate
to
continue.
The
plugin
enables
this
is
called
sim.
Sim
works
as
a
thin
wrapper
of
low
level
runtimes
each
low
level
runtime
can
integrate
to
continuity
by
implementing
their
sim
following
the
defined
sim
api
run
system
provided
by
continuity
project
also
supports
pluggable
logging
destination
feature.
A
B
B
B
However,
in
practice
you
have
to
use
go
language
for
the
controller
native
api,
because
the
native
api
model
depends
on
the
smart,
current
smart
client
library
that
is
implemented
in
go
wrong
ways,
especially
for
pulling
images
from
registering.
Currently,
it's
really
hard
to
use
other
languages
such
as
rust
or
python,
to
implement
your
own
client.
B
B
B
B
B
B
B
Neither
substructure
is
similar
to
surgical
substructure
and
enables
basic
print
as
well
as
denomination
of
the
images,
but
nadas
uses
a
file
format
that
is
incompatible
with
ocr
tables
and
alibaba.
Recently
released
overlay,
bdd,
samsung
programming,
pd
means
broke
devices
overlapping
is
similar
to
urban
effects
but
uses
icelandic
block
devices.