►
From YouTube: Kubernetes WG IoT Edge 20230111
Description
January 11, 2023 meeting og the CNCF ioT Edge Working Group. General discussion of group activity including edge native white paper, KubeCon Europe plans, efforts to catalog edge native related open source projects.
A
Hi
welcome
to
the
January
11
meeting
of
the
cncf
iot
edge
working
group
for
today's
meeting.
It's
going
to
be
relatively
informal
because
I
don't
think
we
have
much
on
the
agenda.
We
do
have
a
item
being
added
now
to
review
the
Edge
native
white
paper
day
on
verbally
submitted
a
topic
which
is
to
discuss
the
we've
been
invited
to
propose
a
talk
for
kubecon
Europe.
That
I
believe
kubecon
Europe
is
in
April
am
I
right,
yeah,
yeah,
I.
C
Okay,
yeah
sounds
good
I'll
go
ahead
and
share
my
screen
and
what
I'm
sharing
is
shortly
before
the
break?
I,
let's
see
here,
I
think
that's
the
window.
Yeah
and
I'll
put
a
link
to
this
stock
in
the
chat
as
well.
It
should
be
visible
and
editable
by
everyone
so
before
the
break,
I
moved
what
or
I
should
say
I
copied.
C
What's
in
the
GitHub
repo
into
a
Google
share
doc,
because
I
thought
that
would
be
easier
in
terms
of
being
able
to
review
The
Dot
comprehensively
and
provide
an
overall
General
edit
for
readability,
completeness
correctness,
Etc,
so
I
got
started
during
the
break
and
then
ran
into
holiday
activities.
But
then,
just
yesterday
got
around
to
completing
my
edit.
So
in
this
stock,
I
marked
all
edits
as
suggested
so
that
everyone
can
see
what
they
are.
C
Sometimes
less
is
more
so
in
there
in
occasions
where
I
thought
we
were
maybe
using
extra
words,
I
I
I
narrowed
that
down
a
little
bit,
but
I
think
it
was
helpful
to
to
just
read
it
from
start
to
finish
again
with
fresh
eyes
and
taking
a
step
back
from
being
deep
into
the
dock
like
we
all
were
for
several
months
and
I'll
go
through
the
edits.
Just
in
a
moment.
There's
there's
not
that
many,
but
once
we're
all
okay
with
the
final
product.
C
What
I'm
recommending
we
do
is
make
these
edits
in
GitHub,
first
of
all,
and
then
secondly,
take
this
draft
back
to
the
cncf
marketing
working
group
marketing
committee,
who
I've
I've
been
a
part
of
now
for
for
several
months,
they've
expressed
a
willingness
to
help
us
co-promote
this
this
paper.
Now
that
it's
reached
this
state
of
what
we're
calling
I
guess
complete
for
now,
we
want
to
you
know,
highlight
the
good
work
that
we've
done.
We
want
to
raise
awareness
across
the
cncf
community,
as
well
as
the
industry
around
these
educative
principles.
C
C
C
So,
as
you
can
see
a
really
light
edit,
but
that
said
I
would
appreciate
one
or
more
people
from
this
group
to
give
those
edits
the
sanity
check
and
make
sure
that
I
haven't
written
over
anything
important
or
changed.
The
meaning
of
anything
in
my
edits
and
reviewers
can
feel
free
to
accept
or
reject
the
edits
as
as
they
see
fit
here
at
the
bottom,
I
updated
the
language
a
little
bit
for
how
to
get
involved.
C
So
we
are
reminding
folks
to
we're
inviting
them
to
come
to
this
group
to
present
their
Edge
related
projects
and
to
bring
ideas
to
this
group
as
the
the
body
that's
creating
and
evolving
this
paper
and
also
participate
in
a
future
iteration
of
the
paper
or
follow-up
papers
down
in
the
author
list.
I
changed
the
order
here
to
make
it
alphabetical
by
company
I
think
that's
usually
how
the
Linux
Foundation
does
it,
and
then
we
have
our
references
and
then
it's
that's
it.
A
D
C
A
Don't
know
it
might
just
be
me,
but
you
know
it
just
sometimes
seems
Promotional,
and
you
know
you
can
kind
of
draw
these
numbers
on
edge
or
kind
of
any
other
topic
in
Tech.
If
you
look
hard,
if
you
unleash
the
Google
search
engine
right,
you
can
find
random
hockey.
Stick
growth
forecasts
on
just
about
anything.
A
Some
of
them
tend
to
be
better
than
others
over
time.
If
you
go
back
and
review
them,
yeah
I'm
not
sure
that
this
one.
How
did
you
come
up
with
this
particular
one
versus
others?
And
you
know
I
I
think
that
there
are
some
that
come
right
out
of
cncf
materials.
That,
maybe
are
you
know,
kind
of
more
vendor
neutral
and
I.
Wasn't
familiar
with
this
one
I'm,
not
saying
it's
good
or
bad.
It's
just
the.
C
C
Know
it's
been
maybe
five
months
since
we
put
our
initial
draft
together,
so
I
I,
don't
recall
the
reasoning
behind
it.
I
guess
the
question
is:
do
we
need
a
a
sentence
that
reminds
the
audience
of
the
the
rapid
projected
growth
of
this
technology
or
if
we,
if
we
do
feel
it's
necessary?
Maybe
we
look
to
get
something.
C
D
Yeah
I'm,
actually
the
one
that
put
that
in
there
and
the
reason
that
I
had
initially
put
that
in
there
is
because
we
were
talking
about
the
definition
of
Edge
and
I,
also
wanted
to
just
kind
of
highlight
where
it
was
going
and
the
growth
in
the
industry,
okay
and
in
my
particular
role
at
VMware.
This
is
one
of
the
areas
that
I
actually
work
in,
and
so
that's,
why
I
put
it
in
there.
But
if
you
guys,
don't
think
it's
necessary,
then
feel
free
to
take
it
out.
A
E
E
A
C
Happy
to
to
discuss
any
other
thoughts
as
it
is
now,
but
what
I'd
like
to
do
is
is
allow
folks
to
maybe
go
through
and
review
each
one
of
these
changes
in
a
little
bit
more
detail
if
this
group
would
be
willing
to
spare
some
time.
So
how
about
we
give
folks
until
let's
say
the
end
of
this
week,
to
make
any
additional
tweaks
as
needed.
Otherwise,
I'll
accept
all
these
changes.
C
Turn
this
over
to
the
the
marketing
committee,
look
to
get
it
updated
on
GitHub
and
then,
as
their
promotional
Cycles
develop,
and
this
gets
inserted
into
it.
I
can
periodically
report
back
to
this
group
on
when
to
expect
the
the
pdf
version.
I
suspect
people
on
this
call
would
find
that
a
useful
document
as
well
to
to
share
amongst
your
teams
or
in
your
own
communities
as
well,
and
then
yeah.
We
can
look
to
see
where
it
where
it
goes
from
there.
C
So
it's
exciting
to
see
kind
of
the
full
evolution
of
what
we've
been
working
on
and
I'm
glad
that
we
we
seem
to
have
found
a
a
niche
that
needed
some
filling
in
terms
of
definitions
and
Leadership
and
Innovation.
So
we're
close
to
calling
this
done
for
now
and
then
yeah
we'll
see
what
we
need
to
do
with
this
into
the
future.
A
Let
me
just
close,
and
let's
verify
that
is
everybody
on
this
call
is
end
of
this
week
enough
and
we
do
have
some
people
who
couldn't
make
this
meeting
like
Kate.
So
maybe,
if
you
wouldn't
mind
putting
a
note
out
on
the
slack
Channel
indicating
the
same
thing
just
in
case,
somebody
didn't
make
this
meeting
and
isn't
aware
that
this
happened
and
it's
going
on
sure
and
anyway
last
call.
A
C
A
Okay,
moving
on
to
the
agenda,
I
put
the
item
of.
We
touched
on
it
briefly
at
the
open,
but
the
kubecon
maintainer
track
talk
proposal.
Dion.
Did
you
have
any
specific
ideas
of
what
direction
you
wanted
to
go
with
that
or
anybody
else
who's
on
this
call.
If
you've
got
anything,
you
know
any
any
ideas
that
you
personally
would
like
to
work
on,
or
maybe
just
throw
it
out
there
for
others
to
put
together
the
talk
proposal,
but
things
that
you
would
like
to
see
as
a
presentation
yourself.
B
So
so,
from
my
perspective,
it
was
just
the
first
idea
that
I
come
come
up
with
like
trying
to
get
all
these
principles
detailed
in
the
in
the
paper
into
something
that
is
more
more
more
practical,
yeah
and,
and
it
was
open
so
I'm
not
really
sure
how
much
time
commitment
I
would
have
to
actually
do
do
a
a
proper
bill,
but
I
I
think
we
can
still
go,
go
with
it
in
some
sense.
If,
if
there's
no,
no
other
other
proposals,
so
Kato
also
suggested
panel
kind
of
session.
C
F
A
F
A
A
A
A
So
we
do
have
a
few
weeks,
but
this
is
probably
something
we
need
to
start
moving
on,
because
you
know
you
it
isn't
going
to
come
together.
Well,
if
we
procrastinate
and
wait
for
the
last
day,
so
maybe
anybody
interested,
let's
start
putting
together
a
chat
on
this
in
slack,
to
try
to
move
this
forward.
A
A
D
A
People
that
if
they've
got
anything
that
they
work
on,
that
they
use
whatever.
Let's,
let's
try
to
get
the
things
in
there,
because
I
think
we
can
perform
a
service
to
the
community
by
putting
together
this
list.
It's
actually
like
I've
I've,
come
across
instant
incidents
where
I'm
looking
for
a
solution
to
a
problem
and
having
one
place
to
look
rather
than
Googling
for
things
like
I,
don't
know
iot
device
management
or
something
that
would
manage
firmware
updates
at.
F
A
Is
something
that
you
can
spend
a
few
hours
on,
because
there
doesn't
seem
to
be
a
good,
comprehensive
source
of
where
to
go
to
the
closest
thing
to
it
is
selected
foundations
like
Linux,
Foundation,
LF,
Edge
and
cncf
do
have
a
portal
where
they
list
the
projects
that
they
maintain
but
kind
of
this
overall
cross
industry
across
the
world.
Catalog
of
these
I
haven't
come
across
it.
It
would
be
nice
if
it
existed,
and
maybe
this
is
an
opportunity
for
this
group
to
perform
a
useful
service
by
putting
something
in
place.
F
I
think
this
sounds
great
I
I.
Definitely,
second,
the
comment
you
made
about
it
being
difficult
to
find
a
master
list
of
what's
out
there.
One
other
thing
I've
found
when
looking
at
these
projects
is
often
it's
hard
to
understand
what
applications
would
sit
on
top
of
these
Edge
systems
and
which
systems
would
be
better
suited
for
one
type
of
application
or
another.
So,
to
cite
a
specific
example,
egex
Foundry,
for
example,
has
more
of
like
an
industrial
consumer
slant
where
you
need
to
do.
F
Protocol
translation,
whereas
something
like
acri
is,
is
better
focus
for
incorporating
devices
that
may
not
be
part
of
the
cluster,
where
you
need
to
run
a
service
on
the
cluster
in
order
to
accommodate
those
third-party
devices,
and
if
there
was
some
way
we
could
capture
that
essence
of
the
of
the
use
case
or
the
application.
I
think
that
could
be
really
useful
to
people
yeah.
A
A
The
original
principle
of
Unix
was
that
you
had
tools
that
do
one
thing,
but
some
of
these
projects,
like
you,
brought
up
at
Jake's,
Foundry,
they're,
relay
rappers,
and
the
project
itself
includes
a
lot
of
specialized
tools
but
really
The,
Edge
X
boundary
itself,
I
view
as
an
integration
of
a
collection
of
tools
more
along
the
lines
of
hey.
It's
a
Swiss
army
knife.
A
If
you
will
versus
some
of
these
being
purpose
built
for
one
application
and
I
think
where
we'd
go
once
we
had
the
list
is,
we
would
break
them
down
into
these
categories
and
there
is
usefulness
to
the
Swiss
army
knife
idea
that
you
know
if
you're,
a
user
and
you've
just
got
a
problem
to
solve
up,
and
you
want
to
go
to
one
place.
You
may
be
the
effort
of
choosing
best
of
breed
fork
and
best
of
Reed
knife
and
then
integrating
it
together,
lays
a
big
job
of
that
integration
of
gluing.
A
These
pieces,
together
on
you
and
going
after
kind
of
a
comprehensive
Suite
of
things
that
have
been
designed
together
and
are
tested
together,
does
have
some
attractive
and
I
think
that
it
that
where
this
will
fall
once
we
get
the
list
is
that
we
will
have
some
of
these
projects
that
turn
out
to
be
more
like
integration
projects.
If
you
will,
you
know
there
are
some
of
these
that
integrate
a
bunch
of
components
at
Edge
and
also
some
that
integrate
Management
in
a
public
cloud
with
Edge,
Leaf
nodes
and
another
form
of
integration.
A
That
is
indeed
useful.
I
mean
you
know
it's
useful
because
there's
so
many
of
them
out
there.
The
other
interesting
thing
that
I
think
would
fall
out
of
this
is
that
I
think
one
of
the
one
of
the
fallouts
of
there
not
being
this
comprehensive
list
is
I'm
afraid
that
maybe
a
neutral
outside
Observer
could
come
to
the
conclusion
that
there's
been
a
lot
of
reinvention
of
the
wheel
going
on
here
that
maybe
five
people
went
and
did
effectively
independent
Solutions
of
the
same
problem
and
had
they
been
aware
that
these
other
things
existed.
A
There
might
be
an
opportunity
to
merge
these
independent
projects
together
or,
if
you
were
an
end
user
you'd
want
to
sort
of
evaluate
these
to
pick
and
choose
which
one
of
these
you
thought
was
best
match
to
solve
your
problem
and
kind
of
looking
ahead
on
the
chess
board.
Two
or
three
moves
ahead,
I
think,
once
you
have
the
list
and
categorize
it
you,
maybe
even
could
do
the
world
some
good
by
doing
kind
of
some
letter
grade.
A
Just
because
you
don't
want
to
you
know
in
the
in
the
current
state
of
the
world,
with
security
risks,
kind
of
anything
that
nobody's
paying
attention
to
in
a
sad
dormant.
For
you,
a
couple
of
years,
just
smells
to
me
like
security
incident,
waiting
to
happen
and
I.
Think
I.
Think
that
there's
a
lot
of
Runway
here
for
taking
this
idea
of
the
catalog
and
advancing
it
to
to
be
a
really
useful
source
for
people
working
in
this
space.
Yeah.
F
I
totally
agree
with
that
I
mean
my
personal
experience
was
last
fall.
I
came
across
k3ai,
which
sounded
like
a
great
Edge
type
application
to
to
deploy
machine
learning
using
k3s
to
to
Edge
devices,
but
just
as
you
said,
I
think
there
was
a
lot
of
Gusto
at
the
beginning
of
the
project
and
then
some
of
the
some
of
the
involvement
trailed
off
over
2022.
So
when
I
came
about
came
to
the
project,
I
found
that
things
didn't
work
and
being
able
to
provide
some
of
that
insight
to
people
would
help
save
them
some
time.
F
My
suggestion
is,
we
might
want
to
consider
an
approach.
That's
not
going
to
be
too
judgmental
for
smaller
fledgling
projects
so
that
we
kind
of
set
appropriate
expectations.
So
a
small
team
of
like
two
or
three
people
don't
get
overwhelmed
by
influx
of
people
suddenly
paying
attention
to
their
project.
We
need
to
find
some
way
of
like
indicating
the
scale
and
Grandeur
of
the
projects
that
they're
trying
to
accomplish
yeah.
A
Yeah,
the
other
thing
is,
if
we
went
so
far
as
to
sort
of
give
grades
on
the
health
of
a
project
that
is,
you
know,
that's
an
opportunity
where
some
people
might
take
offense,
and
it
would
be
key
that
these
not
be
a
tent
opinions.
They
have
to
be.
You
know,
objective
metrics
things
like
you
know
how
many
committers
have
have
been
active
in
the
project
in
the
past
year,
or
you
know
it's
the
same
criteria
actually
that
the
cncf
uses
to
allow
something
to
become
sandbox
or
incubating
as
a
project.
A
They
have
published
standards
for
some
of
the
health
metrics
they're
going
to
take
into
account.
You
know
they.
You
can't
take
any
random
piece
of
source
code
and
expect
the
cncf
to
accept
it
as
a
Sandbox.
There
has
to
be
a
you
know,
a
proposal
with
actual
metrics
collected
and
presented
and
I
think
something
like
that
is
the
direction
this
would
have
to
go
where
it
would
be.
So
you
know
there
are
a
number
of
numbers
that
you
can
get
on
GitHub
things
like.
A
How
long
does
it
take
for
PR's
to
be
accepted,
because
you
know
some
of
these
dormant
projects?
What
happens
is
people
will
even
submit
enhancements
to
it?
But
if
nobody
who
has
approval
authority
authority
is
paying
attention,
they
just
sit
there
aging
and
nothing
gets
done
and
I
think
there's
opportunities
to
go
somewhere
with
this
I
did
bring
out
the
idea
that
if
we
got
this
together
by
kubecon
Europe,
this
might
be
a
really
good
presentation.
A
F
I'd
be
happy
to
toss
in
my
hat
to
help
you
with
this
okay,
one
one
other
metric
that
I'd
recommend
that
might
be
contentious,
so
I
bring
it
up.
Now
is
I
often
look
to
see
if
there's
governance,
governance
that
goes
beyond
just
the
company
that
open
source
the
project
I
think
that
would
be
useful
for
people
to
see
if
others
have
gotten
involved
than
just
the
people
that
originally
wrote
it
yeah.
A
I
think
maybe
columns
that
would
be
interesting
are.
If
there
is
a
governing
body,
you
know
whether
it
be
cncf,
LF,
Edge
Eclipse,
it
should
be
listed
and
then,
if
there
isn't
many
times
maybe
most
times,
there
really
is
one
entity
whether
it
be
a
company
or
whatever
that
sponsored
this
and
I
think
that
sometimes
that
actually
isn't
easy
to
find.
A
You
know
in
a
commercial
sense
to
not
abandon
this
thing
and
there
are
some
who
maybe
don't
like
the
idea
of
it
being
corporate,
meaning
that
they
want
everything
to
be
free,
but
I,
think.
Another
attitude
is
that
you
actually
want.
You
know
you
know
any
software
engineer
has
to
eat
and
get
housing,
and
somebody
has
to
pay
for
that.
A
And
often
these
projects
don't
go
well
when
they're,
there
really
isn't
some
commercial
activity
associated
with
them,
too,
fund
the
operations
so
but
anyway,
if,
if
you
want
to
help
out
on
this,
this
is
great.
I
posted
the
link
in
the
chat
to
the
actual
spreadsheet
feel
free
to
add
columns,
or
you
know
we
originally
had
this
hope
that
maybe
it
was
a
little
naive
that
people
from
these
projects
would
discover
this
on
their
own
and
I.
Think
in
retrospect,
they
don't
even
know
we
exist.
A
So
I
think
that
was
the
last
item
on
our
agenda
and
historic.
We
generally
then,
just
turned
this
over
to
just
kind
of
birds
of
a
feather
discussion,
so
anybody
who
wants
anything
I
I'm
going
to
just
seed
this
birds
of
a
feather
but
just
bring
up
if
you've
come
across.
Anything
interesting
in
the
field
of
iot
Edge
and
one
thing
that
has
been
captivating
me
is
I've
just
been
hearing
all
of
this
I.
Don't
know
if
it's
hype
or
reality,
but
talk
about
this
chat,
GPT
being
the
hottest
thing
in
Tech.
A
Well,
I
came
across
people
who
were
talking
about
having
chat
GPT,
put
together
scripts
for
home
assistant
and
for
node-red,
which
is
kind
of
a
flow
based
development
tool
that
is
often
used
for
Edge
applications
with
devices.
I
think
Deon
you've
talked
about
code
red
in
the
past
in
presentations,
but
it
kind
of
surprised
me
that
people
reported
that
they
said
that
they
were
able
to
verbally
go
to
chat.
A
Gpt,
say
write
me
a
script
to
turn
on
the
light
when
the
temperature
drops
below
32
degrees
in
my
home
automations
system,
and
it
actually
worked,
and
that
just
kind
of
flabbergasted
me
that
they
said
that
it
came
up
with
like
either
work,
fully
working
or
close
to
working
code.
Examples
that
they
could
use
at
some
of
these
tools
now
I,
don't
know
that
anybody
in
industrial
iot
is
going
to
trust
that
with
Refinery
control.
But
often
these
things
sort
of
creep
up.
A
A
F
A
Yes,
certainly
that
I
think
that
was
one
of
the
major
underpinnings
with
why
the
move
happened.
I
think
we
we
had
quite
a
few
conclusions,
come
up
both
in
meetings
and
in
presentations
that
kubernetes
doesn't
solve
all
the
problems
and
part
of
the
reason
why
it
moved
over
is
an
acknowledgment
that
kubernetes
has
fairly
significant
resource
demands
for
low
resource
device
Edge,
for
example.
The
fact
is
on
some
of
these
devices.
They
really
can't
even
support
Linux,
much
less
a
container
runtime
and
then
loading
a
kubernetes,
even
a
thin
kubernetes
on
top
of
it.
A
So
the
move
was
in
recognition
of
that
I
would
contend
in
making
it
broader
based,
not
not
just
in
terms
of
orchestrators
but
Solutions
in
general,
so
this
whole
scope
of
it
being
our
constraints
are
in
my
mind
that
it
be
open
source.
You
know
we're
not
here
to
pitch
commercial
projects.
If
commercial
things
happen
to
have
an
open
source
dotted
line,
that's
fine
if,
but
this
is
open
source
focused.
But
if,
if
that
checkbox
is
done-
and
it's
add
your
iot,
then
it's
good.
F
A
Think
so,
and
once
again,
it
I
would
think
that
those
orchestrators
would
appear
in
that
list
of
Open
Source
projects.
So
I
can
tell
you
that
Nomad
is
Alive
and
Well
in
this
space
as
an
orchestrator
and
I
haven't
personally
used
it.
But
there
are
certainly
many
you
can
readily
just
search
for
YouTube
and
find
people
who
have
given
YouTube
videos
about
using
Nomad
for
Edge
use
cases,
and
it's
my
understanding.
A
It's
potentially
thinner
than
kubernetes
I,
believe
we've
had
a
presenter
in
this
group
from
poor
kainer
who
brought
that
up
in
a
presentation
on
portaner.
You
know:
portainer
supports
hosting
things
at
Edge
nodes
on
just
a
bear:
Docker
runtime,
with
Docker
compose
with
kubernetes
and
with
Nomad,
and
all
of
those
are
potentially
options
and
I
believe
that
that
speaker
suggested
that
Nomad
is
lower
in
resource
demands
than
kubernetes,
so
that
one
didn't
drop.
A
I
think
that
you
brought
up
Apache
mesos
I
think
that
one
is
effectively
dead
for
Edge,
it's
kind
of
losing
momentum,
but
that
one
to
some
extent
had
bigger
resource
demands
than
even
kubernetes.
You
know
it
was
similar
in
that
he
used
zookeeper.
Instead
of
that
CD,
but
the
Zookeeper
component
alone
was
pretty
resource.
Intensive
and
Landing.
A
A
full
mesos
cluster
at
an
edge
node
would
be
a
challenge,
to
put
it
mildly,
and
I
do
have
some
experience
with
deploying
mesos
clusters
so
I'm,
not
that
one
I'm
I'm
speaking
from
personal
experience
and
I
I,
think
even
the
company
behind
mesos,
which
was
originally
mesosphere
now
rebranded,
as
D2
IQ,
has
moved
on
to
being
kubernetes
specific,
but
in
terms
of
orchestrators
there's
anything
from
the
no
orchestrator
option
to
using
just
a
bare
Docker,
runtime
and
conceivably.
The
docker
runtime
could
be
driven
by
something
like
get
Ops.
A
So
in
a
way
you
could
have
a
supervisor
misery
system
populating
a
git
repository
that
really
is
driving
what
happens
at
Edge
nodes,
where
you
could
take
it
up
to
orchestrators
and
I.
Think
it's
a
pretty
Broad
and
open
area
at
this
point
in
time
where
there's
no
one
obvious
winner,
like
kubernetes
I,
would
contend,
sort
of
has
become
dominant
in
the
public
cloud,
a
large
data
center
space
as
an
orchestrator,
but
when
you
get
to
Edge
I
think
it's.
A
It's
still
Greenfield
other
things
that
are
out
there,
that
I'm
aware
of
the
web
assembly.
People
have
a
niche
of
people
in
webassembly
that
are
talking
about
displacing
Docker
as
the
container
engine,
because
the
web
assemblies
are
potentially
smaller
in
terms
of
resource
demand
and
for
sending
them
on
the
wire
to
get
your
apps
out
to
Edge.
And
if
you
circulate
in
that
community
community,
there
have
been
attempts
to
modify
kubernetes
to
get
it
to
orchestrate
web
assemblies.
But
I.
A
Think
that
there's
at
least
a
50
percent
consensus
that
they
think
that
ultimately,
the
kubernetes
is
a
stop
Gap.
Just
to
get
you
a
mechanism
to
work
on
things
for
now
that,
but
that,
if
webassembly
catches
on
there
are
enough
different
aspects
about
it,
that
maybe
a
new
orchestrator
is
called
for
to
handle
webassembly
packaging
out
at
Edge.
A
A
D
A
I
I
think
that
what's
going
on
is
they
still
are
using
oci
containers.
There's
a
there's,
an
awful
lot
of
use
of
oci
containers
for
things
other
than
dockerized
applications.
So
it
is
pretty
mainstream
now
I
think
to
package
webassembly
in
an
oci
container,
but
you
wouldn't
necessarily
then
use
the
docker
runtime
at
the
edge
location
to
run
it
you,
you
might
use
something
else,
but
the
advantage
of
putting
it
in
an
obviously
eye
container
is
you
can
use
Docker
container
repositories.
A
You
know
anything
from
the
docker
Hub
code
itself
to
something
like
the
open
source,
Harbor
project
and
there's
a
number
of
object
options
for
running
container
Registries.
It's
just
like
container
Registries
have
morphed
to
Hotel
charts
as
well
in
some
cases
and
they're
just
sort
of
a
generalized
distribution
mechanism
with
some
governance
so
that
you
can
put
things
out
there
that
get
distributed
to
to
nodes
that
are
going
to
run
them.
A
So
I
think
oci
is
very
much
alive
and
well
even
with
webassembly,
but
the
orchestrator
and
the
runtime
is
something
that
maybe
not
so
much
now.
I
think
that
some
of
the
docker
run
times
actually
have
recently
advertised
the
ability
to
actually
run
web
assemblies
in
addition
to
Doc.
So
that
is
out
there
too,
whether
that's
a
wise
decision
or
not
I,
don't
know
it's.
It's
smells
a
little
Swiss
Army
knife-ish,
so
that,
if
you
maybe
are
only
running
webassembly,
maybe
something
purpose-built
for
just
that
would
work
a
little
better.
A
B
A
F
Yeah
I
know
that,
with
regards
to
the
using
the
container
registry
as
a
general
artifact
registry,
they've
added
support
for
Helm
installing
held
charts
inside
for
the
purpose
of
this
group.
It
might
be
interesting
to
look
into
storing
other
build
artifacts
for
Edge
devices
inside
of
the
oci
registry
as
a
service,
because
you
could
start
to
use
it
for
storing
things
like
firmware
or
or
other
small
artifacts
that
you
might
want
to
distribute
to
your
Edge
nodes.
A
Yeah
another
thing
I
found
is
that
that
oci
containers
are
being
used
for
is
potentially
even
the
OS
itself.
You
know
so
originally
docker
uses
reuses
the
kernel
of
the
host,
so
the
expectation
is
that
it
is
additive
to
whatever
Linux
kernel
is
found
on
the
host,
but
as
I
understand
it,
it
turns
out
that
there
was
nothing
in
the
oci
spec
that
precluded
you
from
putting
a
full
Linux
kernel
in
that
container.
A
It
was
simply
that
when
Docker
came
about
you
know,
nobody
did
it
because
you
were
expecting
to
layer
on
top
of
the
host's
existing
OS.
So
I
believe
that
there
are
Edge
Edge,
targeted,
Linux
distributions
that
utilize,
an
oci
runtime
to
actually
package
the
Linux
kernel
itself
and
use
this
to
manage
to
do
updates
at
Edge
nodes,
which
is
otherwise
a
challenge.
You
know,
if
you,
if
you
say,
I'm,
going
to
run
Docker
applications
out
in
an
edge
node
and
use
Docker
to
send
the
application
there.
A
Well,
Linux
itself
is
going
to
have
cves
discovered.
How
do
you
keep
that
patched,
and
it
would
be
nice
if
you
could
use
exactly
that
same
Paradigm,
that
you're
using
to
patch
the
applications
itself
to
catch
the
operating
system,
to
perform,
updates
and
I
believe
that
there
are
some
Linux
distro
Solutions
I
haven't
used
them
myself
that
try
to
use
that
mechanism
to
distribute
the
OS
inside
an
oci
container.
F
Yeah
and
I
think
in
those
cases
they
may
be
deploying
a
container
with
the
tools
necessary
to
rewrite
the
bootloader
and
they
might
be
reaching
around
and
updating
the
kernel.
That
way
and
the
container
becomes
a
delivery
mechanism,
but
it
would
be
interesting
to
document
those
and
see
what
tricks
they
use.
Yeah.
A
But
if
you
could
do
it
well,
the
thing
is
running,
which
is
often
the
case
for
something
like
a
plug-in
IO
board.
You
know
in
a
a
mainstream
full
out,
x86
server,
for
example,
you
might
have
firmware
on
network
cards
and
on
storage
host
bus
adapters
and
I
think
those
generally
can
be
applied
hot,
but
if
it
is
targeting
kind
of
the
motherboard-
and
you
know,
memory
interface,
controller
or
something
like
that-
that
that
that's
going
to
require
sort
of
the
power
cycle
reboot
to
which
might
be
problematic
to
pull
off
in
the
conventional
application.
F
D
F
Things
like,
instead
of
pulling
down
entire
Ubuntu
base
image
with
200
or
400
packages,
all
in
the
same
tarball.
You
might
just
pull
down
a
single
tarball
for
grep
and
that
essentially
becomes
the
layer
that
you
add
to
your
previous
base
image
check
out,
build
packs.
If
you
want
to
see
the
concept,
it's
still
sort
of
being
worked
out.
A
A
They
they
have
an
Amazon,
hosted
repo
that
has
the
source
code
for
it,
and
I
think
that
I
would
contend
that
that
green
grass
is
a
concept
where
there's
a
cloud
hosted
management
layer,
but
out
at
the
edge
nodes
you
would
run
applications
and
that
the
system
could
be
viewed
as
a
form
of
orchestrator
and
it
is
not
Docker
specific
I
believe
with
green
grass.
Once
again,
I
haven't
gone
and
taken
it
to
the
next
level
of
actually
deploying
it
and
playing
with
it.
A
A
A
But
I
think
in
that
commercial
space
that
Amazon
and
Microsoft
I
think
might
have
things
going
on
there
that
are
open
source,
meaning
that,
if
they're
open
source
with
an
appropriate
license
they're
within
scope
of
this
working
group.
So
you
know
if
we're
keeping
a
list,
they
belong
on
it,
but
they
they
might
be
viewed
as
Niche
if
they
are
running
their
own
ecosystem
and
haven't
gotten
enough
adoption
that
people
are
discovering
ways
to
use
them
independently.
C
And
Stephen
this
is
Brandon
I,
don't
know
if
you've
brought
the
list
to
the
attention
explicitly
to
the
LF
Edge
umbrella
project.
I
see
a
couple
LF
Edge
projects
here
at
crano
and
ngex
Foundry,
but
I
know
that
group
is
at
least
a
dozen
or
so
projects
I
would
imagine.
All
of
them
would
qualify
for
inclusion
on
this
list.
Yeah.
A
I
I
have
a
private,
spreadsheet,
I
kind
of
fork.
This
and
I
was
kind
of
aspiring
that
people
associated
with
the
projects
would
directly
compose
their
own
rows
in
this
table,
but
it
hasn't
happened
so
I'm
on
sometime
in
the
next
few
weeks,
I'm
going
to
go.
Add
to
this
list
in
a
big
dump
and
all
of
the
Linux
LF
Edge
projects
are
on
my
list,
as
well
as
all
the
eclipse
I
believe,
or
at
least
Eclipse
has
a
concept
of
kind
of
active
projects
and
ones
that
have
gone
dormant.
A
Speaking
of
eclipse,
let
me
have
one
other
final
thing.
If
somebody
else
wants
to
bring
up
a
subject
that
it'll
take
me
a
minute
to
look
this
up,
but
there
is
a
book
that
was
written
by
the
person
who
runs
open
source
Community
for
the
eclipse,
Foundation
and
I
went
and
ordered
the
book,
and
it's
due
to
arrive
tomorrow.