►
From YouTube: App Runtime Platform Working Group [Feb 1, 2023]
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Okay,
let's
get
started
welcome
to
the
app
runtime
platform
working
group
meeting
for
February
February,
1st
2023.
We
have
one
thing
on
the
agenda
feel
free
to
add
other
things
on
the
agenda.
If
you'd
like
to
talk
about
them,
let's
start
with
this
BBS
high
bandwidth,
well,
I'm
gonna!
Let
you
take.
B
Okay,
it's
a
topic
coming
from
us.
I
I
can
give
you
few
words
as
a
motivation
and
to
put
the
story
behind
these
item
is
basically,
we
have
a
stakeholder
on
running
on
our
platform,
which
has
really
strange
or
even
weird
setup,
with
a
thousands
of
long
running
processes
in
a
single
space
and
during
the
platform
update,
we
noticed
a
high
CPU
and
high
bandwidth,
outgoing
high
bandwidth,
outgoing
from
the
Diego
API
nodes,
and,
of
course
it's
due
to
the
BBS
process.
B
Running
on
them
unfortunately
seems
like
the
with
deeper
investigated
the
issue,
maybe
a
year
ago,
and
actually
noticed
that.
The
reason
for
this
is
actually
the
watch.
The
growing
amount
of
egress
rules
stored
into
the
Run
info
object,
which,
apart
from
that
is
every
time
encrypted
and
decrypted.
Once
we
somehow
touch
or
add
something
into
this
and
of
course,
the
increased
amount
of
the
egress
rules
affect
the
outgoing
bandwidth
of
the
Diego
API
nodes,
while
the
encryption
and
decryption
is
actually
a
factor
for
the
high
CPU
usage.
B
Just
to
give
you
an
over
a
glimpse.
What
we
currently
use
in
terms
of
virtual
machines,
VM
types,
so
we
have
in
order
to
support
this
use
case.
We
need
to
use
Amazon,
C5
and
9
extra
large
instance
type.
It's
c5n
VMS
are
network
enhanced.
They
have
enhanced
networking
which
actually
give
us
the
upper
the
opportunity,
the
opportunity
or
even
the
or
the
ability
to
to
handle
this
use
case.
B
B
Unfortunately,
c5n
9
extra
Arch
are
very
expensive
notes
and,
in
the
course
course
of
cost
saving
optimizations
whatever
we
are
willing
to
solve
this
problem
because,
most
of
the
time
99
percent,
these
resources
are
idling
and
we
are
actually
not
able
to
utilize.
C
B
And
it's
a
kind
of
a
waste
yeah,
basically
Kaylee
and
to
me
did
this
proof
of
concept.
We
try
to
reproduce
to
locally
and
came
up
with
these
metrics
coming
from
a
Amazon
which
shows
how
the
network
bandwidth
linearly
growed
during
the
platform
update
it's
something
like
eight
eight
hours,
more
or
less
reproducing
the
exactly
the
same
scenario.
We
have
on
productive
systems,
but
yeah
right
now,
I
more
or
less.
Imagine
that
fixing
both
problems
first,
Network
bandwidth
and
the
second
CPU
High
CPU
usage
should
be
a
complicated
topic.
B
A
Yeah,
these
are
some
good
good
research.
A
It
sounds
great,
so
sounds
like
what
you're
asking
for
is
feedback
from
the
broader
working
group,
or
how
can
we
help
you.
B
B
C
A
In
the
piano,
oh
here
here
here,
okay
got
it
sorry,
I
missed
that
okay
right
have
you
can
or,
and
if
you
haven't,
can
you
advertise
this
in
the
our
working
group
stock,
Channel
or
in
the
the
I
guess
the
Diego
side
Channel.
C
No
I
think
we
advertised
it
a
while
back.
We
haven't
got
back
in
the
comments
below
and
we
mentioned
why
exactly
I
think
it
was
Maria.
That
gave
a
comment
and
she
mentioned
that
we
should
not
query
the
database
for
the
egress
rules
in
the
first
place.
So
if
we
can
make
it
so
that
we
don't
query
the
egress
rules
in
the
first
place,
however,
the
egress
rules
and
everything
else
is
stuck
in
the
Run
info
and
the
whole
relief
with
safety
as
a
whole.
C
So
that's
kind
of
the
cause
of
the
problem
and
after
our
last
update,
there
was
no
nothing
from
your
site
and
that's
why
we.
A
A
Got
it
so
you're
re-poking
us
for
some
more
feedback.
C
Yeah,
so
so
we
know
how
to
because
she
mentioned
that
we
we
added
the
for
the
POC.
We
added
the
the
parameter
to
the.
A
B
A
A
B
By
the
way,
Maria,
not
Maria
but
Army-
sorry,
you
may,
if
you
remember,
maybe
four
or
five
working
group
sessions
before
we
discussed
something
about
the
runtime
evacuation
whatever.
So
you
mentioned
that
it's
good
to
have
something
quite
an
issue
as
a
starting
point.
Do
you
mean
to
create
an
issue
for
the
Diego
release
or
somewhere.
B
Mean
where
to
create
the
issue
itself,
describing
some
kind
of
a
high
level
requirements
whatever.