►
From YouTube: 2022-04-05 CNCF TAG Observability Meeting
Description
* Meshery.io and its founder Lee Calcote. The Meshery project describes itself as "...the open source, cloud native management plane that enables the adoption, operation, and management of Kubernetes, any service mesh, and their workloads."
Services meshes provide a wealth of Observability signals. Meshery aims to facilitate good practices, patterns, and lifecycle management of services meshes.
In addition, we'll be covering updates & news related to:
* New short-format serial video series covering current events in our domain by Hendrik Rexed of the "It's Observable" YouTube channel.
* observe-k8s: a nascent community curated compendium of clonable case-studies
* Amazing new book! "Understanding Software Dynamics" by Richard Sites
the latest on our soon-to-be-unveiled mascot & logo!
How to join: cncf.io/calendar, https://tockify.com/cncf.public.events/detail/388/1649174400000
More info: github.com / tag-observability
#cncf #observability #meshery #kutrace #observe-k8s
A
B
D
D
D
D
D
D
Okay-
well,
it's
five
past,
I
guess
we'll
the
rest
will
will
get
here
when
they
get
here
so
welcome
everyone.
This
is
the
first
meeting
for
tag
observability
in
april.
This
is
a
cncf
sponsored
event,
so
the
code
of
conduct
does
apply.
Please
don't
put
anything
in
chat
or
do
or
say
anything
that
would
be
in
violation
of
that
code
right
so
sign
in.
If
you
haven't
already
we'll
cover
some
admin
bits
and
some
other
things
prior
and
we'll
come
back
to
mesheri.
D
So
first
for
evan
bits,
there's
been
alolita
and-
and
I
and
some
others
were
talking
about
when
we
meet-
and
there
was
a
suggestion
that
we
move
to
sort
of
every
other
meeting
being
some
sometime
around
now,
which
is
really
good
for
the
west
coast
of
the
us,
the
east
coast
of
the
us
or
north
america,
rather
as
well
as
central
europe.
However,
it's
really
not
good
for
the
the
other
time
zones.
D
B
D
Cool
okay,
so
we'll
do
that
on
the
list.
Secondly,
there's
a
really
old
issue:
issue
number
37
for
our
logo-
that's
just
been
sort
of
lingering,
but
in
in
talking
with
the
the
cncf
folks
that
have
been
working
on
logos
for
all
of
the
other
tags.
They
all
seem
to
be
animal
based,
and
so
you
know
there
were.
There
were
successive
rounds
of
of
ideas
around
around
that
and
folks
seem
to
like
the
owl.
D
I
I
kind
of
like
an
owl
as
a
mascot
as
well
versus
some
of
the
other
choices
you
know.
Does
anyone
want
to
write
up
something
about
owls
and
and
and
and
how
they
might
be
applicable,
I'm
kind
of
a
nerd
for
animals
and
birds
and
things
so
it
could
all
be
in
my
head.
But
you
know
owls
are
wise
they're,
you
know
there's
allusions
to
heisenberg
and
things
like
that.
You
know
they
fly
softly.
D
D
D
B
B
It
might
be
just
to
get
those
telemetry
signals
from
a
bear
system
on
cube
running
and
get
that
piece
of
infrastructure
laid
down
and
then
focus
on
instrumenting
demo
applications
to
provide
more
application
level
information
into
that
observability,
rather
than
just
the
stack
itself,
and
I
think
that's
kind
of
where
we're
at
right.
Now
any
questions
or
comments
from
anyone.
B
B
D
D
C
Hi
there
I'm
here
yeah
yeah,
so
we
started
an
initiative
for
the
moment.
It's
like
a
pure
pilot.
There's
nothing
really
going
ongoing
on
the
cncf
youtube
channel
that
we
want
to
create
open
source
or
open
observability,
I
would
say,
but
we
extend
it
to
open
source
in
general,
because
there
are
other
project
concerns
here,
not
really
dedicated
to
observability.
But
the
idea
is
to
give
a
continuous
update
to
the
market
and
to
increase
adoptions
on
various
projects.
C
So
we
started
the
first
pilots
few
weeks
back
and
we
are
going
to
record
another
one
in
two
days.
The
idea
we're
improving
are
our
the
process
and
again
the
idea
is
to
the
to
deliver
video
about
10,
to
15
minutes
max
with
five
minutes
of
news
sharing
presenting
few
new
projects.
What's
the
value
of
that
sharing
some
things,
and
then
we
pick
one
of
the
topics
from
one
weekly
news
and
we
discuss
with
michael
from
amazon.
C
It
was
a
blast
and,
and
then
we
finished
with
I'm
starting
to
design
like
a
weather,
forecast
dedicated
to
open
telemetry,
and
the
idea
is
to
keep
track
on
the
various
various
project.
Libraries,
which
version
how
stable
they
are
any
news
related
to
the
project,
so
people
can
basically
follow
what's
going
on
and
which
version
they
should
pick
if
they
want
to
get
started
in
terms
of
compatibility
and
so
on
and
again
the
idea
is
at
the
moment
we
are
piloting.
C
So
if
you
want
to
contribute
again
the
we
don't
want
to
be
10
10,
because
otherwise
we
were
difficult
to
manage,
recording
and
bus
production
it
is
to
to.
But
if
you
want
to
contribute
on
the
content
and
if
you
want
to
share
news
because
at
the
moment
we
are
mainly
going
after
news
and
searching
for
each
project.
C
D
C
So
don't
the
pilot
will
change
to
be
honest
because
we
did
it
live,
but
I
think
this
is
another
content
for
live
streaming,
so
we
could,
we
will
probably
do
post-production.
So
at
the
end,
this
will
change
soon
with
another.
The
ideas
to
replicate
the
experience
of
a
tv
news,
old
style,
tv
news.
So
that's
the
so.
With
the
with
the
live
nature,
it
was
difficult
to
make
it
to
have
that
that
experience
so
we're
gonna
change
and
do
post-production
instead
so
record
and
then
put
the
background
and
so
on.
C
So
at
the
moment
I'm
gonna
do
it.
But
again,
if
there
are
people
who
wants
to
just
yeah
be
part
of
the
project
they
want
to,
they
like
to
do
post
production
and
the
idea
is
we
can
collaborate
on
this
one
I
mean
I
already
have
some
some
some
stuff
on
my
plate,
so
it
could
be
great
to
share
that,
of
course,
the
work,
but
again
I
I'm
I'm
used
to
do
a
lot
of
post-production.
So
if
you
are
not
used
to
that,
then
I
can
do
it.
C
One
thing
is
at
the
moment
the
weather
forecasts
that
I'm
going
to
discuss
at
the
moment.
I
will
do
the
weather
forecast
for
sure,
but
if
someone
feels
to
to
put
yourself
in
front
of
a
camera
with
a
green
screen
and
make
it
yeah
make
it
fun
feel
free.
Otherwise
I
will
do
it,
but
actually
I
I
have
michael
who's
the
special
reporter,
so
we
make
a
discussion
and
then
there's
the
four
weather
forecasts
section
that
could
be
presented
by
someone
else.
D
Absolutely
I'm
happy
to
bang
a
drum
here
too
as
well.
I
think
it's
been
something
I've
kind
of
repeated
a
bunch
over
the
last
year,
but
ever
since
we
got
the
youtube
channel
and-
and
you
know
just
the
go
ahead
from
the
cncf
and
support
you
know
to
to
produce
content
such
as
what
hendrick's
doing
the
one
sort
of
prime
directive
to
borrow
star
trekism
is
really
to
be
creative
and
be
fun.
You
know,
there's
you
know
you
can
use
it.
D
So
what
henrik's
doing
here,
I
think,
is
awesome
because
it's
catchy
and
it's
novel
and
it's
not
just
another
blog
post
or
not
just
another.
You
know
info
dump
of
things
that
are
happening
but
he's
actually
making
it
engaging
and
making
it
stick
out
from
the
rest.
If
you
will
so
you
know,
we
were
kind
of
asked
specifically
to
unleash
the
creative
forces
that
we
have
in
the
tag
around
this.
D
So
so,
if
anybody
does
have
a
passion
for
this
sort
of
video
format
or
just
the
overall
concept
of
informing
the
community
or
finding
interesting
things,
please
reach
out
to
her.
I'm
gonna
make
a
very
brief
plug.
I'm
almost
done
with
this
book,
which
is
understanding
software
dynamics.
It's
wonderful!
It's
it's
sort
of
bringing
back
old,
I
don't
say
old
ideas,
but
ideas
that
were
used
up
to
20
years
ago
in
in
in
my
case,
in
the
embedded
system
space.
D
D
It
records
at
that
point
why
it
woke
up
so
threads
typically
run
until
they
are
waiting
on
something
you
know
whether
it's
an
io
or
a
lock
or
some
other
weightable
resource.
So
what
this
profiling
approach
does
is
sort
of
build
something
analogous
to
a
distributed
trace,
but
this
is
all
you
know
way
down
inside
inside
the
kernel,
and
so
you
get
a
look
at
every
time
a
thread
wakes
up.
Why
did
it
wake
up
with
a
relationship?
D
You
know
an
arc
back
to
the
thing
it
was
waiting
on
and
unlike
profiling,
they
can
just
look
at
one
process,
oftentimes
or
or
something
like
that.
You
you
get
a
holistic
view
of
all
of
the
interrupts,
all
of
the
threats,
not
just
one
process
but
on
on
the
whole
system
and
their
interactions,
and
so
this
is
a
really
rich
source
of
of
data
that
can
help
us
really
understand.
What's
going
on
like,
why
did
this
code
run
a
little
bit
slower?
D
You
know
it
could
have
nothing
to
do
with
that
code.
Something
else
could
be
slamming.
You
know
I
o
on
the
underlying
node
a
completely
unrelated
work
workload,
but
that's
having
an
impact.
In
addition,
it
does
a
very
good
job.
This
approach
of
recording
explicitly
when
nothing's
happening
right,
like
everything,
is
running
slow,
but
when
you
go
in
there
and
then
look
at
what's
going
on,
perhaps
everything
is
basically
waiting
and
sleeping
all
the
time.
D
D
D
D
Her
talk
was
called,
I
believe,
beyond
github
stars
and
pull
requests
or
something
to
that
effect
and-
and
she
went
into
you
know
how
do
we
assess
the
health
of
a
project?
How
do
we
assess
the
health
of
a
community?
You
know
it's,
it's
not
just
a
popularity
contest
with
stars
and
and
much
like
quantifying
the
output
of
an
engineering
team.
D
So
we
started
looking,
I
we
started
looking
at
other
groups
within
the
cncf
that
are
doing
something
similar
and
if
and
about
a
month
and
a
half
ago
or
two
months
ago,
catherine
paganini
from
the
business
value
sub
committee,
who
is,
is
the
group
that
has
created
the
glossary,
glossary.cncf.edu
and
and
other
other
things
like
it.
They
have
a
goal
of
providing
inhuman,
speak
to
business,
decision
makers
and
other
technology
leaders.
D
You
know
what
do
all
these
jargon
terms
mean
and
how
can
we
help
normal
humans
understand
some
of
our
vocabulary
and
understand
the
ecosystem,
and
so
so,
under
that
banner,
they
had
an
ongoing
effort,
what
they
had
and
still
have
an
ongoing
effort
to
kind
of
walk
through
the
sector
and
help
provide
some
clarity
like
what's
a
vendor?
What's
a
company,
you
know
what
does
this
mean?
D
If
you
can't
see
it
you'll
be
able
to
soon
it's
an
internal
repo
to
the
cncf,
but
I
won't
go
into
great
detail
here,
but
the
long
and
short
of
it
is
called
landscape
graph
and,
in
short,
it's
a
graph
data
model
of
the
landscape.
The
landscape
we
have
now
is
is
very
rectangular.
If
you
will
it's
a
giant
spreadsheet,
with
a
really
cool
app
on
top
of
it
that
if
you
know
what
you're
looking
for
you
can
go,
find
information
about
any
given
product
or
cncf
member
organization,
companies,
vendors,
etc.
D
However,
if
you
don't
know
what
you're
looking
for,
if
you
want
to
understand
the
correlation
between
things
like
who
contributed
to
these
three
projects
and
what
else
did
they
contribute
to
or
another
example
might
be
for
the
cncf
sandbox
for
all
projects
in
the
observability
domain?
Who
are
all
the
contributor
contributors
to
them
in
aggregate
who
employs
all
of
them
right
and
who
funded
those
employers
right?
So
you
can
look
at
like
trends
like
that
that
sound
a
lot
like
social
networking
ad
tech
phenoms.
D
You
know
you
know,
you
know
these
kinds
of
things
where
graph
databases
for
the
last
decade
and
a
half
have
really
revolutionized
how
we
assess
a
community.
You
know
social
networking
and
the
relationships
between
things.
So
what
I've
done
is
I've
taken
an
initial
whack
at
pulling
the
data
model,
which
is
a
massive
json
file
into
a
neo4j
database
so
that
that
can
form
a
data
model
that
can
that
can
be
then
used
by
all
manner
of
things.
D
What
packages
do
they
all
bring
in
and
what
versions
are
they
and
cross
reference
that
with
all
of
the
known
cves?
So
we
can
say,
like
you
know,
we
can
provide
to
both
the
toc
and
to
the
to
the
community.
An
assessment
of
the
cncf's
project
and
and
build
up
an
understanding
of
what
dependencies
they
have
that
may
or
may
not
be
safe
or
unsafe.
So,
to
answer
these
kinds
of
questions,
a
relational
data
model
falls
over
pretty
quickly
generally
around
two
or
three
friends.
D
If
you
will
a
relationship
hops,
so
so
that's
in
a
nutshell,
some
ongoing,
very
early
work
and
in
the
next
few
days,
certainly
by
the
end
of
the
week,
I
would
expect
that
repo,
hopefully
to
be
public,
if
you're,
not
a
ts,
a
cncf,
tlc,
contributor
or
part
of
the
cncf
organization
on
github.
If
you
want
to
see
a
brief
preview,
you
can
go
to
my
github
and
there's
a
landscape.
D
D
Okay
with
that,
we
are
at
the
halfway
point
here,
which
is
perfect
lee
yeah,
you
know.
Well,
I
haven't.
I
haven't
introduced
you
yet
so
yeah
briefly,
so
so
lee
is
the
founder
of
a
project
called
measuri
which
describes
itself
as
the
open
source
cloud
native
management
plan
plane
that
enables
the
adoption,
operation
and
management
of
kubernetes
any
service
mesh
and
the
workloads
and,
as
we
know,
service
pressures
are
a
huge
source
of
signal
signal
data.
A
Yeah,
well
I
I
hope
I
hope
that
you're
excited
we'll
see.
There's
we've
got,
I
don't
know,
there's
a
couple
of
there's,
maybe
a
couple
of
novel
ideas
in
here
about
how
to
visually
comprehend
your
infrastructure,
your
workloads,
so
my
just
to
add
to
matt's
background
a
little
bit
to
the
extent
that
it's
relevant
here
so
the
while
my
prior
well,
my
current
full-time
role
is
kind
of
focused
toward
measuring
toward
a
few
other
open
source
projects.
A
My
last
full-time
role
was
leading
technology
strategy
for
solarwinds,
and
so
observability
is
a
real
focus
of
mine.
That
was
let's,
let's
not
talk
about
the
other
solarwinds
incident.
Let's
not
talk
about
solar
winds,
making,
I
don't
know
just
uplifting
the
the
value
of
any
security
oriented
startup,
but
the
the
that
was
that
was,
after
my
time.
A
A
A
You
know
the
more
that
you
look
at
a
thing,
the
more
frequently
the
more
overhead
that
is,
and
so
there's
a
lot
of
questions
around
overhead
and
understanding
overhead
and
in
context
of
the
value
being
provided.
That's
a
large
part
of
what
service
mesh
performance
as
a
project
is
so
these
two
projects
measure
in
service
mesh
performance,
one
is
a
spec
project.
The
other
one
is
the
capability
they
entered
play
with
service
mesh
interface,
smi
smi
is
focused
on
interoperability
across
service
meshes,
and
so
is.
A
A
Some
of
the
things
that
measuring
looks
to
do
is
to
empower
people
with
patterns
and
best
practices.
There
is
well
an
up
and
coming
book
on
service
mesh
patterns.
The
patterns
that
are
described
inside
that
book
are
available
to
measuring
and
importable
by
measuring,
so
there's
something
of
a
surface
mesh
catalog
that's
coming
forth,
where
oh
mesh
users
mesh
would
be
able
to
take
and
apply
these
patterns.
A
There's
this
catalog
is
in
just
in
draft,
and
there
are
individual
patterns
that
are
web
assembly
filters
at
some
point,
probably
evpf
programs
again
with
both
observable
with
monitoring
use
cases
and
then
other
use
cases
for
those
those
programs.
Those
filters
to
apply
in
your
infrastructure,
meshi
helps
people
get
those
configure
them
and
deploy
them.
A
Let
me
let
me
go
into
mesh
where
you
poke
around
some
of
the
interfaces
and
do
probably
what's
the
worst
job
of
introducing
the
project
that
I
have
to
date,
and
so
inside
of
so
so,
one
of
the
reasons
I'm
going
to
do
a
bad
job
is
because
it's
because
part
half
of
my
cluster
is
down
I'm
having
problems
with
the
networking
in
my
cluster.
So
one
of
the
contributors
to
measures
on
the
call
he's
going
to
help
me
show
the
other
part
of
of
what
I'm
missing.
A
So
we're
missing
some
data
in
the
dashboard
that
you
would
otherwise
normally
see
here.
One
of
the
things
that
meshri
does
by
way
of
trying
to
help
people
get
at
maybe
pre-existing
dashboards
or
creating
new
dashboards,
primarily
from
grafana
and
the
metrics
that
they're
they're,
pulling
in
from
prometheus
meshri
will
discover
the
kubernetes
environment
it'll,
discover
the
service
mesh
environment,
it'll
provision
service
meshes
or
discover
existing
ones.
A
A
A
So
in
in
this
sense,
there's
not
necessarily
a
lot.
That's
novel
to
this
I
would
say
a
novel
item
is
that
you're
able
your
users
are
able
to
add
their
own.
You
know
their
own
metrics
that
they'd
like
to
monitor
and
they
do
that
by
way
of
defining
using
grafana's
board
syntax
they're
able
to
import
grafana
boards.
That's
great.
A
The
novelty
that
comes
in
probably
by
way
of
observability
for
mescheri
has
to
do
a
bit
with
a
visual
topology
called
mesh
map.
There's
two
modes
of
mesh
map.
One
mode
is
a
designer
mode
in
which
you
can
visually.
You
know,
design
and
configure
your
like
kubernetes,
environment
or
even
more
specifically,
like
your
service
mesh
deployment.
A
What
meshrey
will
do
is
allow
you
to
meshwi
has
knowledge
of
each
kind
of
any
number
of
versions
of
a
given
service
mesh
that's
been
released,
it
lets
you
drag
and
drag
and
drop
your
entities
over.
This
is
not
going
to
make
sense
what
I'm
kind
of
drawing
on
the
fly
here,
but,
but
basically
do
do
this
type
of
design
come
in
fill
in
the
specifics
about
a
given
component
and
kind
of
visually
configure.
It.
A
In
this
case,
if
you
were
wanting
to
maybe
apply
a
filter,
you
might
want
to
drop
that
filter
onto
maybe
your
ingress
gateway
in
this
case
and
have
a
particular,
maybe
some
additional
signals
generated.
So
you
could
then
see
them
in
your
in
the
charts
that
we
were
just
looking
at
to
the
extent
that
you
wanted
to
maybe
chain
a
couple
of
filters
together
that
you'd
go
through
and
configure.
You
know,
you'd
select
what
filter
it
is
that
you
wanted
to
run.
A
A
In
this
particular
case,
this
is
a
kubernetes
deployment,
so
that
represents
a
little
bit
differently
to
the
extent
that
you
might
drag
and
drop
particular
components
inside
of
the
inside
of
a
deployment
anyway.
So
so
this
is
the
part.
That's
really
frustrating
because
this
other
mode
of
measure
of
mesh
map,
this
visual
topology
is,
is
a
visualizer
which
is
much
more,
this
much
more
of
a
read-only
style
like
visualization
of
your
actively
running
infrastructure
and
looking
at
the
telemetry,
and
that
kind
of
thing
this
is
what's
frustrating
is
this
is
broken.
A
This
is
like
my
environment's,
not
messed
up,
but
pranav
if
you
might
be
able
to
share
so
pranav,
is
a
mystery
contributor
and
works
on
a
number
of
different
components
of
measuring,
so
the
visualizers
to
look
like
this.
It's
a
right
now.
You
know
kind
of
a
color
coding
of
the
environment,
you're
running
so
there's
some
objects
in
kubernetes,
some
in
in
the
measuring
name
space,
since
he
has
mesh
redeployed.
A
A
Pranav,
if
you
right
click
on
a
pod,
we
wanted
to
show
folks
how
how
logging
is
being
done
so
yeah.
If
you
people
are
able
to
open
up
an
interactive
terminal
against
a
a
pod.
It's
actually,
if
you
pranab,
if
you
minimize
this,
but
you
right,
click
on
the
rec,
the
rounded
rectangle
around
yeah
there
you
go
yep.
So
it's
a
little
this.
This
needs
to
be
updated,
not
on
the
hex,
the
octagon,
but
the
rounded
rectangle,
that's
around
yes,
that
yeah
there
you
go!
So
that's
the
pod.
A
What
yep,
if
you
initiate
connection
there
what'll
end
up
happening,
is
that
for
any
containers
that
are
inside
that
pod,
an
ssh
set
or
a
terminal
session
will
be
established,
but
the
same
thing
goes
for
logs.
So
if
people
want
to
quickly
initiate
a
logging
or
a
log
stream,
they
can
right
click
on
that.
If
you
choose,
maybe
a
different
pod,
pranav
and
right
click
and
do
log
streaming
instead.
The
concept
here
is
that
people
will
be
able
to
readily
apply
filters.
A
They
can
do
there's
a
magnifying
glass
in
here
to
kind
of
quickly
search
through
the
logs
to
find
what
they
need
to
look
for
or
with
the
mesh
grid,
doesn't
have
it
yet
but
sort
of
predefined
filters
that
people
you
know
things
that
you're
looking
to
search
for
in
advance,
and
so
I
wanna
pranav.
Actually,
if
I'm
gonna
re-share
a
couple
of
other
aspects
of
mastery,
but
I'm
gonna
stop
talking
so
people
are
gonna.
Ask
questions,
there's
kind
of
a
lot
to
measure.
A
D
A
A
number
of
integrations
that
that
kind
of
make
this
immediately
useful
out
of
the
box
yeah
good,
great
question
so
so
mesh
rehab
measure
is
an
extensible
platform,
so
some
of
the
things
that
we
were
just
showing
are
plugins
or
our
areas
of
configurability
for
for
measuring
and
one
of
those
areas.
One
of
the
prominent
things
is
this
concept
of
adapters,
and
so
meshri
has
an
adapter
for
each
of
the
following
service
meshes.
A
Actually,
this
one
needs
to
be
promoted
and
this
one
needs
to
be
demoted,
but
so
there's
about
10
in
total.
I
think
this
lists
11
this
one
that
needs
to
be
demoted
needs
to
go
away,
so
we'd
have
about
10,
so
cilium
surface
mesh,
I
think,
is
it-
is
in
beta
there's
a
couple
of
folks
from
isoven
that
we've
been
working
with
to
make
sure
that
the
adapter
best
represents
the
capabilities
of
each
mesh.
A
Some
of
you
might
be
asking
wait
a
second
like
wait.
So
if
you're
familiar
with
smi
service
mesh
interface,
it
has
a
set
of
currently
four
but
maybe
a
fifth
api
specification
that
defines
kind
of
a
lowest
common
denominator
set
of
api
interfaces,
which
is
to
help
people
or
tools
interact
with
service
meshes
in
a
uniform
way.
That's
kind
of
at
the
core
of
what
smi's
value
proposition
is,
and
it's
quite
helpful
measurey
is
quite
close
to
the
smi
project.
A
So
this
is
kind
of
a
global
view
of
the
tests
that
have
been
run,
performance
tests,
so
load,
that's
been
generated
against
an
active
environment
and
then
statistical
analysis.
That's
been
run
or
done
in
evaluation
of
that
load,
and
so
the
kind
of
the
you
know:
p50
p75,
p90
p99.
These
things
are
calculated
for
a
duration.
So
when
someone
runs
a
performance
test,
they
choose
an
endpoint,
they
choose
a
particular
configuration
of
a
service
mesh
and
measure
will
run
that
this
helps
people.
Both
you
know,
compare
against
one
another.
A
A
So
let
me
see
if
I
can
show
kind
of
a
quick
comparison
just
randomly.
So
if
two
different
tests
have
been
run,
you
can
come
in
and
choose
between
those
two
and
choose
to
compare.
So
I
don't
I
don't
know
the
particulars
of
them
and
maybe
they're
of
the
environment
change.
Maybe
there
weren't
any
environmental
changes,
but
but
people
you
can
come
through
and
kind
of
try
out
one
configuration
see
which
one
it
performs
better
for
you,
and
so
you
can
see
this
ties
into
that
service.
A
C
A
A
Yes,
is
the
answer
to
your
question
by
the
way
like
for
the
run
test.
It's
like
mesherie
itself
has
three
different
types
of
load:
generators
built
in
so
fortio,
wrk2
and
nighthawk.
They
each
each
behave
like
to
most.
Maybe
they
don't
care,
they
behave
a
little
bit
differently
or
they
do
their
analysis
a
little
bit
differently.
Some
of
them
start
counting
at
different
points.
A
C
A
So
you
can
do
distributed
performance,
testing
and
kind
of
distributed
analysis
and
have
that
have
that
the
statistical
analysis
come.
You
know,
data
sync
back
centrally
and
kind
of
coalesce,
so
you
can
look
at
it.
That's
one
thing,
the
the
second
thing
that
that
the
community
is
doing
around
nighthawk.
Is
these
adaptive
load
tests,
which
is
to
say
if
which
is
to
like
let
people
run
an
optimization
algorithm,
more
or
less,
which
is
like.
A
If
you
have
the
question
of
something
like
hey
I'd
like
to
make
my
service
mesh
deployment
or
my
my
application
infrastructure
as
as
resilient
as
possible,
you
know
knowing
that
the
network
fails
occasionally
I'll,
go
ahead
and
configure
some
retries.
So
I've
got
10
services
I'll
put
on
one
retry
per
service.
C
It
it
would
be
cool
to
have
a
also
the
ability
to
do
like
argo,
like
a
workflow,
where
you
run
a
test
and
in
the
middle
of
the
test,
you
start
changing
your
retry
logic
or
anything,
and
then
you
measure
during
so
then
you
have
the
during
the
test.
The
overall
test
results.
You
can
see
those
specific
events
and
what
was
the
impact
in
terms
of
for
the
user
would
be
great.
A
Totally
heinrich
it's
like
you're
speaking
to
the
road
map,
because
mastery
is
that
almost
that
it's
v0.6
release
and
before
it
gets
to
v10.
There's
like
there's
like
two
major
architectural
components
to
come
forth.
One
of
them
is
a
workflow
engine
and
that's
a
great
example
of
how
someone
would
use
the
the
workflow
engine
that
we're
looking
at
is
or
have
been
for
a
long
time
is
temporal.
A
The
second
architectural
component
is
a
policy
engine,
so
something
like
oppa
or
caverno,
or
something
because
to
your
point
like
not
only
will
they
maybe
want
to
change
something
based
on
have
a
workflow
in
there
to
change
it,
but
they
also
might
want
to
have
a
policy
that
does
an
evaluation
to
say.
You
know,
then
so.
A
A
The
other
thought
here
is
that
those
that
are
you
know,
working
on
microservices
or
developing
against.
You
know
developing
kubernetes
based
applications
over
time
as
service
meshes
become
ubiquitous,
that
they
would
be
able
to.
You
know,
use
the
combination
of
mastery
and
docker
desktop
to
have
kind
of
a
checkbox
for
a
local
kubernetes
deployment
or
a
local
service
mesh
deployment,
and
then
you
know,
go
go
play
and
figure
out
what
the
right
configuration
is
based
on.
A
A
It
was
we
wanted
to.
I
I
don't
know
you
know,
I'm
like
well
heinrich's
thoughts,
that's
those
are
most
welcome.
Those
suggestions
about
and
like
the
specific
mention
of
k6
and
the
other
one
and
locust
part
of
our
challenge
by
the
way
for
the
three
for
fortio
wrk
ii
and
nighthawk
from
mesheri
meshes
written
in
golang
fortio
is
written
in
golang
as
well,
and
so
the
integration
there
is
quite
there's,
there's
no
bugs
like
it's.
A
D
D
A
Yeah
totally
yeah
the
yeah.
The
I
mean
pranav
is
a
great
example
pranav,
and
I
did
not
plan
this.
He
just
showed
up
to
the
meeting
he's
contributing
to
mystery.
He
had
it
running.
He
could
show
it
like
he's
a
great
example
of
there's
a
there's,
quite
a
there's,
a
very
healthy
community
around
mastery.
Maybe
it's
too
open
to
change,
because
it
there
is
a
lot
that
is
desired
to
be
done
in
the
project.
A
D
Yeah,
I
I
can
also
say
I
put
a
link
to
the
messaging
overview
deck
into
the
meeting
notes
as
well.
You
know
lee's
got
a
tall
order
here
in
terms
of
summarizing
what
measure
is
in
20
minutes
or
less,
but
but
the
project
is
quite
broad
and
deep
and
there's
a
lot
of
content
there,
both
on
measury
and
you
know,
one
of
the
contributing
entities
layer.