►
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
Alright
welcome
John
is
not
with
us
this
week,
so
I
am
luke
Tillman
and
I'm,
a
technical
evangelist
for
datastax,
so
I'm
going
to
be
running
this
week
in
Cassandra
for
for
this
week
and
with
us
today
we
have
Patrick
McFadden,
who
is
chief
evangelist
for
Cassandra
datastax
and
also
Joel
nitin,
who
were
super
excited
to
have
he's
going
to
talk
to
us
a
little
bit
about
Jepson
testing
with
cassandra
and
he's
also
a
data
stack
simply
so
you
know
we
you
guys,
probably
if
you've
listened
before.
You
know
how
this
works.
A
We
kind
of
go
through
some
of
the
blog
posts
that
that
are
linked
to
from
from
Planet
kisses
the
planet,
Cassandra
blog
post,
and
there
were
two
that
we
kind
of
wanted
to
take
a
little
bit
of
a
look
at
so
the
first
one.
Was
this
blog
post
talking
about
getting
started
with
spring
boot
and
Cassandra
so
spring
super
popular
in
the
java
community,
I'm
sure
if
you're
a
Java
developer.
Listening
to
this,
you
know
exactly
what
I'm
talking
about
when
I
say
spring
boot
and
Patrick.
B
I
mean
yeah,
I
mean
it's
not
just
people
excited
about
their
like
you're,
either
spring
boot
or
you're,
not
now
I'm.
It's
that
it's
that
in
version
control
framework
stuff,
that's
pretty
cool,
but
I
think
you
know,
just
in
general
spring
has
got
such
a
good
following
in
the
java
community,
but
it
I
think
it
is
not
as
well
known
that
spring
boot
includes
a
native
support
for
Cassandra
and
that's
that's
pretty
cool,
especially
when
you
use
bring
data
and
that's
been
around
for
a
while
yeah.
A
I
didn't
realize
that
the
spring
data
support
was
there.
You
know
I
know,
but
I've
heard
you
know.
Dree
hi
has
a
competing
sort
of
map,
ER
mapper
implementation
Achilles,
but
if
you're
used
to,
if
you're,
if
you're
used
to
using
spring
data-
and
this
looks
pretty
cool
like
this
looks-
like
I
didn't
even
know,
this
actually
worked
so
yeah.
B
The
blog
post
is
really
neat
I.
First
of
all,
it
starts
out
on
the
right
foot
because
he's
using
CCM
to
build
a
small
cluster
which
I
thought
was
awesome
and
then
from
there
it's
just
it's
real
simple,
it's
a
very
short
blog,
but
you
can
get
a
quickly
operating
and
powerful
running
in
spring,
with
full
Cassandra
support.
Yep.
A
Yeah
pretty
good
getting
started.
So
if
your
java
developer,
especially
if
you're
new
to
Cassandra,
definitely
check
this
one
out,
absolutely
yeah,
okay,
so
the
other
blog
post,
then
we
wanted
to
talk
about.
Was
this
post
on
the
smack
stack?
So
it's
talking
about
how
to
get
started
on
D
cos,
which
I
didn't
realize,
but
D
cos
is,
what's
actually
powering
a
sure
container
service
that
just
went.
You
know
that
Microsoft
just
announced
as
a
container
service
coming
out
of
beta,
and
this
is
sort
of
the
thing
under
the
covers.
A
That's
that's
running
it,
but
anyway
I
think
we
wanted
to
talk
more
specifically
about
the
smack
stack,
because
this
is
something
that
we're
seeing
a
lot
of
people
out
in
the
field
showing
interest
in
and
lots
of
deployments
and
Patrick.
Again.
You've
talked
a
ton
about
this.
I
know
you've
gotta
talk,
even
that
you
gave,
I
think,
at
the
at
the
Cassandra
meetup
group
out
there
in
the
bay
area
about
this
seems.
B
C
B
C
B
I
think
a
lot
of
people
get
tripped
up
on
the
aqha
part
and
sometimes
Manzo's
a
little
bit,
because
they
don't
think
of
these
things
all
kind
of
fitting
together,
but
they
do
so
yeah.
Those
are
five
letters
that
it
really
is
is
put
together
in
a
way
that
makes
up
a
data
pipeline.
So
if
you
think
of
all
the
needs,
you
have
its
the
hole
it's
to
collect,
organize
process
and
store,
so
the
collection
is
the
kafka
part
where
the
organization
is
Kafka
as
well.
B
B
A
And
so
you
see
people
doing
this
for
not
just
like
streaming
pipelines
but
also
like
you
know,
since
you've
got
spark
you're,
even
capable
of
doing
like
batch
batch
analysis
on
the
back
ends.
Like
I
mean
this
is
this
is
partially
for
real-time
sort
of
data
ingestion
and
calculations
stuff
like
this,
but
you
know
you're
not
limited.
You
know
it's
kind
of
a
flexible
architecture
where
you
could
do
either
or
but.
B
You
don't
have
to
use
all
of
the
parts
of
the
same
time.
What
it
is
is
it's
a
tool
box
of
parts
and
when
you
need
it,
you
need
it,
but
it
I
think
the
best
part
of
it
is
that
it's
just.
We
already
have
some
predetermined
organization
here.
So
if
you
need
to
learn
more
about,
let's
say
how
to
how
to
process
data
in
Cassandra
using
spark.
There's
a
wealth
of
information
about
that.
We
talk
about
that
all
the
time.
A
Yeah
and
so
one
other
thing
to
mention:
I,
don't
you
know
it's
not
I,
don't
think
it's
linked
on
this
particular
blog
post,
maybe
we'll
put
a
link
in
the
planet
Cassandra
blog
post,
but
we
do
actually
have
a
demo
that
you
Patrick
and
Helena
Edelson
awesome,
Helena
Adelson
put
together
that's
kind
of
like
at
least
three
of
these
parts.
I
think
maybe
even
four
of
these
parts,
or
of
the
smack
stack
put
together
called
killer,
whether
that's
on
get
up.
A
B
A
Yeah,
okay,
so
what
kind
of
segue
now
we've
kind
of
covered
the
blog
post
talk
a
little
bit
to
joel
about
what
it
is
he
he
does
a
datastax
and
Jepsen,
and
that
kind
of
thing
so
Joe,
maybe
start
by
just
telling
us
like
what?
What
is
your
job
at
datastax?
What
do
you
do?
You
know?
What
does
he
work
on
yeah.
A
So
I
know
in
particular
one
of
those
you
know
kind
of
part
of
that
role
is
working
a
lot
with
jeff's,
and
could
you
explain
like
what
is
Jepsen
for
somebody
who's
listening
out
there
and
has
no
idea,
you
know
what
I'm
talking
about
when
I
say
Jepsen,
you
know
what
what
is
jeff's
in
and
how
does
that?
How
does
that
sort
of
apply
to
Cassandra
or
your
job
working
on
Cassandra
sure.
C
So
so,
Jepsen's
a
lot
of
things
and
I
think
that's
kind
of
one
of
the
first
things
to
establish
in
this
discussion.
Is
you
want
to
make
it
as
productive
as
possible?
Is
there's
really
four
things
I
like
to
think
of
it
as
Jepsen
and
I'm
going
to
make
it
clear
which
one
I'm
talking
about
here
so
there's
a
series
of
talks
that
Kyle
Kingsbury
is
given
on
how
he
tested
these
distributed
systems?
C
There's
some
blog
posts,
there's
some
tests
written
for
this
framework
and
then
there's
the
core
framework
itself
and
that's
what
I've
worked
with
and
the
core
framework
of
Jefferson
is
this
closure
library
that
has
these
facilities
to
make
it
easy
to
test
distributed
systems
under
a
variety
of
failure
conditions?
So
when
you're
testing
a
distributed
system
like
Cassandra
part
of
what
makes
cassander
so
attractive,
is
that
we
want
it
to
be
resilient?
These
real-world
failure
conditions
we
wanted
to
scale
well
and
remain
operationally
simple
under
the
sorts
of
things
that
happen
in
real
data.
B
Is
going
to
stay
here
that
I
Joel,
you
and
I
talked
about
this
often,
but
you
know
it's
it's
something
that
whenever
we
we
talked
to
new
users
or
people
who
are
experienced
in
this
field
of
working
with
large-scale
systems
that
original
blog
post,
Jepsen
blog
post
about
Cassandra
comes
up
all
the
time
and
it's
a
few
years
old.
Now,
at
this
point,
I
think
it's
probably
worth
saying
you
know
this
is
this
is
a
really
the
reboot
of
that.
Would
you
say
I
mean
taking
that
and
going
much
further.
C
Yes,
sir
preview
is
an
interesting
word.
I
think
that
blog
post
is
is
a
cool
artifact
of
a
trend
that
was
happening
at
the
end
of
2013
on
the
into
2014.
Was
this
increasing
awareness
of
how
we
could
understand
analyze
and
test
distributed
systems
and
and
bridging
that
gap
from
sort
of
the
academic
applications
to
the
practical
applications
and
and
that
blog
post
is
sort
of
an
artifact
of
where
that
was
in
time
and
where
cassander
wasn't
at
that
time
in
that
specific
set
of
tests?
B
So
I
was
you
know
when,
when
I
talked
to
I
think
I
people
you're
right.
I
think
people
are
more
way
more.
Savvy
developers
are
way
more
savvy
about
nino.
They
know
what
serializability
is
when
four
years
ago
they
didn't
and
they
weren't
thinking
about
that,
because
when
we
use
relational
systems,
acid
gave
us
this
sort
of
comfort
zone.
You
know
when
you
say
commit
rollback.
You
know
what
you're
getting
right,
although.
C
I
wouldn't
know
that
even
then
I
don't
think
everybody
understood
or
or
fully
fully
grasped
those
bumper
rails
as
well
as
as
well
as
they
might
have
thought,
I.
Think
you're,
probably
right,
I
think
you
really
know
it's
a
confusion
about
no
sequel.
Databases
is
not
peculiar
to
no
sequel.
It's
an
artifact
of
the
fact
that
they're
complicated
systems
with
complicated
jobs-
that's
true
yeah.
A
Yeah
to
your
point,
I
think
like
a
lot
of
people,
don't
like
there's
there's
this
whole
thing
about
transaction
isolation,
levels
and
yeah.
There's
this
whole
other
thing
that
I
think
most
most
developers,
probably
don't
even
don't
even
think
about
when
it
comes
to
relational
databases,
so
I
think
I
probably
agree
with
you.
We
can
stay.
It's
not
unique
to
Cassandra
this
sort
of
the
sort
of
problem.
Yeah.
B
C
Sure
so
so
these
Jepsen
tests
that
I
have
better
that
are
updated
for
modern
Cassandra
and
the
modern
Java
driver
and
all
that
are
up
on
get
up
and
they
and
they
are
open
source
and
there's
something
we're
honest.
We
try
to
understand,
do
features
we're
developing
or
increase
your
confidence
and
features
we
already
have
so
that's
sort
of
where
I'm
at
is
that
really
these
benefits
a
benefit
core
Cassandra
and
at
datastax.
C
I'm
really
invested
in
cork
sandra
is
only
one
component
of
the
things
that
we
can
offer
in
datastax
enterprise,
but
those
benefits
alone
aren't
aren't
isolated
to
what
datastax
the
building
on
and
really
really
are,
I
think
important
to
anybody
who
who
wants
to
have
confidence
in
their
database
and
I.
Think
as
someone
who
has
occasionally
under
mistakes
of
my
own
or
mistakes
of
other,
been
responsible
for
databases
in
production,
that's
something
that
helps
me
sleep
better
at
night.
A
So
you
said
you
know,
as
word
like
rolling
out
new
features
and
things
like
that.
These
tests
come
in
handy.
Can
you
think
of
a
time
you
know
I'm
thinking,
maybe
of
materialized.
Views
was
a
feature
where
this,
where
these
tests,
like
came
in
super
handy
at
finding
failure
cases,
can
you
talk
about
like
maybe
some
time
where
you
recently,
where
you
know
having
this
suite
of
tests
or
having
your
your
work
on
Jepsen
has
been
like
super
helpful
yeah.
C
Totally
so
so
I've
been
sort
of
in
the
stabilization
of
30
in
general.
As
we
prepare
for
release,
it
was
kind
of
useful
across
the
board,
but
again
with,
as
you
mentioned,
a
big
new
feature
like
materialized
views
if
it
becomes
really
valuable.
So
let
me
step
back.
Just
a
little
I've
been
made,
I've
made
it
clear
that
Jepsen
is
this
framework.
We
can
use
to
write
tests
for
a
distributed
system,
but
I,
don't
think
I've
really
highlighted
its
strengths
and
and
the
type
of
testing
it
does.
B
B
C
Size,
but
on
top
of
that,
you
kind
of
have
this
gap.
Jepsen
can
fill
where
you're
doing
randomized
a
black
box
testing
of
a
system
with
injected
failures,
and
so,
when
you're
developing
a
big
new
feature
like
materialized
views,
you
might
not
be
familiar
with
the
ways
it
can
interact
with
the
sort
of
failure
modes
that
you
might
have
seen
through
history
in
an
older
feature.
That's
not
sort
of
this
brave
new
world.
C
So
when
you
really
want
to
gain
confidence
in
that
I
think
randomized
testing
can
really
increase
confidence
in
a
feature
because
if
it
explores
a
far
larger
component
of
the
state
space-
and
it
also
makes
it
clear
how
that
feature
will
interact
with
the
system
at
the
boundaries
where
users
are
using.
So
you
need
to
give
this
black
box
testing.
You
also
have
to
gain
understanding
of
how
the
feature
interacts
with
these
really
cool
modern
driver
implementations
that
are
really
sophisticated
systems
on
their
own.
B
The
as
they've
added
you,
a
type
of
testing
your
you're
talking
about
is
that
that
kind
of
testing
that
is
really
almost
silent,
because
it's
not
what
people
think
of
you
know
when
you
think
of
a
test
like
a
unit
test.
Oh
there's
a
pass/fail,
but
those
DUHS
really
a
use.
The
word
Heisenberg
you
know
distributed
systems
are
the
classically
hard
problem
when
it
comes
to
troubleshooting
problems,
because
you
have
no
idea
what
a
system
is
affecting
a
no
system.
B
C
That's
one
of
the
things
that's
a
one
of
my
favorite
part
of
Jetsons
design
and
I.
Think
something
Kyle
did
really
well
in
building
this
library
we
can
use
is
that
when
you're
running
this
test,
it's
recording
your
observations
about
the
state
of
the
environment
in
the
state
of
the
system
and
the
operations
you're
using
to
modify
it
to
this
history
and
you're
actually
checking
that
history
at
the
end,
because
when
you
find
a
bug
in
a
distributed
system,
especially
a
bug,
that's
evaded
regular
unit
testing
or
integration
testing.
C
It's
usually
pretty
sophisticated,
if
not
in
its
underlying
cause,
in
its
presentation
and
as
a
result
of
these
really
sort
of
complicated,
specific
inner
leavings
that
you're
finding
through
this
randomized
testing.
So
if
you
don't
have
this
big
log
of
all
the
operations,
you've
performed
good
luck.
Finding
and
again
good
luck
reproducing
it
again.
So
when
we
run
these
sort
of
tests,
it's
it's
really
nice,
because
it's
a
harder
to
dismiss
them
as
oh.
A
So
I
got
one
more
question
and
I'll
ask
you
to
kind
of
maybe
put
on
your
prognosticator
hat
for
for
a
second.
So
what
do
you
think
like?
Where
do
you
think
the
this
kind
of
work
that
you've
done
already
is
headed,
or
maybe
just
this
sort
of
area
of
testing
and
distributed
systems
and
failure,
modes
and
stuff
like
that
like?
Where
do
you
know?
What
are
you
excited
about
working
on
next
or
where
do
you
kind
of
see
this
going
in
the
future
sure.
C
So
this
is
a
really
exciting
area
for
me
right
now,
because
I've
been
working
on
the
streaming
systems
for
a
while,
but
in
terms
of
training,
I
was
trained
as
a
mathematician.
So
this
wild
wild
west
of
software
can
be
a
very
scary
place
for
me,
sometimes
and
I'm
excited
that
there's
interest
in
in
ramping
that
down
and
really
understanding
these
systems
and
how
they
run
so
I'll
share
my
thoughts
in
a
second
but
in
terms
of
a
high-level
summary
in
writing.
C
So,
frankly,
I
think
there's
there's
still
a
lot
of
a
lot
of
low-hanging
fruit
in
terms
of
testing
all
of
these
systems
we've
built
just
from
a
randomized
black
box
perspective
that
acknowledges
the
reality
of
failure
conditions,
but
on
top
of
that,
there's
a
really
great
research
from
Peter
overo
about
lineage,
driven
fault
injection
that
doesn't
more
targeted
exploration
of
these
different
failure.
Spaces.
C
There's
really
cool
stuff
out
there
from
a
more
academic
perspective,
where
we
either
use
something
like
a
TI
plus
driven
proof
system
to
actually
prove
properties
which
has
a
huge
learning
curve
and
I.
Think
a
lot
of
people
are
still
trying
to
navigate
how
they
might
use
it
in
actually
building
software
or
kapatic
slide
down
that
ladder.
Rigger
there's
either
exhaustive
or
targeted
model
checking
so
we're
working
at
now
to
rhythmic
level
and
understanding
it
and
some
of
those
I
don't.
C
I
don't
know
how
much
you'll
be
able
to
bring
them
into,
or
anyone
for
that
matter
can
really
bring
them
into
what
they're
doing
tomorrow.
But
I
think
they
can
take
a
lot
of
inspiration
in
those
in
how
they
design
their
testing
frameworks
and
how
they
try
to
explore
the
state
space
of
their
software.
A
C
Yes,
so
I'll
be
talking
about
a
lot
of
these
same
things
in
more
detail
about
Jepson
testing
Cassandra
at
devoxx
in
London,
that's
June,
eighth
or
jun.
9Th
will
be
my
talk,
I'm
not
sure
which
one
maybe
there
will
be
time
for
a
live
demo,
if
not
tracked
me
down
after
I'd
love,
to
give
you
on
I'd
love
to
talk
about
it.
One
last
note
I
know
we're
short
on
time,
but
I
have
to
sneak
this
in
I.
C
Think
there's
also
a
perception
that
things
like
Jepsen
and
styles
of
testing
like
Jepsen,
are
only
important
to
people
or
writing
databases
and
look
I
won't
argue
they
aren't
important
to
us.
They
definitely
are
they
help
me.
Do
my
job,
but
I'd
also
encourage
people
to
really
look
at
that
testing
when
they're
testing
their
own
systems,
they're
building
on
top
of
Cassandra
or
any
other
distributed
system,
because
they
they.
C
A
Yeah
agree:
a
thousand
percent,
especially
these
days
with
the
whole
microservices
thing
being
a
you
know,
being
sort
of
the
trend
as
far
as
architectures
go
like
yeah
you're
right
a
lot
of
people
at
the
system
level.
You
know
not
just
the
database
level
are
dealing
with
distributed
systems
in
their
in
their
data
centers,
so
super
important,
yep,
yeah.
B
C
C
B
Joel
I
really
appreciate
you
coming
on
today:
I
it
was
kind
of
late
notice
when
I
want
a
pink
on
this,
but
I'll
tell
you
I
personally
I'm
one
of
your
biggest
fan
I'm
like
having
a
little
star
moment
here,
so
I'm
glad
you
could
come
on.
I
would
love
for
you
to
come
back
to
this
would
be
great.
I
should
also
mention
you're,
going
to
be
at
the
Cassandra
summit,
I'm
sure
yes,
but
yeah.