►
Description
Link to blog: http://bit.ly/1qMQLEs
A
All
right
looks
like
we
are
live
another
edition
of
this
week
in
Cassandra,
as
you
can
see,
John
Haddad,
your
normal
host
is
not
with
us
this
week.
He
actually
took
a
couple
of
days
off
and
if
you're
wondering
to
yourself,
you
know
like
he's
an
evangelist
like.
Doesn't
that
mean
he's
like
off
all
the
time
yeah
that's
kind
of
what
the
job
is
typically
like,
but
he
does
actually
get
vacations.
Sometimes
you.
A
So,
as
you
can
see,
Patrick
McFadden
chief
evangelist
he's
with
us
and
then
Adam
Hudson
from
data
scale
is
also
with
us.
He's
our
guest
this
week
we're
going
to
be
talking
a
little
bit
about
data.
Modeling
Adams
been
doing
a
series
of
blog
posts
on
data
modeling,
so
that
should
be
really
interesting.
A
But
if
you're,
following
along
on
planet,
Cassandra
blog
post,
there's
a
couple
a
couple
of
the
posts
in
the
list
there
that
we
want
to
call
out-
and
the
first
thing
we
wanted
to
talk
about-
is
Patrick-
had
a
blog
post
this
week
about,
like
writing
a
successful
abstract
for
the
Cassandra
summit.
Cfp
call
for
presentations,
that's
actually
coming
up.
I
think
the
deadline
is
early
next
week,
like
the
first
of
June.
Maybe
is
that
right.
B
Yeah
wednesday
june
first
is
a
deadline,
and
so
I
wrote
the
blog
post
because
I
get
a
lot
of
questions
over
and
over
like
oh,
I
was
thinking
about
it,
but
I'm
not
really
good
at
writing
abstracts,
and
so
I
hope
that
it
helps
a
few
folks
with
that
I
mean
it.
This
isn't
a
contest.
If
you
need
help,
ask
if
you
want
to
show
someone
like
myself
or
Luke.
If
you
want
to
show
us
your
abstract
before
you
submit
it
and
say,
is
this
a
good
one
or
what
else
would
you
add?
B
That's
perfectly
legal.
This
is
not
a
contest,
but
a
few
points
about
in
that
that
blog
that
I
really
feel
are
important.
Is
you
know
the
be
detailed,
your
abstract?
The
the
review
committee
for
the
Cassandra
summit
is
made
up
of
Allah
people.
It's
mbps
its
people
from
community
and
it's
just
people
who
care
about
what's
happening
with
the
patch
of
Cassandra,
so
they're
going
to
read
through
these
abstracts
and
there's
a
lot
of
them,
and
we
only
have
so
many
slots
and
if
yours
doesn't
stand
out,
it
has
a
chance
of
getting
left
behind.
B
So
that's
really
what
I
want
you
to
think
about
and
if
you're
on
the
fence
of
whether
you
want
to
return
in
an
abstract,
you
know
talk
about
something
think
about.
You
probably
do
have
a
story
and
I
say
this
really
generally,
but
it
always
works
out.
You
look
at
what
you
have
done
and
it
is
unique,
even
though
you
may
have
done
something
that
sounds
very
similar.
Every
implementation
is
different
and
if
you
think
about
what
you've
done,
no,
they
could.
B
You
know
Adam
you've
been
to
summit,
you
know
what
it's
like
in
the
hallways
right,
that
conversation
you're
really
starting
a
conversation,
and
you
may
have
some
nugget.
That's
really
important,
so
be
really
detailed.
Put
some
detail
into
your
abstract,
but
really
just
tell
your
story:
I
did
those
are
the
ones
that
stand
out
the
one,
the
abstract
that
says
we
we
put
cos
and
reproduction
and
we
learned
things.
That's
just
I.
Think
I've
actually
seen
that
one.
Have
you
seen
that
one
Luke
yeah
I.
A
You
know
and
and
it's
kind
of
frustrating
sometimes
like
in
particular,
like
you
know,
if,
like
you,
you
know
of
the
speaker
or
like
you
know,
you
know
the
company
they
work
for
or
something
like
that.
You're,
like
hey,
you
know,
like
I,
know,
they're
doing
really
cool
stuff
with
Cassandra.
What
did
they.
B
Subtract
exactly
and
that,
but
that's
something
we
can
I
mean
in
some
some
cases
there,
because
if
there's
enough
people
in
the
speaking
committee
that
know
who
it
is,
but
you
can't
count
on
that
and
you
really
shouldn't
and
most
times
those
talks
just
get
rejected.
I'm.
Sorry,
there's
not
enough
here.
If
this
is
your
abstract
I,
don't
even
wanna
know
what
you're
it's
going
to
be
like
it's
going
to
be
that
bad.
B
So
that's
the
thing
and
you
want
to
think
about
what
would
you
want
to
see
if
you're
going
to
the
summit
or
you
go
to
any
any
conference?
What
would
you
like
to
see?
So
those
are
just
some
general
tips,
but
regardless,
no
matter
what
get
it
turned
in,
it's
gonna
be
soon
Adam.
You
got
yours
in
yet
working
on
it
working
on
it
all
right!
B
A
C
Now
no
I
can
talk
about
it,
so
you
know
data
scale,
a
company
that
I
work
for
is
we're
in
the
Cassandra
as
a
service
hosting
space.
So
we
do
a
lot
of
monitoring
of
everybody's
notes
all
at
once,
so
in
the
tools
that
we
are
using
and
currently
we've
implemented
it
with
data
dog,
so
that
is
kind
of
where
I'm
going.
C
You
know
the
things
that
when
people
have
problems,
they're
always
I'm
the
point
person
that
gets
to
help
them
out
so
I
get
to
see
a
whole
bunch
of
interesting
problems,
so
I'm
constantly
digging
into
the
metrics
and
the
jmx
being
so
that's
kind
of
where
I'm
going
and
we
use
data
dog.
We're
really
proud.
B
A
So-So
one
other
point:
I
wanted
to
throw
out
there
about
the
CFP,
because
I
was
going
back
and
forth
on
chat
the
other
day
with
someone
of
employees
of
datastax.
He
was
working
on
a
proposal,
but
he
asked
me
you
know
about
like
how
serious
like
this,
you
know.
Does
my
talk
have
to
have
a
paper
behind
it
because
you
know
like
cfp,
like
in
a
lot
of
places,
is
actually
call
for
papers.
So
you
know
how
serious
does
my
talk
have
to
be
and
and
I
said
to
him.
A
A
Like
some
people
take
the
tack
of
humor
some
people,
you
know
it
absolutely
can
be
serious
and
can
have
a
you
know
can
have
a
paper
or
something
behind
it.
If
that's
what
you're
going
to
submit
but
overall
just
do
something
that
you're
passionate
and
that
you
know
that,
hopefully
our
super
knowledgeable
about
that.
You
think,
like
patrick
said,
other
people
will
want
to
hear
about.
A
That
would
be
my
advice
if
you're
putting
together
your
last
minute
CFP,
so
alright
switching
gears
a
little
bit
another
looking
at
another
one
of
the
blog
posts
in
the
list
on
the
plant
sandra
blog
post,
if
you're,
following
along
so
Ryan
stila,
who
works
for
datastax
road.
Another
excellent
blog
post,
he's
kind
of
a
we've,
had
him
on
actually
on
this
weekend,
Cassandra
before
and
he
wrote
another
excellent
blog
post
about
properly
testing
Cassandra
testing
your
sander
cluster
before
you
get
to
production
and
kind
of
like
what
you
know.
B
It's
interesting,
you
see
the
background
here
where
I'm
sitting
in
a
video
studio
right
now
today,
and
it
wasn't
because
of
this
particular
hang
out.
It
was
actually
because
I'm
filming
the
performance
and
operation
tuning
course
for
Academy
datastax
Academy
and
I
just
did
a
module
on
cluster
sizing
and
I.
It's
one
of
the
points
that
I
making
that
it's
not
a
hard
science,
if
you'd,
somewhat
subjective
and
it's
based
on
a
lot
of
things
like
workload
and
what
capacity
you
have
in
two
nodes,
you're
using,
but
there's
a
lot
of
understanding.
A
A
You
know
with
people
with
varied
kind
of
workloads
in
in
production,
so
I
didn't
realize
some
of
the
like
some
of
the
variability
that
you
could
get
depending
on
like
compaction
strategy,
and
you
know
I
always
knew
let
you
know
like
you
know
that
definitely
had
a
big
impact.
But
you
know
just
the
kind
of
wide
swing
between
you
know
read
heavy
workloads
where
you
might
do
a
terabyte
per
node
versus
right
heavy
workloads.
Maybe
you
can
get
away
with
way.
Denser
kind
of
two
three
terabytes
per
know.
B
Think
the
thing
to
remember
too,
is,
if
you
screw
it
up,
you
do
have
some
outs
like
you
can
add
notes
that
you
can
decrease
the
amount
of
notes
being
able
to
variable
during
that
knob
on
your
cluster
size
is
an
option,
but
I
always
tell
people
is
make
sure
you
have
that
option.
I!
Think
that's
where
you
guys
work
here.
Adam!
Isn't
that
right,
the
old
knob
for
cluster
size.
You
get.
B
B
C
A
So
so,
just
a
one
thing
that
that
Ryan
mentions
in
this
blog
post,
which
kind
of
segues
I
guess
a
little
bit
and
to
Adam
and
what
you've
been
writing
about
recently,
and
so
he
talks
a
little
bit
about.
You
know
one
one
thing
that
he
sees
all
the
time
in
people
that
have
problems
in
production
is
data
models
that
are
sometimes
he
says,
data
models
that
are
sometimes
naively
lifted,
one
for
one
from
Oracle
data
models
that
weren't
scaling.
A
You
knows
and
they
expect
to
pick
them
up
and
pick
that
data
model
up
and
move
it
over
to
Cassandra
and
all
of
a
sudden.
You
know
they
get
this
magic
scaling
fairy
dust,
sprinkled
on
it,
because
Cassandra
is
so
awesome
and
you
know
it'll
all
just
well.
I'll
just
work
like
they'll
all
be
magically
happy,
and
that
is
definitely
not
a
recipe
for
success
with
Cassandra
and
add.
C
Definitely
a
lot
of
that
relational
stuff
that
comes
to
mind
when
you
look
at
it,
but
you
can't
just
pick
up
that
that
model
and
apply
it
so
adding
that
extra
layer
of
you
know
going
after
your
queries.
First
building
your
models,
you
know
knowing
your
entity
is
super
important.
You
know
thinking
back
from
you
know:
we've
all
got
that
DBA
background
least
most
of
us
do.
Then
it
hadn't
normalized
things
and
bring
things
up
to
the
3rd
4th
5th
degrees
of
normalization.
C
Those
are
great
now,
but
then
turning
those
things
into
real
queries
that
are
actually
going
to
be
efficient.
You
know
putting
them
and
I.
Do
a
lot
of
I
do
a
lot
of
professional
services
in
my
position,
so
I
get
to
go
in
and
see
hey
our
data
models,
not
quite
working
out
where
we're
not
performing
very
well.
C
Well,
let's
see
oh
yeah
look
at
how
wide
rows
you
if
you've
implemented
here
because
of
XYZ,
you
know
we're
just
trying
to
go
in
and
then
figure
out
exactly
what
the
data
model
is
that
they're
trying
to
achieve
and
knowing
where
they
came
from,
whether
it
be
my
sequel
or
or
some
other
relational,
that
they've
they've
adopted
Cassandra
to
jump
right
in
so
I.
Think
in
modeling
really
there's
there's
a
lot
of
meat
on
the
bone.
C
C
A
For
somebody
who's
listening
to
this,
who
doesn't
know
like
what
you're,
what
you're
talking
about
when
you
say
logical,
physical
conceptual
like
could
you
break
down
just
kind
of
real
quick,
like
you
know
that
you've
done
two
parts
in
this
in
this
series
so
far
and
I
think
you
did
the
logical
and
the
conceptual
data
model,
but
you
have
done
the
physical
yet
that'll
be
on
this
weekend
or
next
week
or
in
the
next
few
weeks.
So
I'm.
A
C
C
What
not
and
then
you
you
think
through
well
that
conceptual
is
going
to
have
certain
attributes
characteristics,
so
a
person
may
have
name
and
age
and
gender,
and
things
like
that.
Then
you
put
it
into
once.
You've
built
those
relationships,
and
you
know
purpose.
They
may
have
a
relationship
with
a
another
person,
whether
that
be
via
genealogy
or
you
know
some
some
other.
You
know
marriage
or
something
like
that.
C
What
you
do
is
you
build
that
into
logical,
so
you
take
that
and
you
start
to
put
it
into
what
you
would
think
of
as
a
table
form
so
you're
starting
to
build.
Oh
well
now
I've
got
a
person
table
and
we're
starting
to
link
the
idea
of
an
entity
to
a
table,
and
now
all
of
our
attributes
are
becoming
columns
and
then
that's
where
the
logical
is,
but
you
don't
want
to
leave
it
just
that,
because
you
want
to
add
the
Cassandra
part
to
it,
which
is
you
want
to
think
of
it
inquiries?
C
So,
if
you're
coming
from
that
DBA
world,
you
want
to
think
of
it
like
what
are
the
views
that
I
use
a
lot?
You
know
where
you
have
a
whole
bunch
of
joints
together
in
the
relational
world,
because
you're
not
going
to
have
any
of
those
joined.
You
want
to
put
all
that
data
all
in
one
row.
So
that's
where
you're
starting
to
put
it
into
logical
form
and
then
we'll
talk
about
the
physical
afterwards,
which
physicals
we
were
into.
What
does
it
look
like
I'm
disk?
How
does
it?
How
does
it
perform?
C
A
Things,
like
maybe
data
locality,
like
you
know
how
do
I
pick
a
partition
key
so
that
I'm
only
so
I'm
hitting
a
few
nodes
as
possible
yeah?
How
does
it
look
on
disk?
You
know
+
tab,
keys,
yep
makes
total
sense.
Yeah,
Patrick
and
I
I
know
like
Patrick
every
every
Cassandra
day
we've
ever
done.
You
always
give
the
data
modeling
talk
and
we
tend
to
give
sort
of
a
condensed
version
of
that
when
we
kind
of
when
we
do
those
Cassandra
day
fox
or
you
do
since
you
do,
though,
the
data
modeling
talk,
yeah.
B
Well,
I
mean
it's
really
just
kind
of
breaking
through
that
barrier.
I
think
most
of
the
and
I
would
do
this
raising
hands
thing.
Where,
like
how
many
of
you
have
done
relational
data
modeling,
it's
like
a
hundred
percent,
you
know
it
so,
okay,
we
can
all
start
from
there
and
it
does
that
our
talk
or
I
do
to
talks
usually.
But
a
couple
hours
is
not
enough,
but
it
really
does
start
opening
doors
and
that's
all
you
can
really
hope
for.
B
Like
Adam
said
you
know
breaking
through
the
okay,
this
isn't
so
scary,
but
it's
the
difference
and
it
would.
It
will
help
developers
work
through
the
path
that
they're
going
to
take
and
I
would
hope
that
if
they
really
want
to
get
I
tell
people
this
all
time.
If
you're
getting
paid
to
do
this,
then
you
should
do
this.
You
know
you
should
go
to
academy
and
take
the
data
modeling
course
there,
because
you
need
to
cover
all
the
bases
you
need
to
be
professional
and
understand
all
the
parts
and.
A
C
The
lion's
share,
so
the
majority
of
what
we
do
or
what
I
go
out
and
do
is,
is
fixed
thing.
Their
data,
or
you
know,
trying
to
looking
at
their
data
is
one
thing
and
you
can
kind
of
understand
it,
but
stopping
and
asking
them.
What
are
you
trying
to
do
and
then
going
okay
now?
This
is
really
not
how
you
want
to
approach
that
you
know
because
you're
you're
doing
this
and
you're
you're
looking
at
too
many
nodes
here.
C
That's
why
you're
you're
kind
of
hitting
a
bottleneck-
and
you
can't
do
you
can't
do
this
or
you
really
need
to
you-
know
trying
to
use
one
table
for
too
many
things
like
no,
no,
no,
no,
no
duplicate
as
much
data
as
you
want.
That's
that's
the
key
you
wanna,
you
know,
spread
the
same
data
out
all
over
the
place
just
to
make
those
queries
so
much
faster.
So
the
majority
of
what
of
what
we
do
is
going
out
and
doing
data
modeling.
A
C
A
A
They
have
a
senior
Cassandra
DBA
position
open
as
well,
so
if
you're,
a
DBA
out
there
looking
for
work,
definitely
check
that
out
as
well
and
then
lastly,
cfp,
like
we
mentioned
at
the
beginning
of
this
week
in
Cassandra,
we're
coming
up
on
the
CFP
deadline:
Patrick
yeah
Epis
in
yeah
yeah
right
away.
Let's
do.