►
Description
In this episode, Aaron catches up with Loren Sands-Ramshaw of Temporal. In it they talk about durable execution, Loren's book on GraphQL, getting through system design interviews, and his upcoming talk at Cassandra Forward. Also, this is the first episode that we're trying out a video podcast format.
A
A
Welcome
everyone,
thanks
for
joining
us
I'm
here
today,
with
Lauren
Sands
ramshaw
learn,
did
I
say
that
right.
B
First,
first
coded
on
my
TI-83
Plus
calculator
in
high
school
and
it
felt
like
magic
and
I
took
a
C-Class
and
and
loved
that
so
I
majored
in
CS
in
college
first
job
was,
was
being
a
TA
and
then
tutoring
and
then
let's
see
full
full-time,
Java,
Dev,
SQL
and
then
I
spent
a
couple
years
working
as
a
security
engineer
at
the
NSA
okay
and
then
after
that
was
a
decade
of
Open
Source
and
starting
startups.
B
Consulting
full
stack,
mobile
and
web
dev
and
writing
a
book
on
graphql
and
then
after
I
finished
the
book
I
spent
a
few
months,
not
a
few
months.
Maybe
three
months
I,
don't
know
it
was
a
long
time
applying
to
to
big
check.
Full-Time
jobs
and
I
had
a
few
offers,
but
was
waiting
on
Google
because
their
process
was
the
longest
by
far
and
while
I
was
doing
that,
I
got
a
a
a
a
new
newsletter
from
swix
Sean
Wang
about
iPhone
for
system
design.
B
My
favorite
part
of
the
interview
process
for
big
Tech
was
the
system
design,
questions,
interview,
questions
sure
and,
and
this
blog
post
sort
of
explained
temporal
and
and
pitched
it
as
like.
This
is
the
thing
that
solves
a
lot
of
your
system,
design
problems
or
distributed
system
problems
in
general
and
and
I
was
like.
B
Oh,
my
God
I
can
use
this
as
the
answer
to
lots
of
my
interview,
questions
and
it's
it's
so
much
simpler
and
easier
to
to
use,
and
he
offered
me
a
Consulting,
gig
sort
of
learning
it
and
and
explaining
it
improving
like
examples
and
stuff,
and
that
was
like
a
dream
job
so
I
did
that
and
then
I
wound
up
joining
full
time.
I
worked
on
the
the
node
runtime
and
now
I'm
switched
to
devrel.
A
B
Sure
yeah
I
guess
the
the
backstory
of
that
was
been
a
long
time.
Dx
Enthusiast
I
was
looking
for.
Let's
say:
I
was
coming
from
like
say
in
2012
and
13
I
was
looking
for
something
that
was
full
stack.
B
Javascript
I
was
coming
from
rails
and
Sinatra
on
the
back
end
and
I
loved
Ruby
and
I
liked
it
more
than
JavaScript,
but
I
was
like
I
have
to
be
JavaScript
on
the
front
end
I,
don't
want
to
rewrite
code
twice
so
I
was
like
I
need
something
full-time
job
I
was
thinking
like
Derby
JS
I
wound
up
settling
in
2014
on
on
meteor
JS,
the
full
Tech
JavaScript
framework,
and
that
was
like
amazing
and
still
to
say,
probably
the
fastest
way
to
I
I
could
build
a
Greenfield
like
mobile,
or
what
about
so
I
loved
having
a
really
fast,
efficient
developer
experience.
B
I
came
across
graphql
and
I
was
like.
Oh
my
gosh.
This
is
this
is
a
great
improvement
over
rest
and
whenever
I
guess
back
then
now
now
when
I
learn
something
new,
I
go
straight
to
Youtube,
but
back
then
I
used
to
always
buy
the
book
and
there
wasn't
a
book
that
I
liked
about
graphql
out
at
the
time
that
I
was
trying
to
learn
it
yeah
I
I
think
this
is
a
great
thing.
B
I
think
it's
gonna
like
improve
developer
experience
for
everyone
in
the
world
and
I
wanted
people
to
know
about
it
and
be
able
to
learn
it
efficiently.
So
I
was
like
I
could
write
a
book
and
and
I
I
did
it
took
a
bit
longer
than
I
thought
it
would.
It
was
like
a
five-year
project
and
wow
almost
900
Pages,
it's
a
it's,
a
very
big
book.
B
I,
don't
think
I
could
be
printed,
I
think
it's
too
long
for
a
print,
but
it's
a,
but
it's
an
ebook
and
a
website
version
and
and
yeah
I
guess
like
I,
it's
the
I
think
the
most
most
comprehensive,
graphql
single
resource
out
there
and
it
it
teaches.
Like
goes
through
the
spec
teaching
you
it
and
then
it
goes
through
a
different
different
front
ends
and
back
end
as
sort
of
a
long
tutorial
style
where
it
teaches
each
concept
with
examples.
Okay,
oh
excellent.
A
Excellent
so
now
you're
going
to
be
joining
us
for
our
online
virtual
event
called
Cassandra
forward.
Why.
B
Sure
so
the
title
of
the
talk
is
code
that
can't
fail,
backed
by
Cassandra.
A
B
Yeah
so
I
start
out
explaining
what
code
that
can't
fail
means
and
that's
basically,
what
temporal
provides
is
a
high
level
of
reliability
for
your
code
and
so
I
explained
durable
execution,
which
is
like
the
category
that
we're
in
which
is
a
a
new
programming
model.
B
And
then
I
go
on
to
talk
about
some
about
how
we
Implement
durable
execution
internally
using
Cassandra
and
then
a
few
Cassandra
learnings.
Okay,.
A
Okay,
so
Lauren,
what
could
you
tell
us
about
about
durable
execution.
B
Sure
so
it's
a
it's
a
new
sort
of
model
of
programming
in
which
you
are
programming
at
a
higher
level
of
abstraction,
where
you
don't
have
to
be
concerned
with
a
number
of
different
distributed
system
concerns
and
you
can
and
and
we
can
and
I
guess,
once
you
write
a
durable
function,
it
is
guaranteed
to
complete
no
matter
what
happens.
Okay,
and
that
includes,
like
your
process,
crashing
your
process
being
killed
by
the
OS.
B
Maybe
it's
out
of
memory,
the
machine
losing
power
code
redeploying
and
you
need
to
like
restart
or
also
transient
failures.
So
if
you
have
a
downstream
service,
that's
temporarily
unavailable
or
you're
like
having
having
network
issues.
B
This
sort
of
takes
care
of
all
that
for
you
and
automatically
retries
with
that
special
back
off
and
does
so
reliably
and
at
high
scale.
So
yeah,
and
also
like
opens
up
some
new
possibilities
like
sleeping
for
30
days
and
not
reliably
working,
and
you
can
also
like
send
rpcs
to
your
your
durable
functions,
to
give
them
instructions
or
or
get
back
information.
You
can
also
have
them
run
an
indefinitely
long
period
of
time
and
you
can
treat
local
variables
as
as
durable
or
as
personable.
B
B
A
Okay,
had
you
worked
with
Cassandra
much
before
this.
B
No
I
had
learned
a
little
bit
for
like
system
design
interview
questions.
B
A
All
right
all
right
was
there.
Was
there
a
can?
Can
you
think
of
a
time
that
you
know?
Maybe
you
were.
You
were
working
on
something
with
Cassandra
and
you
kind
of
ran
into
an
issue
with
it.
Just
just
curious
how
you
you
know
what
it
was
and
how
you
worked
your
way
past
it.
B
Yeah,
so
I
can
I
can
go
through
like
one
of
the
things
for
my
talk
is
there's
there's
an
old
like
2013,
blog
post
from
data
stacks
on
the
cues
being
an
anti-pattern,
for
example,
right
right,
and
we
we
have
a
number
of
of
cues
in
our
system
and
so
I
I
basically
went
over
the
post
and
and
and
shared
the
issue
of
having
having
a
lot
of
taking
a
while
to
go
through
a
lot
of
tombstones
after
you've,
Deleted
things
from
the
queue
and
adding
to
the
to
the
where
Clause,
so
that
you
can
tell
Cassandra
where
to
start
looking.
A
Yeah
yeah,
okay,
okay,
so
so
what
would
you
say
your
your
experience
was
with
Cassandra
was
it
you
know
hey.
This
is
hey.
This
is
great
or
maybe
hey.
This
needs
a
lot
of
work
or
you
know.
I
know.
I,
know
that
developer
friendliness
is
something
that
that
we've
been
working
on
to
improve
for
for
quite
a
while,
so
I'm
I'm
curious
to
know
you
know
from
a
from
a
newcomer.
You
know
what
was
your
you
know.
What
was
your
perspective
on
that.
A
B
Cells,
columns
and
and
tables
column,
families
and,
and
things
like
that,
I
found
that
the
the
day
Stacks
glossary
was
was
a
great
help.
B
I
I
guess
more.
More
generally,
like
the
experience
of
my
company
has
been
really
positive
in
that
okay,
it,
it
is
allowed
us
to
scale
to
a
higher
right
load
than
SQL,
and
that's
that's
enabled
us
to
to
have
a
provide
a
high
scale
cloud
service.
A
All
right
Lauren!
So
if
there's
something
that
you
could
tell
our
listeners
that
you
know
when
they,
when
they
tune
into
Cassandra
Ford,
what
could
they
expect
to
learn
from
your
talk.
B
Well,
yeah
I'll
I'll
definitely
give
a
more
in-depth
explanation
of
what
durable
execution
is
with
some
practical
code
examples
and
then
we'll
see
how
we
Implement
that,
like
the
system,
design
of
the
temporal
server
internally.