►
From YouTube: Introducing Saturn, web3 CDN - Ansgar Grunseid
Description
Saturn is a Web3 CDN in Filecoin's retrieval market built with IPFS-addressable data.
A
Good
afternoon
my
name
is
Ansgar
grunzide
and
today
we're
going
to
go
a
little
adventure
and
journey
into
the
world
of
Saturn.
Now
what
is
Saturn
Saturn
is
a
new
content,
delivery
Network
for
ipfs
and
filecoin.
Let's
get
started
so
every
great
product
starts
with
a
problem
starts
with
an
opportunity
and
Saturn
began
with
this
opportunity.
Ipfs
can
be
faster,
I'm
sure,
you've,
all
retrieved,
content
from
ipfs
and
gone
and
made
yourself
a
cup
of
coffee
and
come
back
and
still
waited
a
little
bit.
That
is
the
core
problem
that
Saturn
is
going
to
solve.
A
So
what
we've
seen
here
with
filecoin
is
file
coins
assembled
an
absolutely
massive
storage
network,
but
leaves
a
lot
to
be
desired
as
far
as
retrievals
go.
You
know
so.
We've
seen
filecoin
assemble
17
expabytes
of
storage,
300
petabytes
stored,
but
the
core
problem
is
where's
web3,
YouTube,
where's
YouTube
built
on
top
of
filecoin
no
web
2
scale
has
been
achieved
yet
on
top
of
ipfs
and
filecoin
specifically,
and
why
is
that?
Well,
the
real
reason.
One
of
the
core
pieces:
that's
missing
is
content
delivery
to
be
able
to
build
an
app
like
YouTube.
A
You
need
to
have
fast,
reliable
content
delivery.
You
can
store
a
file
on
filecoin
that
works
beautifully.
But
now
imagine
you
store
video
like
a
YouTube
video
on
filecoin,
and
you
know
that
video
is
going
to
be
segmented
across
ipfs
stored
on
six
to
ten
storage
providers
in
the
filecoin
network.
But
now
imagine
that
video
goes
viral
and
you
have
hundreds
of
thousands
of
people
all
around
the
world
trying
to
watch
that
video.
A
It
can't
keep
up.
So
this
is
where
Saturn
comes
in
oh
and
there's
one
of
the
piece
here,
which
is
that
we've
already
seen
how
much
like
demand
there
is
for
faster
ipfs
retrievals.
You
know
you
can
just
look
at
the
Myriad
of
irrigated
content.
That's
stored
on
ipfs
every
day
through
gateways
like
nft,
Storage,
web3,
storage,
Etc
and
all
this
content
is
available
in
ipfs
and
available
the
IPS
Gateway,
but
apps
are
waiting
and
chopping
at
the
bit
to
be
able
to
access
this
data
quickly.
A
A
The
big
change
that
Saturn
has
realized
and
embraced
is
that
of
cryptographic
incentives.
So
crypto
incentives
are
rewiring
the
internet.
We
see
this
all
over.
You
know
the
world
of
web
3
we're
centralized
proprietary
services
are
being
replaced
by
decentralized,
open
ones.
You
know,
filecoin
is
a
great
example
of
this,
where
filecoin
is
disrupting
and
replacing
many
of
the
use
cases
of
Amazon
Glacier,
S3
Etc,
and
so
we
at
Saturn
see
this
windfall
see
this
changing
world
and
have
leaned
into
this.
A
Fundamentally
Saturn
is
a
crypto
incentivized
content
delivery,
Network
one
of
the
world's
first
and
these
crypto
incentives
are
what's
used
to
grow
and
scale
the
network
and
we'll
get
into
that
in
a
second
with
some
of
our
goals,
but
the
other
key
piece
of
Saturn.
Is
it
like
all
these
cryptographic
networks,
they're,
trustless
and
they're,
open
source
and
we'll
get
into
some
of
the
technical
details
of
how
we
can
build
a
Content
delivery,
Network
and
a
world?
A
Is
trustless
and
permissionless
and,
lastly,
like
all
these
other
projects
in
terms
of
being
open
source,
Saturn
2
is
open
source.
All
our
work
is
on
GitHub.
We
encourage
all
of
you
to
go.
Take
a
look
through
our
code,
submit
bug,
requests
and
get
involved
with
the
community
and
we'll
get
there
later
too.
A
So
we've
gone
through
the
problem.
We've
gone
through.
What's
changed
our
little
insight
with
crypto
incentives?
What
is
our
solution?
What's
the
product
we're
going
to
build
and
that
product
is
a
crypto
and
incentivized
content
delivery?
Network?
So
if
we
juxtapose
content
delivery
networks
of
the
past,
you
know
we
talk
about
Akamai,
Amazon,
cloudfront,
cloudflare
Etc.
What
these
networks
look
like
is
centrally
managed
top-down
constructed,
Global
networks.
That
means
there's
a
a
room
somewhere
or
a
bunch
of
Amazon
execs
get
together
and
they
say:
where
are
we
going
to
build
our
next
data
center?
A
There's
a
small
team
of
people
that
decides
where
the
network
grows
and
how
the
network
grows
Saturn
fights
of
an
unfair
advantage
on
the
other
side,
what
we
have
built
is
a
Marketplace
for
Content
delivery
capacity
like
a
Marketplace
for
bandwidth.
So
we
have
no
Central
organized
essential
organization
or
Central
team
that
decides
where
to
build
infrastructure
where
to
grow
infrastructure.
A
Instead,
we've
built
Marketplace
Primitives
into
the
network
itself,
where
anyone
all
around
the
world
can
just
spin
up
a
server,
install
Saturn
software
and
add
a
node
to
the
network,
share
content
with
the
network
and
be
remunerated
in
filecoin
for
return
or
in
return.
Excuse
me,
they
don't
have
to
ask
my
permission.
They
don't
have
to
ask
any
of
your
permissions.
They
can
just
go.
A
Do
it
and
that's
a
fundamental
difference
between
prior
content,
delivery
networks
and
our
unfair
advantage
and
we're
going
to
lean
into
this
to
scale
to
millions
of
nodes
globally,
not
just
tens
of
nodes,
hundreds
of
nodes,
thousands
of
nodes,
millions
of
nodes
and
we'll
get
into
kind
of
the
way
Saturn's
network
is
orchestrated
with
the
different
players
and
different
pieces.
They're
different
tiers,
just
like
in
your
computer,
there's
multiple
tiers
of
cash.
A
So
what
are
the
goals?
The
first
and
primary
goal
is
accelerate
ipfs
content
again
when
you
were
going
through
a
neat
demo
here
shortly,
but
when
you
retrieve
content
on
the
IPS
Network
oftentimes,
it's
slow
and
then
there's
the
ipfs
Gateway
as
well,
and
the
IPS
Gateway
leaves
some
room
for
improvement
as
well.
So
our
primary
goal
is
to
accelerate
ipfs
content,
but
our
larger,
larger,
bigger
goal
is
build
the
world's
best
fastest
lowest
cost
CDN
period
in
human
history.
We're
singularly
focused
on
that
goal
and
how
to
get
there
now.
A
A
The
one
I
have
in
front
of
me
and
the
one
you
have
in
front
of
you
to
phones
and
browsers
and
everything
in
between
imagine
if
we
had
a
little
magical
thread
that
we
could
weave
into
every
device
on
the
internet,
so
they're
all
sharing
communicating
like
built
on
ipfs,
that
is
the
world
we're
building
towards
and
why?
Why
are
we
building
towards
that
world?
We
fundamentally
want
to
advance
ipfs
and
Advance
web
free
adoption.
You
know
we
want
to
hasten
the
YouTubes
of
web3.
A
A
What
that
means
is
right
now,
if
you
go
and
buy
content
delivery
from
a
Content
delivery,
Network
like
Akamai
or
cloudflare
AWS,
that
money
goes
from
your
pocket
into
Jeff
bezos's
pocket,
and
what
we
want
to
do
is
take
a
a
very
clever
little
knife
and
cut
that
money
up
and
redistribute
it
into
everyone
who
contributed
to
the
network.
So
what
does
that
mean?
What
does
it
mean
when
anyone
can
spin
up
a
node
and
earn
filecoin
in
return?
A
What
that
means
is
we
don't
have
these
centralized
powers
and
content
delivery
isn't
centralized,
so
we
want
to
democratize
earnings.
So
how?
How
are
we
going
to
do
this?
How
are
we
going
to
achieve
these
goals
by
building
a
crypto,
incentivized
content
delivery
network
with
these
fundamental
pieces
and
we'll
go
into
some
of
the
technical
details
of
these
here
in
a
second,
so
at
a
high
level.
The
first
piece
is
a
retrieval
client.
A
What's
the
retrieval
client,
this
is
the
piece
of
software
that
fetches
content
from
the
network,
and
we've
done
some
very
clever
things
here.
One
of
the
key
pieces
here
is
we
built
something
called
incremental
verification.
So
imagine
this
in
a
trustless
world
with
a
trustless,
Content
delivery.
Network,
you
ask
the
network,
please
give
me
a
picture
of
my
favorite
Kitty,
but
how
do
you
know
that
the
content
delivery?
You
know
the
Saturn
node
operator,
who
we
don't
know
who
they
are?
You
don't
know
who
they
are?
They
really
just
don't
like
cats.
A
Instead,
what
they
do
is
they
want
to
send?
You
their
favorite
like
puppy
picture
instead,
so
you
request
a
sid
of
a
kitty,
but
you
get
back
a
picture
of
a
puppy.
How
do
you
know
without
trusting
the
node
that,
like
they
handed
you,
the
correct
content
you
requested
or
the
incorrect
content
you
didn't,
and
so
one
of
the
things
we've
built
is
in
the
browser
piece
of
software
that
runs
in
what's
known
as
a
service
worker
and
what
it
does.
It
incrementally
verifies
a
car
file,
as
is
retrieved
from
Saturn.
A
A
You
requested
the
Sid,
so
you
have
the
root
Sid
of
the
dag
of
that
file
and
then,
as
that,
dag
is
streamed
back
from
this
trustless
L1
that
this
trustless
node
in
Saturn's
Network,
you
can
see,
does
was
this
first
block
match
the
root
Sid
of
what
I
requested
and
then
all
of
the
leaves
under
that
Sid.
Do
they
match?
You
know
in
the
classic
Merkle
tree,
and
so
you
can
incrementally
verify.
A
The
content
you
requested
is
the
content
you
receive
without
having
to
trust
that
party
whoever's
running
the
node,
and
this
is
a
huge
key
piece
of
what
makes
Saturn's
tick
and
lets
us.
Let
anyone
around
the
world
run
a
node
without
having
to
trust
them.
But
what
do
these
nodes
look
like
so
Saturn's
Network
right
now
is
bifurcated
into
two
buckets
of
nodes.
A
The
first
node
is
what
we
call
L1
nodes
or
layer,
one
nodes,
and
these
are
Big
beefy
servers
and
data
centers,
so
we'll
get
to
kind
of
the
requirements
later
and
of
what
this
looks
like
but
think
about
it.
These
are
the
machines
that
sit
on
the
edge
of
the
network,
not
on
people's
houses,
but
in
data
centers,
with
large
capacity
to
serve
lots
of
users
reliably
and
when
I
say
reliably,
I
mean
the
note
flicker
up
they
don't
flicker
down.
A
These
are
like
semi-permanent
to
very
permanent
parts
of
the
network,
and
next
up
are
L2
nodes
and
again,
if
we
go
back
to
one
of
Saturn's
goals
is
to
unite
Humanity's
Hardware
L2
nodes
run
on
desktops
laptops,
phones,
Etc,
and
they
run
with
software
that
you
can
download
so
coming
soon.
You'll
be
able
to
download
a
piece
of
software
called
station,
which
is
an
electron
desktop.
App
and
embedded
inside
of
station
is
a
little
piece
of
software
for
Saturn.
A
We
call
it
the
L2
or
the
level
2
of
the
network
and
then
running
on
your
machine
with
your
full
permission.
You
know
when
you
can
pause
it
and
stop
it.
If
you
want
is
a
little
cash
and
part
of
Saturn
and
see,
if
you
can
imagine
we'll
go
to
the
diagram
here
in
a
second,
but
if
you
can
imagine
we're
building
a
giant
global
scale,
kind
of
BitTorrent
swarm
of
l2s
that
support
l1s.
A
So
just
like
in
your
CPU
cache,
you
have
level
one
cache
and
level
two
cache
and
you
like
first
check
level,
one
cache,
there's
a
cash
miss
you
go
to
level
two
same
thing
with
Saturn
when
you
send
a
request
to
the
network.
You'll
get
routed
to
the
nearest
L1,
big
beefy
server,
and
it
is
a
cache
Miss.
Then
you'll
fall
back
to
this
kind
of
swarm
of
l2s.
And
how
do
we
power
this?
How
do
we
take
that
little
magical
thread
and
like
tie
all
this
together
was
two
pieces
here.
A
One
is
called
the
orchestrator
and
the
orchestrator
is
a
piece
of
software
which,
when
a
new
node
joins
a
network,
says
okay,
what
is
your
capacity?
How
much
storage
capacity
you
have?
How
much
network
capacity
you
have
and
where
can
we
best
fit?
You
in
the
network
and
the
logs
are
another
key
piece
which
is
to
pay
people
because
people
remunerated
for
their
contribution
to
the
network.
A
We
need
to
keep
track
of
like
how
much
data
was
served
when
you
were
running
an
L2
on
your
machine
and
you
send
an
L1,
you
know,
let's
say
10
gigabytes
that
month
you
need
to
be
remunerated
for
contributing
10
gigabytes
of
the
network,
and
so
we
keep
logs
of
all
this
and
it
fundamentally
in
terms
of
logs.
You
are
earned
file
coin
per
retrieval.
You
know,
there's
multiple
metrics
involved
that
calculate
how
much
you
earn
things
like
time
to
First
Bite
how
quickly
you
were
able
to
send
content.
A
You
know
how
fast
you
were
to
upload
that
content.
How
available
you
are?
Did
you
go
offline?
A
lot.
Did
you
go?
You
know
down
a
lot
and
there's
a
last
piece
here
which
is
kind
of
anti-fraud.
So,
let's
walk
through
what
this
kind
of
architecture
looks
like
at
a
high
level
now
of
the
diagram,
so
on
your
left,
yeah,
it's
your
left!
If
you
look
in
the
little
blocks,
that's
like
the
retrieval
client.
So
this
might
be
a
browser
or
an
in-game
app
or
like
a
potentially
even
a
desktop
app
in
the
future.
A
A
We
can
see
kind
of
this
bittorrent-esque
swarm
of
L2
nodes.
So
imagine
this,
for
example,
the
L1
box
being
like
a
server
and
a
data
center
in
New,
York
City,
and
then
all
the
l2s
or
all
the
users
running
this
station
a
little
desktop
app
on
their
machine
in
the
surrounding
New
York,
City
boroughs,
maybe
New
Jersey,
maybe
Connecticut,
Etc
and
they're
all
in
support
of
all
the
l1s
nearby
and
then
the
background
here,
the
very
far
to
the
right.
We
have
the
filecoin
storage
providers
and
this
is
kind
of
Bedrock.
A
So
when
someone
uploads
a
file
and
they
makes
it
available
on
ipfs
or
stores
in
a
file
coin,
that
is
where
the
content
is
stored
and
Saturn
sits
in
front
as
an
accelerator
of
that
content
and
then
off
and
up
and
to
the
right.
We
can
kind
of
see
the
orchestrator
which
again
is
kind
of
this
software.
That's
tying
it
all
together
and
weaving
it
all
together
and
all
these
nodes
submit
logs
and
these
logs
are
processed
to
calculate
payments
and
also
to
do
anti-fraud.
A
A
So
what
we're
going
to
do
here
in
this
demo
is:
we
are
going
to
Benchmark
in
escalating
performance,
requesting
a
Sid
from
the
ipfs
network
via
the
ipfs
client,
to
requesting
this
content
from
an
IPS
gateway
to
requests
and
content
from
Saturn
and
we're
going
to
Benchmark
the
three,
and
so
what
you
see
before
you
are
nodes
that
are
about
the
SSH
into
around
the
world.
In
the
upper
left,
we
have
Sydney
upper
right.
We
have
Taipei
bottom
left,
we
have
Toronto
and
bottom
right.
We
have
Portland,
so
these
are
VMS
running
in
those
cities.
A
Okay,
so
we
take
a
quick
look
at
this
command.
What
this
command
does
is
there's
a
local
ipfs,
node
running
in
this
VM
or
all
four
of
these
VMS,
and
this
first
little
piece
clears
the
cache.
So
we
go
to
network
we're
not
going
to
fulfill
this
Sid
from
cache,
and
then
we
are
sending
a
curl
request
or
using
curl
to
send
a
request
to
this
local
ipfs
node
to
request
this
Sid
and
which
is
this
Sid.
This
is
what
the
Sid
looks
like
it's
a
little.
A
Well,
you
tell
me
what
that
is,
but
a
little
guy
in
a
little
box.
So
this
is
the
sit
it's
about
a
six
and
a
half
megabyte
video
file.
So
here
we
go.
Let's
slam
enter
here
and
we're
gonna
see
how
quickly
we
can
retrieve
this
file
from
the
ipms
network
around
the
world.
So
here
we
go.
We
see
some
progress,
so
bottom
left
Toronto
finished.
First,
it
finished
with
the
time
to
First
Bite
of
1.2
seconds.
So
the
first
bite
of
that
video
was
retrieved
from
ipfs
in
Toronto
1.2
seconds.
A
A
A
So,
as
you
can
see,
Toronto
finished
really
quickly
again:
140
milliseconds
and
the
rest
of
them
are
sitting
there.
Spinning
trying
to
fetch
this
content.
There
we
go
Portland
2.3
seconds
and
the
first
two
are
sitting
there.
Spinning.
Let's
move
on
they're,
not
fetching
the
content
very
quickly.
So
now,
last
but
not
least,
Saturn.
A
So
if
you
take
a
look
here,
it's
the
same
exact
curl
command,
except
instead
of
ipfs.io,
is
the
URL.
It's
strn.pl,
that's
Saturn's
little
production,
Slug
and
let's
see
boom.
So
there's
Saturn
live
running
right
now
so
again
to
compare
the
performance.
Here
we
take
a
look
Sydney
35,
milliseconds
Taipei
53
Toronto
under
20
Portland
46..
So
these
are
live
numbers
from
live,
Saturn,
Network,
nodes
running
around
the
world,
being
orders
of
magnitude
faster
in
the
ipfs
network
and
the
ipfs
Gateway.
And
you
can
go
verify
this
yourself.
A
You
can
go
spin
up
VMS
in
all
these
locations
and
start
submitting
requests
to
Saturn.
This
is
all
live,
but
don't
wait.
There's
more
here
we
go.
Oh
I
have
even
better
news
than
what
you
just
saw.
We
actually
just
launched
Saturn
two
days
ago,
two
days
ago,
Saturn
went
live,
so
what
that
means
is.
Yes,
all
of
you,
beautiful
people
sitting
here
in
this
room
can
go
run
a
Saturn
node
right
now.
So
what
does
that
actually
mean?
A
Well,
if
an
L1
node
is
what
we
launched,
and
so
that
means
you
can,
if
you
have
a
server
Go
download,
Saturn,
run
Saturn,
share
bandwidth,
accelerate
ipfs
content
and
earn
file
coin
in
return,
and
now
the
requirements
for
an
L1
node
are
relatively
steep
right
now,
10
gigabits,
a
second
upload
more
than
a
terabyte
of
SSD
and
more
than
32
gigabytes
of
RAM.
We
understand
that
kind
of
the
10
cubits
a
second
might
be
a
bit
steep
to
start.
A
However,
we
want
to
start
high
and
see
where,
like
the
network
demand,
is
you
know
what
we
want
to
avoid?
Is
people
adding
nodes,
let's
say
with
a
gigabit
a
second,
so,
like
imagine,
you've
been
your
favorite,
you
know
Ferris
wheel,
you
know
you'll
be
this
tall
to
ride,
but
we
won't
avoid,
is
stating
that
really
low
and
that
a
bunch
of
people
get
on
the
ride.
You
know
they
put
the
harness
on
they
get
clicked
in
and
then
halfway
through
the
ride.
A
Like
sorry
you're,
not
fast
enough
for
the
network
enough
to
kick
you
off,
while
you're
screaming
through
the
air
300
miles
an
hour.
That's
what
we
want
to
avoid,
so
these
are
the
requirements.
If
you
have
a
server
with
that
matches,
these
requirements,
I
encourage
you
go
set
up
a
Saturn
L1
you
can
make
between
0
and
600
following
a
month,
zero
to
three
thousand
dollars
a
month
and
the
more
content
you
serve
to
the
network,
the
more
you
earn
and
our
website
is
saturnstrn.network,
so
you
can
go.
Take
a
look
at
that
site
later.
A
So
what
is
it?
Excuse
me?
What
does
it
mean
to
run
an
L1?
What
does
the
world
look
like
when
you
run
an
L1?
Well,
first
of
all,
you
need
to
set
up
a
filecoin
wallet.
We
got
to
pay
you,
so
you
can
go
to
set
up
a
wallet,
it
can
be
a
shared
wallet,
coinbase
or
binance
or
whatnot,
or
you
can
run
your
own
and
then
you
need
a
Linux
server
that
can
run
Docker
and
we
have
an
ansible
recipe.
A
A
A
So
all
the
things
you
care
about
running
a
Content
delivery,
node
and
the
last
piece
here
is
when
you're
running
a
node
you'll
get
paid
monthly.
So
during
that
month,
you're
contributing
content,
accelerating
ipfs
content
accruing
earnings
submitting
logs,
and
then
you
get
paid
in
file
quantity
of
the
month
for
how
much
you've
contributed
to
the
network.
A
So
what
is
Saturn's
roadmap?
Look
like
so
what
we
did
just
two
days
ago
again
was
we
launched
with
the
L1
nodes
and
coming
up
our
testing
L2
nodes
and
how
to
integrate
these
desktop
nodes
into
the
network
and
then,
what's
next
year
kind
of
our
immediate
priority
for
2023
is
to
scale
the
L1
Network.
So
right
now
we
have
about
90
to
100
nodes
and
again
that's
up
about
50.
A
That's
doubling
Network
size
as
of
like
24
hours
ago,
something
like
that,
and
but
we
want
to
achieve
again,
the
world's
largest
fastest,
cheapest
CDN
and
to
put
in
perspective
cloudflare
has
about
240
points
of
presence
around
the
world.
Amazon
cloudfront
has
about
300,
and
we
want
to
reach
that
and
blow
past
that
again
we
want
to
lean
into
crypto
incentives
to
get
us
way
bigger
way
faster
than
any
other
CDN
in
the
past.
That's
our
priority
for,
like
the
first
quarter,
2023
and
the
other
piece
here
is.
A
We
want
to
achieve
sub
second
time
to
First
Bite
globally
for
ipfs
Content
worldwide.
So,
whatever
Sid,
you
want
to
request
from
ipfs,
if
you
can
request
it
from
like,
as
we
saw
in
the
demo,
from
ipfs
from
the
FPS
Gateway
or
for
Saturn,
we
want
every
single
Sid,
you
request
to
be.
You
know,
retrievable
in
under
a
second
well
China
first
bite
under
a
second
and
then
one
thing
we
look
very
much
towards
doing
is
working
with
the
ipfs
Gateway
team
to
figure
out
how
we
can
help
accelerate
existing
traffic
to
ipfs.io.
A
That's
something
we're
really
excited
about,
and
then
moving
on
to
Quarter
Two
2023.
We
want
to
integrate
L2
nodes
again,
these
are
the
desktop
nodes
into
the
main
Network.
So
that
means
any
of
you
not
don't
just
have
to
run
a
server
to
participate
in
Saturn.
You
can
download
station
run
it
on
your
laptop
or
desktop.
It
can
contribute
to
the
network
and
earn
file
coin
too,
and
then
we
want
to,
like
you
know,
marry
the
two
together
where
L1
nodes
will
cache
missed,
l2s
and
then
on
a
quarter
three
of
2023.