►
Description
Learn more about the Filecoin Retrieval Market to better understand how Filecoin deals are negotiated.
A
Hello,
my
name
is
patrick
woods,
heads
and
I'm
a
technical
program
manager
at
protocol
labs,
and
we
are
talking
today
about
far
coin
retrieval
market.
The
agenda
is
we're
going
to
begin
with
some
motivation,
not
that
we
need
any
motivating
for
everyone
listening
in
to
farc
on
orbit,
because
I'm
sure
everyone
is
super
excited
about
the
amazing
presentations
they've
seen
so
far
we're
going
to
motivate
the
topic
of
retrieval
markets.
A
We're
then
going
to
have
an
ecosystem
spotlight
with
one
of
the
really
exciting
teams
in
our
ecosystem,
for
retrieval
markets
mile
and
then
for
the
second
half
we're
going
to
have
a
q
a
with
mile
and
find
out
yeah
what
they've
been
up
to
and
their
thoughts
on
retrieval
markets
in
general.
A
So
we
begin
with
the
motivation.
The
internet
is
like
a
road
network
with
traffic
jams
and
crowded
roads
in
a
road
network.
Let's
say:
let's
take
amazon
books,
for
example,
they
have
distribution
centers
all
around
the
world
and
so
that
they
can
deliver
books
to
people
who
have
ordered
them
online
as
quickly
as
possible,
and
the
same
can
be
said
for
the
internet
and
the
way
that
the
internet
provides
files
around
the
world.
So
they
can
be
retrieved
very
quickly
is
by
something
which
is
called
a
cdn.
A
content,
delivery
network
and
cdns.
A
A
I
also
want
to
mention
here
that
web
2
or
traditional
cloud
providers
they
separate
file
storage
pricing
from
file
retrieval
pricing.
So
you
pay
to
store
your
files
in
amazon
s3,
and
then
you
pay
also
to
retrieve
files
from
amazon
s3,
and
then
you
may
pay
on
top
of
that
for
the
cdn
layer.
A
A
But
you
know
that's
a
very,
very
condensed
metric
on
the
right
hand,
side
we
have
performance
in
africa
where
we
can
see
that
the
retrieval
times
for
the
cdns
is
is
much
lower
more
than
twice
as
slow
for
the
top
provider,
and
this
data
is
from
cdnperf.com
and
even
here
I
think,
we've
condensed
the
data
quite
a
lot,
because
they're
optimized
to
perform
in
cities,
but
in
more
sparsely
populated
regions
and
territories.
A
It
can
take
much
much
longer
for
users
to
retrieve
their
data,
and
here,
in
fact,
we've
got
a
list
of
countries
that
don't
have
a
single
point
of
presence
for
one
of
these
web
2
cdn
providers
and
lots
of
them
are
small
islands.
So
perhaps
it's
not
particularly
necessary.
They
can
fetch
that
data
in
a
performance
time
from
from
next
door
countries,
but
there
are
also
some
pretty
big
land
masses
here.
We've
got,
for
example,
the
democrat
democratic
republic
of
the
congo.
We've
got,
greenland
we've
got
the
central
african
republic,
sudan.
A
These
are
big
land
masses
without
a
single
cdn
pop,
and
that
makes
us
ask
the
question
why
why
has
they
not
got?
One
of
these
cdm
points
to
presence
and
it's
I
imagine
to
do
with
the
business
model.
It
doesn't
make
sense
for
these
companies
to
build
a
data
center
there
to
go
to
power
supply
to
these
countries
and
to
you
know,
to
offer
this
service
when
there
isn't
the
appetite.
A
So
we've
spoken
about
the
geographic
issues,
I
suppose,
with
web
2
cdns,
I'm
now
going
to
speak
a
bit
about
the
the
market.
The
pricing
of
retrieving
data
from
in
this
case,
amazon,
s3,
and
this
this
slide-
is
taken
from
a
blog
from
cloudflare
called
aws,
egregious
egress,
which
is
a
great
name,
and
they
estimate
with
their
calculations
that
customers
in
u.s,
canada
and
europe,
who
are
retrieving
files
from
s3,
are
paying
an
8
000
markup
to
retrieve
their
data
to
80
times
what
it
costs
amazon.
A
To
put
on
that
service
of
retrieving.
That
file
is
what
they're,
if
what
you're
paying
to
retrieve,
which
is
a
huge
markup.
A
A
The
first
part
of
this
division
is
just
to
build
the
web3
cdn,
it's
a
fundamental
building
block
of
web3,
and
if
we
can
provide
this
service
so
that
the
app
developers
can
can
have
a
cdn
in
the
web3
stack
and
it
can
be
the
same
sort
of
retrieval
times
and
price
as
a
web.
2
ccdn
and
that's
a
huge
win.
But
then
can
we
go
one
step
further,
because
actually
web3
properties
and
constructs
may
allow
us
to
outperform
traditional
cdns
in
certain
ways.
A
For
example,
the
content
addressable
nature
of
data
on
filecoin
means
that,
firstly,
these
points
of
presence
won't
have
won't
have
to
worry
about
kind
of
deduplication
of
data.
They
can
just
store
the
file
once
not
multiple
times,
and
we
also
don't
have
to
worry
about
caching
validation.
If
data
becomes
stale,
we
we
know
because
it's
been
updated
and
it's
a
new
cid.
A
So
we're
getting
a
lot
of
you
know
by
the
kind
of
nature
of
content.
Addressable
data
we're
getting
a
lot
of
stuff
free,
which
is
actually
really
useful
for
a
cdn.
We
also
it's
permissionless,
so
anyone
should
be
able
to
spin
up
one
of
these
points
of
presence
in
the
web3,
cdn
and
incentivized
by
by
crypto
economic
mechanisms.
A
A
This
won't
be
a
problem
anymore,
because
we've
got
a
much
more
distributed
network,
a
lot
of
points
of
presence
from
whom
we
can
fetch
the
data
and
then
also
smart
redistribution
of
content,
points
of
presence,
there's
a
really
low
barrier
to
entry
to
spin
up
a
point
of
presence.
So
if
there's
a
market,
people
can
spin
one
up,
they
can
start
serving
files
and
they
can
start
to
understand
where
data's
being
requested
and
it's
much
more
kind
of
dynamic
than
the
kind
of
static
more
hub
and
spoke
approach
of
the
web.
2
graph.
A
So
I
hope
I've
set
the
scene
for
for
retrieval
markets,
it's
about
this,
this
web
3
cdn
and
how
we
can
incentivize
people
to
come
point
to
presence
in
the
network
and
the
work
we're
doing
at
protocol
labs
on
retrieval
markets
is
actually
a
real
ecosystem.
Focused
works,
yeah,
work
stream,
so
we've
got
so
many
amazing
teams
in
the
ecosystem,
all
contributing
and
one
of
which
is
mile-
and
we
have
alex
and
thomas
from
mile
here
to
speak
to
us
today.
B
Yeah,
thank
you
patrick
for
that
great
introduction
and
thank
you
valcoin
for
having
us.
So
as
patrick
mentioned,
I'm
alexander,
camuto,
co-founder
and
ceo
of
mile
so
very
succinctly.
Mile
is
a
content
delivery
network
that
is
resilient,
scalable
and
peer-to-peer
to
suit
the
long-term
needs
of
web3
applications.
B
So
as
a
member
of
the
falcon
community,
why
should
you
care
about
us?
Well
we're
building
the
cdn
on
falcoin
and
we're
bringing
in
new
capabilities
to
the
ecosystem,
so
a
cdn
and
retrieval
market
like
ours
can
really
accelerate
retrieval
from
storage
miners
and
by
doing
so
can
extend
filecoin's
capabilities
beyond
cold
storage
to
hot
storage
and
caching
as
well,
and
secondly,
because
the
requirements
for
running
a
retrieval
miner
are
much
less
than
a
storage
miner.
We're
also
lowering
the
barrier
to
entry
to
become
a
node
on
the
falcoid
network
and
to
earn
fill.
B
B
We're
really
developing
interfaces
that
allow
anyone
to
easily
extend
our
cdn
by
spinning
up
a
new
node
in
an
instant
and
what
we're
aiming
for
is
for
this
ease
of
use
to
create
a
distributed
network
of
caches,
with
no
single
point
of
failure
and
where
the
cost
of
adding
a
new
node
is
close
to
zero
dollars
and
we're
really
aiming
to
bring
content
as
close
as
possible
to
end
users
for
maximum
performance
by
way
of
this
network
and
at
the
core
of
what
we're
building
is
the
golang
interface
of
the
mile?
B
A
node
can
serve
content
that
they
have
cached
to
a
requesting
content
and
payments
whereby
a
node
can
pay
another
node
in
fill
for
delivering
content
so
on
their
own.
These
operations
are
really
simple,
but
you
can
kind
of
compose
them
and
coordinate
them
across
the
network
to
create
more
complex
capabilities,
but
paired
with
these
pop
nodes.
B
The
first
is
that
it
doesn't
rely
on
centralized
gateways
to
retrieve
content,
so
it
enables
a
pure
peer-to-peer
connection
between
a
browser
and
a
pop
node,
and
this
is
kind
of
a
rare
feature
in
the
peer-to-peer
space,
and
the
second
is
that
we've
enabled
sponsored
crypto
payments
whereby
an
app
developer
or
company
can
sponsor
and
pay
for
their
users
payments
when
they
retrieve
content.
So
we
have
an
early
version
of
this
client
running
up
on
our
website
at
mile.network.
A
Thank
you
thanks.
So
much
for
that
intro
guys
and
yeah
I'd
like
to
begin
by
just
saying
it's
been
yeah
a
pleasure
kind
of
starting
to
work
together
recently
hearing
more
about
what
you've
been
up
to
a
mile
and
seeing
the
really
nice
uis
that
you've
built
firstly
for
your
your
website
and
for
your
products,
so
it
always
makes
it
easy
to
get
into
into
in
the
product.
When
you
can
look
at
it.
A
It
looks
nice
in
the
first
place,
so
yeah
first
question,
I'd
like
to
ask
is
just
kind
of
speaking
generally
we've
touched
on
this
a
bit
already,
but
why
do
you
think
we
need
a
web3
cdn.
B
Yeah,
I
mean
a
very
fair
question.
I
mean,
as
you
mentioned,
you
know,
centralized
cdns
kind
of
serve
their
current
purpose,
really
well
in
the
areas
that
they
are.
You
know
they
deliver
content
in
a
really
reliable
fashion.
B
I
think
some
of
them
serve
trillions
of
requests
a
day,
so
nothing
really
to
spit
at
in
terms
of
performance,
but
the
problems
that
we
for
ceo,
twofold
and
you
kind
of
touched
on
this.
The
first
is
that
you
know
these
cdns
kind
of
serve
the
the
world
unevenly
and
they
haven't
really
expanded
beyond
big
population
centers-
and
you
know
the
us,
the
eu
and
asia,
because
it
it
really
isn't
profitable
for
them
to
do
so,
and
the
second
is
that,
because
of
the
management
of
this
infrastructure,
is
super
centralized
when
it
fails.
B
It
fails
very,
very
visibly,
as
you
mentioned,
with
the
the
fastly
downtime
in
in
june
and
in
our
quest
to
kind
of
make
the
the
web
ever
more
reliable.
B
We
should
kind
of
aspire
to
reduce
the
risk
of
these
kinds
of
failure,
and
it's
really
our
thesis
that
a
decentralized
cdn
can
kind
of
supplement
existing
cdns
by
you
know
providing
a
dense
network
of
content
delivery
points
that
can
expand
to
where
existing
cdns
can't
go.
They
can
do
so
at
a
really
low
cost
and
because
they
don't
have
a
single
point
of
failure,
they
can
provide
a
more
reliable
baseline,
basically
for
content
delivery.
A
Awesome
thanks
yeah.
I
love
the
idea
of
the
kind
of
zero
cost
to
spin
up
a
pop
node,
the
idea
that
you
know
you
it
could
be
on
someone's
phone
or
or
laptop.
You
know.
These
are
just
super
computers
that
people
are
carrying
around
so
that
the
network
becomes
much
more
dynamic
and
okay
yeah
low
barrier
to
entry.
So
you've
also
spoken
a
bit
about
this,
but
would
you
be
able
to
go
into
a
bit
more
detail
about
what
mile
are
currently
focused
on
and
what's
kind
of?
What's
interesting?
You
at
the
moment.
B
C
Yeah,
we've
we've
been
working
on
kind
of
making
it
easy
to
run
clients
in
the
browser
so
that
it
doesn't
really
disrupt
the
regular
front-end
experience
on
how
you
would
normally
build
your
ui,
and
you
know,
consumer
content
and
we've
actually
had
two
options
available.
Now
one
of
them
is
running
the
the
client
into
a
service
worker.
C
So
you
it's
essentially
acting
as
a
proxy,
so
you
just
make
normal
http
requests
as
if
you
were
targeting
an
ipfs
gateway,
for
example,
and
then
the
node
that's
running
in
a
different
worker
in
a
different
thread,
essentially
in
the
background,
we'll
kind
of
intercept
those
requests
and
then
find
the
provider
and
then
make
those
requests
immediately.
C
So
you
actually
don't
need
to
hit
any
gateway
or
anything
like
that
and
you're
really
running
a
node
inside
of
the
browser
and
then
for
those
who
don't
want
to
add,
for
example,
a
service
worker,
because
it
still,
you
know,
adds
a
little
bit
of
you
know,
download
time
on
the
initial
load
time,
for
example,
or
others
who
don't
want
to
use
this,
we're
also
kind
of
running
those
clients
inside
of
cloudflare
workers.
So
it's
kind
of
an
intermediary
where
it's
already
running
on
the
edge
and
can
already
find
very
closed
nodes
from
there.
C
So
you
know
it's
as
if
you
would
request
us
as
well
normal
http
requests,
but
then
you
just
spin
up
a
client
node
in
the
browser,
sorry
in
on
the
cloudflare
edge
really
quickly.
So
that's
also
something
we're
using
to
make
it
easier.
But
still
you
know
quite
decentralized
to
to
query
content
from
our
network.
A
Awesome
thanks
so
much
it's
nice
to
go
into
a
few
more
of
the
kind
of
technical
details
there,
as
well,
just
to
give
people
a
flavor
of
what
this
is
really
about.
So
I
know
that
you've
been
traveling
the
world
recently
looking
to
to
deploy
props
top
nodes
around
the
place.
So
my
next
question
is:
who
wants
to
use
this?
Who,
who
do
you
find
your
users
who's
interested?
What
what
sort
of
yeah
categories
of
users
are
you
looking
at.
B
You
know
deeply
crypto
native
organizations
like
daos
who
have
you,
know
community
treasuries
and
cryptocurrency,
and
they
need
to
be
able
to
deploy
those
funds
to
then
you
know,
buy
out
infrastructure,
basically,
whether
it's
origin,
servers
or
cdns,
and
because
you
know
everything
we
do
is
is
essentially
enabled
by
crypto
payments,
we're
a
great
fit
for
those
sorts
of
organizations
who
want
to
function
autonomously
and
leverage
that
those
their
treasuries
basically
to
hire
out
infrastructure
on
the
decentralized
app
end.
Obviously,
nft
is
really
content
heavy.
You
know
nft
platforms,
really
content.
C
Yeah
and
to
add
to
that
as
well,
I
think
we've
been
doing
also
a
lot
of
research
into
who
is
running
the
points
of
presence,
and
you
know
what
the
experience
should
look
like,
because
you
know
it's
not
we're
not
looking
at
a
different
way
where
people
are
gonna,
build
server,
farms
and
that's
in
that
sense.
We're
really
looking
at
you
know
who
are
users
inside
of
urban
environments,
or
you
know
people
with
regular
laptops.
What
that
experience
is
going
to
look
like
for
them
to
run
those
nodes.
Are
they
going
to
trust?
A
Wilson
thanks
so
yeah
next
question
is
looking
forward.
What
are
the
biggest
question
marks
around
kind
of
where
this
is
going,
the
research
that
you've
that's
got
to
be
done?
What
keeps
you
up
at
night
yeah
for
how
this
progresses.
B
Well,
yeah
a
few
things,
but
I
as
as
toba
mentioned
the
longer
term
vision,
obviously
is
one
where
you
know
the
network
and
the
nodes
are
as
distributed
as
possible,
and
you
want
to
bring
it
as
close
as
possible
to
end
users
because,
what's
really
key
in
content,
delivery
like
physical
distance,
is
actually
kind
of
king
and
and
performance
in
this
space.
B
If
you
get
content
as
close
as
possible
to
your
end
users,
you
know
so
basically,
all
of
that
to
say
that
building
this
very
distributed
network
with
lots
of
nodes
is
going
to
require,
like
fine-tuned
and
very
carefully
considered
crypto
economic
incentives
to
incentivize
people
to
serve
content
on
the
network,
so
making
sure
that
they're
getting
paid
an
amount
and
whereby
they
would
basically
contribute
excess
resources
to
the
network
is
something
that
we're
actively
working
on
and
researching,
but
we're
also
looking
at
leveraging
potentially
like
non-economic
incentives.
B
Obviously,
this
is
a
bit
of
a
softer
topic,
but
if
you're,
a
user
who
wants
to
contribute
to
a
decentralized
application
that
you
really
care
about,
maybe
there's
incentives
already
there
for
you
to
contribute
your
your
access
storage
to
cash.
You
know
some
of
the
web
assets
for
the
applications
that
you
already
care
about
so
kind
of
balancing
those
two
incentives.
You
know
with
the
softer
incentives
with
the
crypto
economic
ones,
is
something
that
we're
looking
into
for
sure.
C
You
know
on
the
falcon
blockchain,
so
that's
definitely
something
where
we're
also
researching
actively
at
the
moment,
but
we're
confident
there
are
solutions
out
there
and
there's
other
groups
looking
you
know
on
the
actor's
side
to
improve
the
performance
and
the
user
experience
around
payment
and
state
channels
on
that
coin.
So
we're
definitely
actively
testing
these
and
looking
what
is
needed
to
to
get
to
that.
A
Awesome
yeah,
I
think
I
know
the
work
you're
leading
to
and
it's
very
exciting
for
these
other
things
happening
in
the
ecosystem.
So
final
question:
what
would
you
recommend
to
other
teams
who
are
looking
to
get
into
this
retrieval
market
space
yeah?
We
want
to
build
the
ecosystem.
We
want
more
teams
to
to
get
interested
to
get
involved.
There's
so
many
different
pro.
You
know
problems
we
can
solve.
So
what
would
you
recommend.
C
I
I
think
you
know
everyone
should
come
introduce
themselves
in
in
the
retrieval
market
channel
on
the
falcon
slag.
First,
first
off,
you
know
and
then
really
kind
of
look
at,
and
you
know
obviously
start
a
conversation
and
look
at
where
the
holes
are
because
there's
still
some
gaps,
and
I
think
it's
really
important
to
figure
out
what
people
are
not
working
on
and
just
kind
of
fill
those
gaps
so
that
we
don't
duplicate
the
work
and
so
yeah
really
communicating
on.
What's
interesting.
What
are
your
strengths
and
figure
out?
C
You
know,
where
is
your
place
in
the
ecosystem
and
that's
kind
of
what
we've
been
looking
at
and
collaborating
with
other
teams
and
figuring
out
how
we
can
all
you
know,
leverage
each
other's
building
blocks
and
then
you
know
help
each
other
when
you're
stuck
that's
also
pretty
useful
yeah.
B
For
sure
I
guess
for
for
context,
there
is
there
are
other
teams
working
on
the
retrieval
market
problem
and
we're
all
kind
of
tackling
different
aspects
of
it.
So
there's
a
yeah,
there's
a
great
community
feel
and
obviously
there's
some
outstanding
problems
with
all
these
teams
that
we
can
all
help
direct
direct
you
to
like
problems
that
need
solving.
A
Awesome
well
thanks.
So
much
guys,
it's
really
exciting
the
work
that
you're
doing
I'm
really
excited
to
see
where
it
heads.
I
can
also
see
some
people
coming
up
next
from
the
ecosystem
to
speak
so
yeah,
it's
just
the
ecosystem.
Around
retrieval
markets
is
really
exciting.
If
anyone
does
want
to
get
involved
head
to
retrieval.market,
it's
that
easy
to
remember
it's
just
you
know
put
it
into
your
url
bar
and
press
enter
and
from
there
you
should
be
able
to
find
anything.
You
need
to
get
involved
thanks
very
much.