►
From YouTube: Improve Data Durability with Reed-Solomon Encoding
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
Opportunities
for
data
encoding
in
file
coin
again,
that's
me:
rob
mcginnis,
there's
a
lot
more
past
work,
but
relevant
past
work
includes
etherstore
is
a
distributed.
Storage
solution
that
utilized
the
front
office,
workstations
and
organizations
to
essentially
provide
an
alternative
for
an
on-site,
nas
storage
or
a
secondary
target.
A
So
you
add
up
all
the
spare
storage
space,
all
the
spare
hard
drive
space
across
a
number
of
front
office,
workstations
and
looks,
and
actually
walks
and
talks
just
like
a
regular
drive,
except
it's
obviously
slightly
more
more
complicated
than
that
and
then,
which
is
really
the
the
project
that
requires
the
most
data
storage
right
now,
we're
using
roadside
cameras
to
detect
infrared
cameras.
A
It
works
day
and
night
to
detect
large
wildlife
on
the
roads
and
then
alert
drivers
in
real
time
using
you
know,
flashy
lights
reason
being
I
moved
from
new
york
out
to
wyoming
about
two
years
ago
and
yeah.
We
are
blessed
we're
surrounded
by
very
large,
very
cool
animals,
but
we
see
some
moose
hit
on
the
side
of
the
road.
A
You
know
a
couple
times
a
year
that
and
far
more
deer
than
that
and
trying
to
come
up
with
a
solution
that
is
is
portable
and
that
is
is
helpful
for
both
drivers
and
and
wildlife
was
something
I
was.
I
was
happy
to
be
able
to
to
work
on
here.
So
you'll
see
the
the
rig
for
that
in
just
a
second.
A
This
is
less
interesting.
I
think
this
is
ether
store,
three
tiers
of
local,
an
in
organization
and
then
web-based
version
backup.
So
it's
like
read
through
right
through
caching,
with
all
sorts
of
things
all
chopped
up
in
in
interesting
ways,
but
the
blue
star
stuff
is
what
I
find
most
interesting
right
now,
moose.
Obviously
you
can
tell
we're
using
a
flir.
We
got
a
floor
camera
and
a
couple
of
other
brands
that
we're
we're
testing
out,
and
this
is
the
trailer
that
has
just
arrived.
A
A
bunch
of
community
organizations
got
together
and
we
gathered
some
funding
to
purchase
a
solar
trailer
and
you
can
see
the
really
tall
mast
there
that
you
mount
the
camera
to,
and
information
comes
down
to
my
little
neural
network.
In
here
we
snap
a
single
frame
takes
about
a
second
and
a
half
to
process
each
frame
right
now,
it'd
be
nice
if
it
were
faster
but
that's
to
be
worked
on,
and
then
you
can
see
these
yellow
lights
here
that
aren't
hooked
up
to
anything.
A
Yet
that
will
be
alerting
drivers
that
there
is
presently
a
a
large
animal
in
the
middle
of
the
road
or
on
the
side
of
the
road,
and
they
ought
to
slow
down.
But
it's
not
just
one
of
those
dumb
flashy
animal.
Beware
lights!
It's
a
right
now,
kind
of
kind
of
thing.
So
how
is
this
relevant
to
what
we're
doing
with
falcoin?
A
Well,
there
is
a
ton
of
data
that
we
need
to
store
for
for
moostar
more
than
more
data
than
I
have
the
local
storage
for
here,
and
I've
got
just
a
ton
of
ton
of
stuff
out
in
the
garage
and
of
course
you
know,
that's
not
resilient.
It's
local
storage,
I'm
right
at
the
flood
plain
here.
So
who
knows
bad
things
can
happen,
and
we
want
to
make
sure
that
all
the
data
that
we
collect
is
stored
in
perpetuity
and
available.
Ideally,
for
anybody
to
access
worldwide.
So
how
do
we
do
that?
A
Well,
we
use
foulcoin,
obviously,
but
there
are
some
some
challenges
with
simple
replication.
So
say
you
want
to
store
one.
A
large
data
set,
you
chop
it
into
say
40
slices,
or
so
let's
say
you
do
two
times
replication.
A
If
you
lose
the
wrong
two
copies
of
that
file,
you're
you're
out
of
luck,
you're,
basically
you
don't.
You
do
no
longer
have
the
ability
to
recover
that
file.
That's
very
frustrating,
and
you
want
things
to
last
for
a
very
long
time.
You
just
need
to
look
at
this.
We're
we're
storing
stuff
on
a
bunch
of
different
people's
systems
all
over
the
world
and
the
annualized
disk
failure
rate
adds
up
to
about
50
within
five
years.
All
of
these
columns
add
up
right.
A
That
means
half
of
all
hard
disks
fail
within
five
years,
which
means
that
if
we're
using,
you
know
non-professional
individuals
to
store
stuff
which
is
highly
it's
likely
now,
things
will
change,
there'll,
be
better
best
practices
and
stuff
shared
out,
but
it's
very
real.
It's
a
very
real
possibility
that,
within
five
years,
two
of
those
copies
of
data
that
will
be
required
for
your
entire
file
to
be
accessible
may
no
longer
be
be
accessible
right.
A
A
A
I've
got
my
contact
details
at
the
end
of
the
presentation
you
can
get
get
in
touch
with
me
on
slack.
If
you
want
to
learn
anymore.
A
A
A
First
of
all,
obviously
recovery,
but
imagine
one
of
those
replicas
is
out
on
you
know
the
slowest
local
internet
connection
you
can
find
my
download
speed's
great
here
in
wyoming,
but
our
uploads
are
tapped
at
you
know
way
way
way
lower
than
the
normal.
You
know.
Ten
percent,
you
know
I'm
just
like
down
at
one
percent,
so
coming
down
is
easy,
but
going
back
up
is
is
challenging
once
you're
done.
Reading
40
of
those
parts
you're
good
to
go.
Okay,
not
free
though
so
there
are
upfront
costs.
A
So
the
time
you've
got
to
pre-process
the
data
and
there's
space.
You
have
to
have
twice
the
amount
of
space
on
your
available
beforehand.
If
you
want
to
encode
things
this
way,
if
you're
doing
just
replication,
you
only
need
the
exact
size
of
the
file.
If
you're
going
to
pre-encode
it,
you
need
double
the
size,
you
don't
always
have
to
do
a
double.
It
could
be
a
little
bit
less.
But
generally
speaking
for
this,
for
this
comparison
with
the
2x
replication,
you
need
twice
as
much
space,
so
that
could
be
challenging.
A
Finally,
upon
retrieval,
the
client
needs
to
post
process
the
data
a
little
bit,
and
this
can
be,
I
mean
a
little
bit
challenging
sorry
not
challenging
easy
to
do,
but
it
takes
a
little
bit
of
time
and
of
the
order
of
minutes,
though,
for
you
know
a
few
gigabytes
and
things
like
that,
it's
basically
straight
line.
So
it's
linear
to
the
size
of
the
file-
and
you
know
so
again,
nothing's
free.
There
are
pros
and
cons
to
to
everything
but
yeah.
A
That's
how
we
are
planning
to
store
the
the
moose
star
data,
and
we
plan
to
you
know,
get
some
of
this
code
out
there,
so
that
other
people
can
can
take
a
look
at
it
and
maybe
utilize
some
of
these
pre
and
post
processing
approaches
in
order
to
make
their
data
more
resilient
in
the
network
feel
free
to
contact
me
on
slack
at
rob.