►
From YouTube: New User Training: 07 Data Ecosystem Overview
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
A
Talks
are
going
to
be
all
about
how
to
store
and
access
data
in
various
ways
at
nurse
and
then
we're
gonna
have
a
couple
of
talks
about
sharing
data
either
through
data
transfer
or
through
this
thing
called
the
spin
early
user
program
is
going
to
talk
about
science
gateways
a
little
bit
and
then
a
little
bit
about
analytics,
because
Python
and
Jupiter
are
pretty
important
part
of
the
data
ecosystem
and
then
at
the
end,
we'll
talk
about
workflows,
kind
of
engine
shifter,
okay.
So
all
these
talks
are
just
20
minutes,
I.
A
A
So
part
of
our
job
is
to
know
about
various
technologies
and
figure
out
how
to
deploy
them
for
users,
and
so
that
means
that
we
do
a
lot
of
adaptation
of
these
technologies
so
that
they
work
in
our
you
know:
crazy,
create
ecosystem.
So
there's
we
kind
of
draw
these
kind
of
semi
arbitrary
boxes
around.
A
What
these
different
pieces
are,
but
they're
kind
of
split
up
into
data
transfer
and
access,
so
you'll
see
like
an
X
because
that's
accessing
stuff
at
the
center
kind
of
next
to
Globus,
managing
workflows,
data
management
and
then
there's
pieces
in
there
that
are
also
kind
of
analytic,
see
pieces
data
analytics,
which
is
kind
of
this
really
broad
topic
about
just
doing
stuff,
with
data
I,
guess
and
then
data
visualizations.
So
today
we're
not
going
to
talk
too
much
about
day,
visualization
or
workflows.
A
Quite
so
much
but
I
hope
that
wasn't
like
I
said
something
really
silly
or
something
I
probably
did
but
but
anyway,
okay.
So
so
anyway,
you
know,
there's
a
whole
bunch
of
logos
here
and
you
know,
if
you
see
something
that
resonates
with
you,
then
you
should
feel
good.
And
if
there's
something
that's
missing
here,
then
you
should
feel
bad,
I
guess,
but
where
you
should
talk
to
us
about,
you
know
why?
Don't
we
have
that
or
maybe
we
do-
and
we
just
didn't
have
it
here.
A
A
The
data
partition,
which
is
the
Haswell
partition
and
then
the
HPC
partition,
which
is
the
the
kml
partition,
doesn't
mean
you
can't
do
data
analytics
on
kano,
but
it
means
that
you
kind
of
have
to
do
a
bit
more
to
optimize
your
code
to
get
the
best
performance
that
you
can
out
of
it
on
KL,
but
Haswell
is,
is
sort
of
the
data
partition.
Most
most
people
doing
data
analytics
are
running
their
stuff
there
on
the
Haswell
partition.
A
Cory
is
also
special
because
it
has
all
of
these
kind
of
nice
features
that
are
really
friendly
to
data
analytics
and
data
processing
users,
people
who
are
doing
anything
with
data
like
experiment,
you
know,
processing
data
from
experiments
or
looking
at
simulation,
outputs
and
doing
something
with
that.
But
there
are
a
lot
of
things
that
are
kind
of
more
forgiving.
I
would
say
to
those
kind
of
users
we
have
like
twice
as
many
login
nodes.
Not
not
all
of
them
are
available
for
users,
but
we
use
some
of
those
other
login
notes.
A
We've
repurposed
them
to
do
other
things.
We
have
a
couple
of
notes
for
running
big
memory,
jobs
which
are
kind
of
important
for
data
analytics
special
nodes
that
are
set
aside
for
pipelines
and
and
workflows.
Cory
is
the
first
system
where
we've
been
running
containerized
environments
in
production.
So
that's
going
to
be
the
last
talk
today.
We
also
have
ways
of
streaming
data
to
and
from
the
compute
nodes
configurable
access
over
the
network
to
the
compute
nodes
from
outside,
which
is
kind
of
new.
We
also
have
this
burst.
A
Browser,
there's
gonna,
be
a
talk
about
the
burst
buffer
and
how
to
use
that
for
storage
and
I/o
later
today,
and
then
we
have
this
really
flexible
workload
manager
in
slurm.
We
switch
to
slow
him
a
few
years
ago
and
it's
given
us
the
ability
to
field
some
kind
of
more
experimental
queues
like
the
serial
queue,
the
shared
queue
transfer
queue,
real-time
queues
and
then
the
interactive
queues.
Have
you
ever
used
the
interactive
queue
on
quarry?
If
you
have
you
haven't
your
life
is
about
to
change
check
it
out.
It's
really
awesome.
A
A
Immediate
access
to
compute
resources
on
Corinne,
so
there
are
some.
There
are
some
guidelines
around
how
many
people
can
use
it
from
from
a
project
and
how
many
nodes
you
can
have,
but
it's
a
life
changing
experience,
I'm,
not
exaggerating
so
a
few
of
the
things
that
that
I
wanted
to
touch
on
that
we're
not
really
going
to
have
time
to
to
go
over
today
in
a
full
talk.
I
think
I
have
like
five
of
these
things.
One
of
them
is,
as
I
mentioned.
A
A
This
is
on-demand
access
to
a
pool
of
nodes
that
can
respond
basically
immediately
as
soon
as
you
want
to
run
a
job
you
you're
able
to
to
do
that
and
it's
you
have
to
apply
to
get
into
it,
and
it's
mainly
it's
it's
for
experiments
that
are
running
at
the
same
time
as
they
need
to
do
the
compute,
but
they
don't
have
the
computer
sources
on-site
to
do
it.
So
it's
shipping
data
over
and
then
doing
something
with
it
and
shipping.
The
answer
back.
A
Noot
noot,
if
you
ever
wanted
an
API
to
talk
to
a
high-performance
computing
center,
you've
got
it
on
this
slide.
You
also
have
a
slide
where
every
word
in
the
title
is
a
acronym
I
guess,
which
is
kind
of
cool,
but
Newt
is
a
is
a
is
a
programmatic
interface
to
to
be
able
to
enable
you
to
script
things
to
script,
doing
things
at
Newark
in
a
nice
and
easy
way.
A
So
if
you,
if
you
have
a
web
browser
or
you
want
to
write
like
a
Python
request,
script
or
something
like
that
to
automate
some
kind
of
workflow,
you
can
do
that
with
new
and
there's
a
there's.
A
little
link
on
all
of
these
slides
I.
Think
that
that
you
can
follow
up
on
science
gateways.
Science
gateways
are
Hardware
where
we
run
user
written
web
applications
that
enable
access
to
users
outside
of
nurse
to
data.
That's
at
nurse
can
these
can
be
completely
public
or
maybe
in
your
own
collaboration,
or
something
like
that.
A
A
We
work
on
optimizing
those
tools
so
that
they
they
work
well
on
our
HPC
hardware
and
we,
you
know
where
we
hope
to
be
in
a
conversation
with
our
users,
about
deep
learning
and
machine
learning,
to
encourage
the
use
of
cutting-edge
methods
and
and
finding
projects
to
to
collaborate
with
so
and
then.
Last
but
not
least,
is
data
visualization
kind
of
the
most
popular
simulation.
A
You
know
visualization
tools
that
are
out
there.
We
support
visited
pair
of
you
and
the
in-car
graphics
library,
but
we
also
provide
support,
for
you
know,
visualization
tools
that
aren't
you
know
quite
so
heavy
weight,
or
maybe
our
you
know,
Python
libraries
or
something
like
that
or
our
libraries
or,
if
you
really
like
new
plot
you,
you
absolutely
can
use
new
plot
yt
root
and
then
specialized
specialized
tools
that
are
kind
of
domain-specific.
A
Like
the
visual
molecular
dynamics,
all
of
those
things
are
are
available
and
if
there's
something
here
that
you
need,
we
can
look
into
installing
that
for
you
and
then
at
the
very
end
I,
you
know.
I
can't
be
emphasized
enough
that
if
you're
doing
visit
nurse,
you
should
really
be
doing
it
over
an
X
okay.
So
that
was
like
just
a
quick
set
of
slides
of
tops.