►
From YouTube: Nebula Graph v3.0.0 Preview
Description
Nebula Graph v3.0.0 will be released in the near future.
This preview will brief you on the highlights for the new features, optimizations in advance such as multi-pattern, cypher parameter support, etc.
You can find the slides here: https://siwei.io/talks/nebula-3-preview/
A
Art,
hello,
everyone:
this
is
weiku,
I'm
the
developer
advocate
of
vsoft
the
team
behind
nablograph.
I'm
super
excited
to
have
this
time,
have
the
chance
to
share
with
you
some
brief
information
of
the
pre
preview
of
the
new
version
release
of
nabla
graph,
which
is
3.0.
Finally,
and
please
be
noted
that
this
slide
is
online,
so
you
can
reach
out
to
this.
A
A
A
You
can
pass
the
parameter
and
your
template
query
together
directly
to
the
nabla
graph
server
and
then
there's
the
birth
side
view
of
the
data
you
can
now
match
return
the
data
without
a
necessarily
created
index.
If
you
just
want
to
sample
some
of
the
data,
so
you
give
a
limit.
So
it's
now
possible
the
data
will
be
scanned
and
pushed
down
to
storage
site
without
requirements
of
index.
A
If
you
specific
the
limit
clause-
and
now
we
also
support
the
bare
vertex,
which
is
the
vertex
without
any
text
attaches
to
it.
The
impervious
version
is
not
the
case,
so
I
will
give
a
introduction
later
and
the
backup
restore
finally
is
released
now
and
we
redesigned
the
previous
day
version
with
the
agent-based
design
to
enable
the
decouple
of
the
ssh.
So
it's
friendly
to
the
cloud
native
deployment
now
some
others
that
are
not
listed
here,
some
of
them
worth
mentioning
is
we
now
support
describe
user.
A
A
The
first
one
is
the
mod
pattern
match
before
we
only
manage
to
support
one
liner
of
the
patent
party
match,
and
now
we
can.
We
finally
support
the
actual
lines
of
either
it's
optional
or
non
non
optional
match,
so
it
will
enhance
the
expression
ability
and
to
enable
some
certain
use
cases.
So
this
is
the
example
that
we
add
a
couple
of
actual
matches
towards
the
one
line
of
this
match
pattern.
A
So
it's
supported
the
next
one
is
the
parameter
so
like
this?
You
can
assemble
your
map
or
dictionary
of
parameters
and
pass
this
template
of
query
like
this
and
those
those
are
the
the
slot
of
the
parameters
and
the
query
them
send
them
to
server
side
is
supported.
You
can
refer
to
this
example,
or
here
this
is
from
the
the
client,
the
nablu
nebula
python.
So
this
is
the
example
and
this
kind
of
convenient
and
also
more
secure
in
certain
cases.
A
A
This
is
possible
and
you
can
check
from
this
explain
of
this
query-
that
underlying
is
actually
scan
of
the
vertex
data
instead
of
depending
on
the
index
like
before
so
and
it's
actually
underlying
push
down
to
the
storage
side.
So
it's
not
that
that
expensive
anymore,
so
this
will
be
quite
convenient
for
the
sample.
If
you
just
want
to
check
some
of
the
data,
we
can
do
that
and
without
introducing
the
index.
A
A
So
this
is
a
bare
vertex
and
you
can
fetch
it
without
any
properties.
Right
and
another
case
that
I
didn't
put
in
this
slice
is
you
can
choose
to
delete
a
tag
from
a
vertex
and
previously
in
inversion
2?
If
you
do
that
to
a
if
a
vertex
with
only
one
tag
after
you
delete
the
tag,
nebula
will
treat
this
vertex
to
be
deleted.
A
You
cannot,
you
cannot
query
after
you
just
delete
the
tag
of
that
vertex
and
it
will
be
deleted,
wiped
during
a
next
compaction
actually,
but
this
is
a
change
in
in
version
three,
so
this
is
supported.
So
after
you
do
that
the
vertex
will
become
a
better
one,
so
it
will
not
be
cleaned
anymore.
So
it
will
be
helpful
for
certain
use
cases
that
you
only
care
of
the
structure
or
topology
of
the
graph
and
though
there
is
no
vertex
or
properties
in
this
graph.
So
you
can
use
this.
A
But
you
already
know
it's
based
on
ssh,
which
is
not
friendly
for
the
cloud
native
environment,
and
now
we
redesigned
it
by
introducing
a
nebula
agent,
and
it
was
as
if
I
recall,
creates,
was
released.
The
first
version
yesterday
and
with
the
help
of
the
agent
we
decoupled
ssh
and
the
the
process
and
the
stability
is,
is
more
it's
better
than
before,
but
we
we
keep
the
user
space
unchanged.
A
A
So
we
think
consider
this
is
a
bad
design
and
it's
not
it's
not
possible
from
this
release
and
we
unified
all
the
format
of
the
yield
clause
on
all
the
ngql,
and
we
also
introduced
the
kiwi
separation,
which
is
a
kind
of
performance,
more
comfort
performance
in
certain
cases
that
I
will
introduce
later
and
the
balance
leader.
Now
you
can
choose
to
balance
leader
in
specific
space
rather
than
the
whole
cluster,
and
there
are
some
other
improvements.
A
That's
not
import,
impacting
our
user
space
like
an
index
execution
plan
is
refactors
and
we
we
revisit
and
refine
our
rafter
implementation
and
and
our
elastic
search
client
supports
tls.
Now
this
is
for
the
food
full
text
search
and
we
read
that
redesigned
the
session
ex
expert
timer,
so
by
default
previously
by
default,
is
zero,
and
now
this
part
is
changed
on
the
geography
part.
A
We
introduced
this
feature
in
last
version
and
there
is
a
s2
function,
and
now
this
s2
index
supports
parameters
in
this
release,
and
this
one
is
the
the
properties.
This
is
a
function
or
syntax
sugar.
In
the
ngo
query.
Previously
we
only
support
it
to
be
in
in
return
clause,
but
now
it's
supported
in
the
where
class
that
I
know
many
of
the
user
is
eager
to
use
this
class,
it's
quite
sweet
and
also
on
the
show
sessions
classes.
A
There's
a
break
change
that
when
there
we
added
the
show
local
session,
which
means
you
only
query
the
ongoing
sessions
on
this
connected
graph
d
instance
and
we
remove
the
show
all
sessions
and
make
the
equivalent
query
in
show
session
and
also
in
server
side.
We
in
this
release,
we
introduced
the
log
rotation
you
have,
you
don't
have
to
use
any
actional
utility
to
implement
the
log
rotation
anymore.
It's
done
from
the
nablograph
server
side
for
other
enhancements,
either
is
listed
here
or
not.
A
I
I'm
not
going
to
dive
walk
through
all
of
them,
so
then
we
will,
but
you
can
check
from
this
release
or
from
github
for
the
matrix.
As
I
mentioned
there,
we
introduced
really
a
lot
in
this
release
and
we
will
for
sure
introduce
more
in
the
future
and
it's
also
integrated
with
our
dashboard
and
exporter.
A
A
So
this
is
helpful.
This
is
proven
to
be
helpful
for
the
case.
If
especially,
your
value
is
huge
and
it's
a
paper.
It's
originally
raised
by
a
paper
named
waste
key
and
actually
roxby
already
implement
this
paper,
and
they
call
it
blob
db.
So
this
is
a
link
of
their
blog,
so
we,
this
is
a.
This-
is
a
function
that
introduced
by
two
of
our
maintainers
in
community
during
the
nebular
hackathon
that
I
will
give
more
information
later
so
during
this
their
implementation
with
their
tests.
A
They
introduced
four
of
the
configurations
and
the
first
one
is
the
switch
of
the
skip
separation
and
seven.
Second,
one
is
the
threshold
like
you,
you
would
like
to
specify,
after
reaching
this
size
of
the
thread
hold
in
in
value
size,
the
value
will
be
persistent
separately,
so,
if
it's
smaller
than
that,
it's
still
together
with
the
key
and
this.
This
is
the
comparison,
compression
method
of
the
value
and
you
can
choose
either
no
or
any
other
compassion
methods.
So
this
is
the
flag
to
determine
why
whether
we
will
do
the
garbage
collection.
A
A
Regarding
upgrade
a
as
we
know
that
when
it
comes
to
the
major
the
from
one
to
two
from
two
to
three,
the
major
version
change,
there
are
some
underlying
data
structure
refactor
to
enable
some,
you
know
some
gains,
so
you
have
to
do
the
data
migration
and
this
is
implemented
by
the
db
upgrader.
This
is
the
binary
in
in
the
package,
so
I'm
referring
to
the
binary
of
this
operator
here.
This
is
the
comment
section
that
we
are
taking
care
of
them
as
before,
as
we
we
were
doing
from
one
to
two.
A
So
it's
also
supported.
So
if
you
are
running
a
2.6
cluster,
so
please
refer
to
the
documentation
to
do
the
upgrade.
It's
not
that
hard,
and
then
we
will
have
some
introduction
of
the
incapable
changes.
Actually
most
of
them
was
already
revealed
in
previously
sized
so
yield
part.
We
can.
We
don't
support
the
omit
of
the
yield
anymore.
You
have
to
specify
all
the
fields
you
would
like
to
return.
A
A
We
actually
changed
the
mechanism
of
the
the
service
discovery.
So
in
this
version
we
have
to
explicitly
add
add
all
the
storage
hosts
after
you
bootstrap
a
new
cluster,
otherwise
even
it's
registered
in
the
method
diamond,
but
it's
not
considered
to
be
activated
to
be
usable.
So
this
is
a
a
new
design.
A
A
Another
thing
is
that
we
redesigned
the
the
data
balance
thing
and
in
this
release
the
balanced
data
is
removed.
So
that's
ex
actually
a
break
change
and
please
refer
more
information
on
documentation
and
if
you
have
any
questions,
just
come
to
our
community
and
reach
out
to
us
in
the
forum
or
slack.
A
A
The
vertex
size
based
on
multiple
factors
like
tag
property
and
degree
and
and
the
edge
color
can
be,
can
vary
based
on
the
the
source,
the
the
source
vertex
tag
and
there
are
more
layouts
supported
like
grade
and
mind
graph,
etc.
There
are
actually
more,
please
check,
really
related
demos,
and
there
are
more
some
more
exciting
features
introduced
in
in
ex
explorer.
Sorry,
the
first
one
is
3d.
A
So
if
we
recall
correctly,
this
is
the
first
time
that
a
a
graph
database
can
have
the
ability
to
explore
the
graph
data
in
3d
and
also
there
is
a
function
called
a
view
snapshot.
So
you
can
persistent
your
explore
the
data
in
the
in
in
the
browser
and
like
even
the
session
expired.
It
still
can
be
recoverable
and
reusable,
and
we
also
bring
the
console
from
the
studio
to
the
explorer
in
this
version.
A
A
So
it's
a
it's
not
not
just
the
dashboard
actually,
and
there
is
an
open
source
version
and
the
enterprise
version.
The
open
source
version
is
more
on
the
matrix
and
the
user
control
and
we
introduced
the
alarm
system,
the
task
management
system
and
the
more
metrics
that
I
reviewed
before
right
in
dashboard
of
this
release,
which
is
quite
exciting
and
for
the
naval
algorithm,
our
non-labor
algorithm
is
a
project
to
enable
the
many
graph
algorithms
running
on
top
of
nabla
graph,
and
we
introduce
a
couple
of
new
algorithms
in
this
release.
A
A
We
support
the
geography
sst
for
exchange
and
also
for
importer.
We
support
the
geography
data
source
in
this
release
and
for
the
clients
update
in
this
release
cycle.
We
have
three
more
clients
added
supported
by
with
the
help
of
our
community
contributors,
we
have
the
node.js
and
dotnet
and
php
so
finally
have
fun
with
that
and
on
the
operator
the
cloud
side,
the
the
kubernetes
operator,
the
the
team,
the
community
is
still
working
on
the
br
support.
A
It's
not
finished,
but
hopefully
we'll
be
finished
in
coming
and
please
watch
the
project
if
you
are
interested
in
so
we
are
so,
let's
recap
on
the
features
we
in
3.0.
We
mainly
introduced
significantly
more
support
on
the
open
server
part
on
mount
pattern
and
and
param
parameter
and
also
in
in
match.
Now
we
support
data
of
the
birth
eye
view
without
defining
introducing
index,
so
you
can
sample
them
by
limit
return.
A
We
support,
bear
vertex
now
and
we
bring
the
backup,
restore
redesigned
and
finally
released
for
enhancement.
A
lot
of
metrics
were
introduced
in
this
release
and
we
support
the
kv
separation
and
the
balance
of
detail.
Balance
of
of
leader
is
now
space
specified,
which
is
more
convenient
for
a
larger
cluster
with
multiple
spaces
we
refined
our
rough
implementation
and
the
execution
plan
of
the
index.
We
support
the
tls
in
the
elasticsearch
and
the
log
rotation
is
part
in
the
server
side.
A
A
A
A
Also,
it's
worth
mentioning
that
we,
we
finished
our
hacksaw
of
nebula
community
a
month
before,
and
there
are
a
lot
of
exciting
projects
contributed
by
different
teams
and,
for
example,
the
kv
separation
is
one
of
them
and
please
check
out
the
website
and
we
will
have
the
2022
hack
song
in
august,
so
stay
tuned.
If
you
are
interested
you
can
you
can
create
your
ideas
during
the
shower.
From
now
on.