All language subtitles for [SubtitleTools.com] RMAN Incremental Backups - Learning Oracle 12c [Video]

In this lesson, we're going to
talk about incremental backups

and RMAN does support
incremental backups.

And incremental backups are very
advantageous to an Oracle DPA

because they allow you to
backup only the changes that

have occurred since
the previous backup.

Now when we use
incremental backups,

we want to remember two things.

First of all, this really
requires a strategy.

We need to think about things
like when our backups occur,

what our rates of
change are in data.

That is to say, how busy is our
database during different days,

and we need to kind
of roll all of that

into a strategy for how we can
best use incremental backups.

And Oracle gives us a
couple of different ways

to run incremental
backups, and so that'll

go into our strategy, as well.

The next thing we need to
remember that's very important

and is not enabled by
default is something

called block change tracking.

So we said that an
incremental backup would only

backup the blocks
that have changed

in the database since
the previous backup.

However, even though
that is the case--

so we would expect that
a incremental backup

would take significantly less
time than a full backup--

but that actually is
not the case by default

because even though
it only writes out

the blocks that have
changed, it still

must read every block to
see if it has changed.

So when we enable what's
called block change tracking,

Oracle actually keeps a
list of changed blocks.

So as blocks change,
they're written

to essentially a changed list.

And then when an
incremental backup

occurs, instead of reading
every block in the database

in order to determine whether or
not it needs to write that out,

it simply looks at the
list, collects the blocks

that it needs, and
writes those out.

So the results can be
significantly faster

for an incremental backup
if block change tracking

is turned on.

So we have two different
types of incremental backups

and really, these exist to
give us the flexibility.

Most people will use the type
that's called differential,

but there is also a
kind called cumulative.

So let's look at a differential
incremental backup.

So the rule here is
that a differential

is going to backup
all the blocks that

have changed since the
last level n or lower.

So when we do an
incremental backup,

the first thing we start
with is a Level 0 backup.

So a Level 0 is going to
backup all of the database.

So it's a complete and
full database backup,

but we must back
it up as a Level 0

if we want it to
be the foundation

of an incremental strategy.

So when we do the first
incremental, if you will,

we'll call that a Level 1.

So we'll specify a Level
0 for the first backup,

and let's say that
completes on Monday.

The next one would be
a Level 1, and so it'll

backup all the blocks that have
changed since the last Level 1

or lower.

In this case, it
would be Level 0.

Then we do a Level 2.

So that will backup
all the blocks

that have changed since
the last Level 2 or lower,

and that happens
to be a Level 1.

But if we were to do after
a Level 2 to do a Level 1,

that will backup
all the blocks that

have changed since the
last Level 1 or lower.

So it would go all the
way back to Tuesday

in order to get those blocks.

Now you can certainly just
do Level 0s with Level 1s,

but the idea of having a
Level 2 in there as well

gives you the flexibility
to have some days where you

skip or overlap data blocks.

The other type is a
cumulative incremental backup

and that's going to backup
the last level n-1 or lower.

So again, a Level
0 is our foundation

for our incremental
strategy, and then we

do a Level 1, which
is going to be

the last Level n the minus
1, which is zero or lower.

Than we do another
Level 1, which

will do all the blocks that have
changed since the last Level

1 minus 1, which means
it goes all the way back

to the Level 0.

And again, another
Level 1 would again

go back to all the
blocks that have

changed since the last Level 0.

So this is why it's
called a cumulative backup

because each night
after the Level 0,

assuming you continue
to do Level 1s,

you'll backup all the
blocks that have changed.

So that's kind of how
incremental backups break down

in their theory, anyway.

So let's take a look at
doing a couple of operations

like block change tracking
and incremental backups.

We'll go into SQL plus and
we'll do a select status

from v dollar block
change tracking.

So currently that
says that is disabled.

So we'll do alter database
enable block change tracking.

And now if we were to select
the status from v dollar block

change tracking, we
could even do select star

from v dollar block
change tracking

and it would give
us the location

of the file where it's writing
the list of changed blocks.

And lets perform an
incremental backup.

We do connect target
to the database

after logging into RMAN and we
do backup incremental Level 0.

Again, we must do a
Level 0 as the foundation

for our incremental
strategy database.

We hit Enter.

And this will run for a bit.

Now the level 0 is complete.

So just to simulate the passage
of a little bit of work,

let's switch a log file--

alter system switch logfile.

So it's as if some
changes have occurred.

And then let's do our
incremental backup--

backup incremental
level 1 database.

So now we're assuming that
we're doing a differential.

If we want to do a
cumulative incremental,

we need to use the
cumulative keyword.

So this backup runs
considerably faster.

And that's really the benefit of
an incremental backup, the idea

that subsequent backups can
back up only the changed blocks.

And that's all that's required.