All language subtitles for [SubtitleTools.com] Real Application Clusters - Learning Oracle 12c [Video]

In this lesson,
we want to examine

the concept of Real
Application Clusters, or RAC,

in an Oracle database.

RAC is a hot topic today
in the Oracle World

because RAC is Oracle's
clustering solution.

And the concept of
clustering involves

taking multiple
computers and having them

act in concert together as one.

RAC is a pay-extra option
for the Oracle database.

So you can have an Oracle
database Enterprise license

and yet not be licensed to
run RAC, or Real Application

Clusters.

It is a pay-extra option.

It is rather expensive.

But that is because
it affords a company

a lot of positive
possibilities when

it comes to their database.

First of all, it's a high
availability solution.

Because in RAC the
software allows machines

to work together, if any one
machine is taken down or fails

or has to even be patched,
the other machines

continue the work.

So it's high availability
for a company's applications.

The next is that it's
a scalability solution.

So if a company
wants to grow, they

have the option in Real
Application Clusters

of simply adding another
machine to the cluster.

So let's use an example for
why a company would want to use

Real Application Clusters.

Let's say that we
have a small company.

We've started a company, and it
does online ordering for books.

Initially, our customer
base, let's say,

is about 100 customers.

And we have an employee
base of about five people.

In that situation,
we would probably

need a fairly small
database server.

We need it to store things
like basically our ordering,

the different orders that
are coming into our company,

and also maybe
employee information.

So we settle on a
small to medium sized

commercial hardware server,
put the Oracle database on it,

and the company begins to grow.

And eventually, the customer
base grows to about 10,000.

And our small computer that we
had when our company started

is no longer viable.

So our only option--

without RAC-- is to get
rid of the old server,

buy a new server, maybe a 4U
higher end commercial server

and run our Oracle
database on that.

Well, then, our company
grows even more.

And let's say our customer
base is about 250,000,

a quarter million.

Now our second server that
we had is inadequate as well.

And so now, we have to go
out and buy another server.

So without the
option of RAC, we can

spend quite a bit of
money trying to keep pace

with the growth of our company.

But with RAC, we can simply
have a scalability solution

that allows us to scale
our database system

upwards by simply adding
machines to the cluster.

So to differentiate RAC from
the more common single instance

architecture, let's
look a little bit

about how that works.

So in a single
instance architecture,

the database operates
entirely on a single machine.

So this is a machine.

And it's composed
of the instance,

which is the background
processes and memory

caches, and the disk.

So the processes
run on this server.

The caches are stored in
RAM that's on this server.

And the disk is on
this server as well.

So this is a single instance
architecture-- very common.

Works fine for
many applications.

But it doesn't operate
as a scalability solution

or a high availability solution.

If this machine's power system,
for instance, were to go out,

your data is inaccessible--

processes and caches
being the instance,

and the disk being the database.

But RAC is a multi-instance
architecture.

So in RAC, we may have
three physical machines.

They are three different
physical machines.

They each have their own
instance, their processes

and caches.

But the disk, or the database,
portion of the architecture

is separate.

So we might actually have
a separate disk unit.

And the most important
thing about the disk

is that it's shared
between all three machines.

So the disk has to be
available to every one

of these separate machines.

And shared disks can be done
a number of different ways,

like storage area
networks or network

attached storage, a NAS.

And it also has to have a
network-aware file system,

or a global file system--

something like NFS, the
Network File System, or ASM

in Oracle, Automatic
Storage Management.

So there is a separation between
the instance and the database.

So each physical machine will
be connected to that storage.

But that alone isn't enough.

Because in RAC, one of the
advances that came out in 9i

was the ability to be able to
let each one of these servers

talk to another.

The instances are shared.

They're separate,
but they're also

shareable through a high-speed
network interconnect,

so something like
a high-speed switch

that will connect all three
of those physical machines

together.

Thus, if any one of
the machines is lost,

processing continues
on the other two.

If two machines are lost,
processing continues.

So this is important
for high availability.

But it's also important with
something like patching.

So if you need to do an
operating system patch on one

of the physical machines, it
needs to be generally taken

down or rebooted.

With RAC, we can do
that and still continue

database processing.

So each one of these machines
has a separate instance that

connects to shared storage.