Hi. Welcome to this lesson on Autonomous Database, Level 200 course.

My name is Bal
Sharma. I'm part of Oracle Cloud Infrastructure product management team.

Let me take a pause here so that you can go through our safe harbor statement.

So objectives for this lesson is going to be around these topics. So first one is,
we will focus on autonomous database in CI as well as several deployment
considerations. We'll also look at Database Cloud Service comparison, because we
have several database services in our data management portfolio. And we'll see how
they differentiate and their use cases.

We will also talk about architecture and best practices for our Autonomous Database
Dedicated. We'll try to understand the client connections for Autonomous
Transaction Processing Database Dedicated.

We'll also talk about the security options, which is available in Autonomous
Transaction Processing database. I will discuss about high availability concepts,
which is applicable to this Autonomous Transaction Processing.

We'll talk about how we can roll out software updates and patching in Autonomous
Database, as well as various developer tool and productivity, which is available on
autonomous database.

We will talk ways to migrate to autonomous database as well as various management

and monitoring options which are available for autonomous databases.

So let's start with the part one, where we will cover autonomous databases in OCI
as well as various deployment considerations.

So when it comes to autonomous database in OCI, we have basically two offerings.

One is serverless as well as Dedicated. And this is a family of products where each
member of the family is optimized by the workload.

We have data warehouse, which is optimized for analytic workload, like data
warehouse, data marts, or even a data lake. ATP, on the other hand, is optimized
for everything else. So it could be transaction processing or mixed workload
environment, and that makes a great platform for any new application deployment.

By autonomous, we mean everything is automatic in terms of provisioning, scaling up

and scaling out the tuning within these databases. When it comes to security and
patching, we have a lot of automation there as well as fault tolerance. Because
these are the characteristics of an autonomous database.

We use machine learning plus most proven technologies such as Exadata, Real
Application Clusters, Data Guard, Database Vault, all sorts of parallel SQLs or in-
memory multi tenant. So these are the enabler for autonomous databases.

With the introduction of 19c, which is available on Autonomous, we have machine

learning, which optimizes the work load. We have several unique features, like
automatically creating indexes for transaction processing. Or when it comes to data
warehouse, we are automatically able to create the data summaries for analytics.

We take advantage of columnar vector processing formats, and it is based on Exadata

Cloud Infrastructure. So both of these offerings, serverless as well as dedicated
we use Exadata behind the scenes.

So Oracle database is instantly provisioned regardless of any database size in case

of server lists, and there are no upfront infrastructure required or, in case of
dedicated customer, get a complete hardware stack with it completely isolated from
any other tenants.

And we have concepts such as virtual cloud networks that enforce, and it provides
you more control over applications as well as database consolidation and patching,
when we talk about dedicated.

In this slide, we are going to discuss how it compares both of these offerings--
serverless versus dedicated. What are the primary goals or benefits you get across
these offerings?

Primary benefit, when it comes to serverless, Oracle is managing or automating

everything for you. In essence, all their deployments, or lifecycle, or software
updates, they are taken care by Oracle. Customer is just able to choose what kind
of compute, how many OCPUs they want to enable, how much storage they need, as well
as the reason, et cetera.

In terms of elasticity, you can start with the lower minimum. So one OCPU can get
you started on autonomous server lists. And a minimum time commitment is basically
just an hour. And you can automatically scale online for a kind of true paper use
basis.

On the other hand, autonomous database dedicated, mostly the primary benefits are
around, it is your own database cloud, it is running on dedicated Exadata
infrastructure. It can run all of your database, any size, scale or criticality. It
provides you higher ISO lesson. It runs inside a secure isolation zone so that it
can provide you highest protection from any other tenants.

So you can figure multiple Exadatas or container databases for intracompany kind of
isolation. It provides you various customizable operational policies. So in terms
of control, or provisioning, or updates, they are controlled by customer, in terms
of availability and consolidation density. So it all comes with a dedicated
offering.

So now looking at serverless Exadata cloud infrastructure, serverless experience is

the easiest thing for a customer to get. Oracle, in managing everything, delivers
real application clusters for both-- basically both serverless as well as
dedicated. Customer just consumes the service. They can activate with one code, one
OCPU, and they can scale out.

So a lower minimum commitment, as I said, just one hour, and it designed for common
compliance applications or public cloud applications. As you see, it is all based
on Exadata infrastructure.

On the other hand, Autonomous Transaction Processing dedicated. So some of the

physical characteristic and constraints. It is designed for more mission critical
databases, where you need more control in terms of consolidation, ability to locate
dedicated environments. It again runs on Exadata hardware. That is the world's
fastest database platform.

It has fastest storage using latest [INAUDIBLE] flash. You have fastest
communication using RDMA or InfiniBand So a smart offload capability of queries of
column [INAUDIBLE] in Exadata storage. And you can transparently do a scaling for
both your OLTP as well as analytic applications.

So the hardware which is available, Exadata quarter rack. It comes with two compute
servers that can go all the way up to 92 OCPUs and roughly 1.5 terabyte of RAM. And
from the three storage cells, or storage servers, which are present in this Exadata
infrastructure, you get roughly 77 terabytes of flasks and 107 terabytes of usable
disk capacity through a spinning disk.
You have cluster or virtual cloud networks. So you can have one cluster per quarter
rack at moment. You can have maximal four-part cluster in a container database. And
in terms of high availability service level agreements on maximum 200 databases, it
provides extreme availability of maximum 25 databases.

Placement logic is based on, like if you have less than 16 OCPU, [INAUDIBLE] open
on one server.

You can always do over provisioning, especially if you are planning for application
testing and then rolling out to production. So those kind of scenario, it can
support overprovisioning. So maximum of 10 autonomous databases per OCPU is just a
guideline.

So in terms of selection considerations for serverless, that is definitely the

easiest way to get you started with or one hour minimum. You can quickly test out
all the ideas or you can terminate when you are done with your work. So you don't
have to worry about any placement or availability domain placement controls.

So they are not there with serverless. And for patch editing or software
versioning, since Oracle takes care of these kind of operations-- so customers
don't have to worry about these.

You can use, or basically configure, Oracle application continuity to minimize any
downtime impact to maintenance. Cloning capabilities are available today in the
server lists. So that is far easier test database creation.

Auto-Scaling is also available for hands-free scaling of variable workload. And

functionally, it is ready for any-- ready for any need to upgrade from serverless
to dedicated for production kind of deployment. So it is always a good idea to test
it out in server lists and then decide on moving to dedicated [INAUDIBLE].

So when it compares with the dedicated, it is definitely a higher cost entry. And
users must subscribe to a minimum of one month of Exadata infrastructure. When
subscribed, you can test out the ideas and quickly [INAUDIBLE] when you complete.
So no ongoing software costs.

It includes the availability domain placement control for latency sensitive

applications. So response latency is in microseconds, and it is consistent over
time. So there are no placement changes.

Quality control for patch editing and software versioning, which is different than
server lists. So Oracle completely deals with all patching, but you can control
what version and when you want to apply those changes.

So it allows you to stage a specific version to your dev test or pre-production

before production. Transparent application continuity automatically minimize any
downtime impact to the maintenance.

Test database creation requires an export or an import from the object storage,
because that is the way you are able to bring any database to Autonomous
[INAUDIBLE]. So cloning and auto scaling capabilities are also available.

So some of the feature considerations, since most of the customers are coming from
kind of environment where they have complete control on things such as creating
table space or user creations or indexes and things.

So here, you need to understand what is different between this dedicated and [?
serverless, ?] because that helps you take a decision when it comes to solutioning.
So in terms of create table space, on dedicated, you can do that. But in
serverless, you have no flexibility. You cannot create a table space. Similarly,
transparent application continuity is available. And in [? serverless, ?] only
application continuity is available.

Cloning support is going to be available in autonomous transaction processing

dedicated. Right now, it is not there. But in serverless, it's available. Vault is
available in dedicated.

Because database vault provides security controls so that operations can just do
whatever they are supposed to in terms of patching without looking into the user
data. So segregation of duties and all is very important. And database vault
provides that.

In terms of profiles, you have access to-- it uses Database Version 19c. In case of
dedicated, but in serverless, you have only preview. But 19c, we also have
available maybe later in serverless.

Auto scaling is not at present. Auto indexing is part of dedicated. And in terms of
RTO and RPI, it's documented and available with dedicated. OCI notification
services are integrated here. Maintenance history and reporting is available in
dedicated.

And for a complete list, the below link provides you details on what is unique to
server lists as well as dedicated. You can go through them.

Some of the other service feature comparison is VCN or Private IP or Azure Secure
Connect. So they all are available with dedicated backup [INAUDIBLE] configurations
are available on dedicated-- you can control the software update as well as
versions, which is unique to dedicated. And SQL or Enterprise Monitoring is
available across both basically. So these tools can be used with both of the
offerings.

So in terms of non-DLS SQL*Net, that is available on dedicated. But in serverless,

it is always going to be encrypted TLS SQL*Net connection.

RAC session or co-location tagging as well as keeping updates during critical

business periods. So these all can be controlled on dedicated environment.

So now moving to a DB Cloud Service comparison. And we'll talk about various
deployment considerations between autonomous versus automated. Automated is where
we have provided a lot of things in automated fashion through scripts or toolings.
But you have no machine learning or AI. So we'll talk about them in much detail.

Primary use case for autonomous database, serverless is basically, for data driven
results, are faster-- fast data-driven results [INAUDIBLE] by the data department.
They store data for SMB customers or developers who wants to quickly deploy a
database for their application development. Or even for data scientists.

So that would be some of the main use case for autonomous database. Because this is
the way we are basically differentiating. Developers, we are always depending on
system admins, or IT admins, to get an environment. And it is to delay the entire
process.

But now, they don't have to have knowledge on how these autonomous databases works.
With a few clicks, they are able to get a functional database running in time. Then
they would be able to do whatever is needed.

So all database operations is fully automated, [? usually ?] on SQL, they don't

have access to operating system or container. It provides you Exadata kind of
performance and availability and customizable for data warehouse or transaction
processing workload.

So that is also a very unique feature. The same database Oracle provides can be
customized or tuned for different kinds of workload we have here. So it is the same
skill set, which is going to probably manage both of these environments.

So server lists, as I said, it's ultra-simple and elastic. And dedicated provides
you customizable private cloud in public cloud.

So on the other hand, if you talk about Autonomous Database service, here we have
automated the database lifecycle. User operates-- they have DB access, they have
operating system or root access to these systems. You will be able to run all the
database versions on automated DB services like 11g or 12c or 18 or 19.

So all of the database features, including Java, is available on automated DB

services.

And two major offerings. One is based on Exadata system. So Exadata cloud service
or even Exadata Cloud customer, they provide scalability, the required performance
for your missing critical workload, as well as availability requirements.

The other offerings is around database cloud service, which is provided on virtual
machines for a smaller kind of workload, or Bare Metal instances which are able to
provide you single server. And database virtual machines provide you real
application cluster as well.

So depending on the workload requirements, you have several offerings available to

choose from between autonomous as well as automated databases.

Since Oracle Database provides you [INAUDIBLE] deployment, so if you look at the
left hand side, it is more of manual activity. So customers have databases
installed on generic hardwares on prem. And then, even though those customers who
have bought Exadata systems, databases are installed on Exadata systems. So that
provides a lot of automation or [? health ?] check or don't have to build the
entire thing from scratch.

So it has automated most of things. But if you look at the right hand side, we are
moving towards most autonomous thing.

So journey to the cloud can have different stages. And Autonomous Database Cloud,
it's also the same.

So some customers are embracing cloud for new deployments or doing legacy latency.
But some other customers, they have huge mixed implementations. But part of their
systems are running on prem, and some are running on the cloud.

So understanding where and how your customer might be using your data management
solution, that helps you really scope out your autonomous opportunities.

So as we see, in the left side, we have most manual implementation. So more of

[INAUDIBLE] on-prem installs. Either they're running on community hardware or
Exadata. That is the land of the traditional IT DBS. They are used to do
maintenance, patching, upgrade. Or you can say that most time consuming stuff.

So in the right side, we have moved through databases running on our cloud
infrastructure. So lift and sift kind of opportunity is still a kind of heavy
workload. And gradually, we move through our existing database [INAUDIBLE] services
to Exadata Cloud or Cloud at Customer and Database Cloud Service, they start to
introduce more automation as well as management capabilities. And that helps
release DBA workloads into more kind of interesting tasks.

So in the far right, we end up with a new Autonomous Database Cloud Services. And
there, you see all the benefits of reducing workload or risk of freeing up your
DBS. So those things are becoming true.

So another advantage is like all Database Cloud Service package includes database
transport and data encryption. Enterprise Manager packs, they include all of the
impacts, the data masking, subsetting or diagnostic tuning [INAUDIBLE] application
testing. They all come with enterprise [INAUDIBLE] version of the software.

And high performance, which is available across this cloud, they are able to
provide multi-tenant partitioning, advanced compression or Database Vault. So all
these cool features.

The extreme performance is adding on top of high performance. And they extend it by
adding real application clusters or database [INAUDIBLE] or Active Data Guard kind
of offerings.

Autonomous Database choice of cloud deployment. As you can see, we are talking
about more of our platform services, where you have databases running on virtual
machines or Bare Metal or Exadata Cloud Service or Cloud at Customer, autonomous
serverless or dedicated.

So there are several options. Like in terms of management, DBaaS, VM, or Bare Metal
is managed by customer. The Exadata Cloud Service or Cloud at Customer is again,
customer controlled. Autonomous is controlled by Oracle, both serverless as well as
dedicated.

In terms of private network, you can have it on your Bare Metal or VMs. Or even on
Exadata Cloud Service or Cloud at Customer, you can deploy a database in private
network.

In serverless, you don't have that scope. And in dedicated, you can still control
them. And databases will be now private network.

So in terms of single or multi-tenant, so it provides both single instance or

multi-tenant. And in serverless, it provides just a single tenant. And in
autonomous dedicated, you have both options available.

In terms of software updates, it is customer initiated in both of these automated

deployments. And in autonomous, it is-- serverless has got automatic updates. And
in dedicated customers, policy controls are possible.

In terms of private cloud, this is not a private cloud. But it is not available in
the Cloud at Customer offerings. But Exadata deployments are availabe in Cloud at
Customer behind their own firewall. And autonomous serverless is, again, not
available. But dedicated is available on Exadata Cloud at Customer.

So it offers availability, different database version support. So you can see DBaaS
as well as Exadata created cloud service that provide support for multiple database
versions, whereas serverless or autonomous dedicated, they only support 19c.

And disaster recovery is possible across Ads and reasons for both of these. So at
moment, for severless, we don't have DR as well as for dedicated. But that might be
coming soon.
So in case of hybrid DR, you can create a hybrid DR on these DBaaS, Bare Metal, or
VMs, as well as on Exa. But for autonomous serverless, you cannot do this. Or even
for dedicated, you cannot have a hybrid kind of solution.

For consolidation use case, definitely these two offerings, DBaaS and Exadata Cloud
Service or Cloud at Customer, is a good platform for consolidation. On dedicated,
you can do that, but not on serverless.

So when it comes to autonomous versus automated database service comparison, the

use case for autonomous databases is mostly for where you need cloud elasticity,
you need machine learning or kind of self driving instant provisioning, always
online kind of operation. There it supports all sort of workload, like JSON
documents, graphs, and more. They are a very good fit for autonomous database.

For Oracle Database Cloud Service, as we've seen, VM, Bare Metal or Exadata Cloud
Service or Exadata Cloud at Customer. So it provides you great availability,
flexible versions as well as features appropriate for a small to large database
deployments with a single instance or rack or automated backups, patching, and
custom controls.

The difference between customer-controlled versus Oracle controlled, that we have

to keep in mind. Exadata and Oracle Database is known. Customers are using in their
own center and using it for consolidation, highest performance. So Exadata is not a
very new technology. It is used across thousands of customers, and they're able to
run their mission-critical workload.

Oracle Database is used for either analytics or basically intuit your [INAUDIBLE]
model. Within your data center, you can run it on any platform.

So with this, I will move to part two of Autonomous Database dedicated. So we'll
see that in our next video. And thanks for watching.