Oracle Cloud Infrastructure White Paper
Oracle Cloud Infrastructure White Paper
INFRASTRUCTURE (OCI)
What Makes it Unique for Your Oracle Workloads in the Cloud
Simon Pane,
Principal Consultant
Oracle ACE
Co Author
Nelson Calero
Internal Principal Consultant
Oracle ACE Director
Oracle Certified Expert
There have been options for running Oracle database workloads in the cloud for
many years. In fact, Amazon Web Services (AWS) was clearly one of the first places
to run Oracle databases in the cloud either on IaaS (AWS EC2) or through their
Relational Database Service (RDS).
Recently Oracle has been promoting their 2nd generation cloud known as Oracle
Cloud Infrastructure or “OCI”. But what makes OCI a more compelling option for
cloud database workloads and what are the limitations?
With the other option of having a managed database service, AWS RDS is really
the main non-Oracle option. RDS takes care of server set up, software provisioning,
database creation, data backup, patching, and DR options if required. All based on
user inputs specified during provisioning. However, RDS is significantly locked down.
There’s limitations in what initialization parameters can be adjusted, limited access
to database internal APIs (some access is wrapped into Amazon-provided APIs), no
access to the underlying operating system, and non-direct workarounds for accessing
or viewing generated log and trace files.
Regardless of the cloud vendor and whether IaaS or a DBaaS option is chosen, a few
things are not possible (at least not in an Oracle-supported manner) in non-Oracle
clouds. Mainly:
• Oracle Real Application Clusters (RAC)
• Oracle Exadata (Engineered Systems)
There are some technical workarounds for building Oracle RAC databases in non-
Oracle clouds, often involving third party software. But the Oracle support position on
such deployments is complex. For a more complete explanation, refer to the Oracle
supplied white paper called “Oracle Real Application Clusters (RAC) Support on Third-
Party Clouds - An Overview and Clarification”
At one point the Oracle Multi tenant option was also restricted in third party clouds but
that restriction has since been lifted, likely as Oracle wants to push customers towards
the container database (CDB) architecture. The non-CDB architecture is currently
available but deprecated and in the future will be removed.
OCI’s managed database service (called the “DB Service”) is also more elaborate
than AWS RDS. Clearly the first few main differentiators are the options for
RAC databases and to build on powerful bare metal (BM) servers in addition to
traditional VMs.
Beyond that, the OCI DB Service provides a lot more access. You’re given full access
to the database, including access to the SYS and SYSTEM users, as well as nd full
“Oracle software owner” and root access to the underlying operating system.
But full access brings the ability to break things. For example, it’s possible to move
listeners to a different port, apply patches manually, or even create new databases
manually using the DBCA. But doing so will effectively “break” the managed part
of the managed service. This means that a database added manually through the
DBCA instead of the OCI-specific tools or web console won’t appear in the OCI
console, won’t be monitored by the new OCI monitoring service, and won’t benefit
from the simplified patching tooling that the service provides. Just because you
can, doesn’t mean that you should.
The line between what can be done and what should be done is a little bit
blurry. The best way to ensure that you stay on the proper side of that line is by
administering the system and the database only through the OCI-provided tooling
such as the OCI web console, OCI-specific command line utilities including dbcli,
dbaasapi, and the OCI CLI utility, or the REST APIs.
That last point is one that often takes people by surprise. Some hear the message
that RAC is available and is supported on OCI which is technically true. But only
in certain configurations. Specifically 2-node RAC databases on the VM-based
DB Service or as 2, 4, or 8 node RAC databases with ExaCS. What you cannot do
is build your own custom RAC with a size and shape of your choosing based on
IaaS. OCI still lacks the shared block volume storage, dedicated interconnect, and
multicast networking options required to support that.
OCI supports a BYOL model and a license-included model. With the latter, rather
than first choosing the addition and then adding in the required Option Packs
(choosing out of more than a dozen), the license-included database services are
grouped into four editions:
• Standard Edition
• Enterprise Edition
• Enterprise High Performance
• Enterprise Extreme Performance
The last two are new and are cloud-unique offerings. Each includes several of the
aforementioned Option Packs.
And a nice plus is that all four editions include Oracle Database Transparent Data
Encryption to support database-based data encryption at rest in the cloud.
Between the new edition-based packaging and new cloud-based price points
including consumption-based (i.e. “pay as you go”) models, it makes using the
Oracle database arguably more cost-effective in OCI. Exact costs change at
The recently announced interconnect to Microsoft Azure. This really gives customers
the best of both worlds allowing them to take advantage of Azure-specific offerings such
as Azure Active Directory and Office365 while not having to compromise on their Oracle
database configurations.
Prior to this, customers wanting to move into Microsoft Azure, but with complicated
Oracle database infrastructure, had some limitations:
• They could keep the Oracle part of their estate on-premises, hence, making
a hybrid-cloud deployment, meaning that they couldn’t fully divest of their
existing data centers.
• They could simplify their complicated Oracle workloads and run single-
instance only databases on Azure IaaS.
The new Microsoft-Oracle cloud interconnect adds the third option of an easier-to-
implement and a more tightly coupled multi-cloud deployment. Hence, they can fully
divest of existing data centers and use Azure for what it’s best for and OCI for what it’s
most suited for.
For this reason it’s probably a big step in the right direction for Oracle and their
customers. And will likely lead to more adoption of OCI in this multi-cloud model. And as
OCI adoption grows, so does its capabilities, scale, locations, and reliability.
Initially the Azure-OCI interconnect is in the popular U.S.-Virginia (Ashburn) region only
but this will likely expand rapidly.
CONCLUSION
There’s no doubt about it: OCI offers by far the most options and arguably at the best
price point for running non-trivial Oracle database workloads in the cloud. It matches
the functional capabilities of other clouds from a “build-it-yourself” perspective on IaaS
with their own uniqueness in the underlying infrastructure (i.e. compute shapes and
VM networking differences). It offers a DB Service which far exceeds the capabilities of
Amazon’s RDS by giving options of VM or BM offerings and full access to the database
and underlying OS. Plus the option of RAC on VMs. At the high end of the performance
perspective, the ExaCS service gives customers the full power of an Oracle Exadata-
based database but in the cloud without the procurement and long term commitment
With over 20 years of experience managing the world’s most complex Oracle
systems, Pythian’s certified experts can help you maximize your investment in
Oracle technologies. Pythian’s certified Oracle experts configure your architecture
to suit your unique processes and continuously maximize the performance of your
environment with our renowned 24x7x365 support. By combining the platform
with the people who know it best, we can ensure constant system optimization,
network monitoring, and performance that you can depend on.
Simon Pane
Principal Consultant
Oracle ACE
Nelson Calero
Internal Principal Consultant
Oracle ACE Director
Oracle Certified Expert
info@pythian.com twitter.com/Pythian
linkedin.com/company/pythian +1-866-798-4426
ABOUT PYTHIAN
Founded in 1997, Pythian is a global IT services company that helps organizations transform how they compete and win by helping them turn data into valuable
insights, predictions and products. From cloud automation to machine learning, Pythian designs, implements and supports customized solutions to the toughest
data challenges.
OFFICES
Ottawa, Canada London, England
New York City, USA Hyderabad, India