Idc Oracles Autonomous Database 4497146
Idc Oracles Autonomous Database 4497146
THE
FUTURE
IDC PERSPECTIVE
Oracle's Autonomous Database: AI-Based Automation for
Database Management and Operations
Carl W. Olofson David Schubmehl
EXECUTIVE SNAPSHOT
FIGURE 1
This IDC Perspective examines the announcement and initial launch of Oracle Autonomous Database. It
also looks at how a search vendor is planning to use Oracle Autonomous Database in its business
operations and why.
Key Takeaways
• Oracle has recently announced its latest database product, a self-managing autonomous database.
• Oracle is at the vanguard of a movement to make enterprise software easier to deploy, use, and
administer, using artificial intelligence and machine learning to provide automation capability requiring
little or no human intervention to manage the software.
• While Oracle Autonomous Database is one of the first products in this new automated category, it won't
be the last. Other vendors are building similar tools for third-party management of existing enterprise
software packages, and many software vendors will also add this kind of capability to their offerings.
Recommended Actions
• Actively evaluate tools such as Oracle Autonomous Database for suitability in domains such as process
automation, digital transforrnation, and other business processes that require database access. lf this
seems to be a fit, organizations should begin evaluating Oracle Autonomous Database now, rather than
waiting and seeing how other organizations are faring.
• Start with low-risk pilots to identify the best opportunities for full-scale implernentation.
Develop rnetricslmeasurement practices to determine how and where these systems are succeeding.
Introduction
Oracle recently announced Oracle Autonomous Database, an Oracle Cloud DBaaS offering based on
Oracle Database, but with special features that enable it to self-tune and self-heal. This service
relieves database administrators (DBAs) of the remaining operational tasks that normally accompany
even a cloud-based database deployment. Those tasks include advanced tuning functions, database
security, and troubleshooting. DBAs will soon be able to focus more time on design and development
activities instead of spending time administering their database installation. Database administration
will be fully automated, providing Oracle with the opportunity to guarantee 99.995% availability
(including unplanned maintenance, bug fixes, configuration changes, and security patches) and no
more than 30 minutes of downtime a year. The product will be tuned continuously, using machine
learning (ML) models of observed behavior as patterns of user and system's behavior change.
Collectively, through its primary research, IDC has found that as much as 75% of the total cost of
database management (including software and hardware costs amortized over five years) can, in
some cases, be labor. Any DBaaS eliminates a large amount of the operational work (running
backups, applying patches, etc.), but this capability redirects most of the remaining labor from "drudge"
work to high-value work that directly benefits the enterprise. In the first version of this capability,
Autonomous Database (version 18c) will only address analytic (that is, non-transactional) databases.
However, the plan is to support all Oracle Database workloads over the course of the next year.
Database administrators are constantly creating and dropping indexes, changing data storage
allocation parameters, defining and redefining partitions, and using various other techniques to boost
performance. They must also monitor the database for any problems and perform preventative
maintenance. In addition to this, the processes involved in applying software upgrades and patches
can be highly disruptive and time consuming.
Although most modern RDBMSs, including Oracle Database, have developed many features and
functions that are designed to lessen the burden of these activities, it remains the case that, at the
highest level of size, complexity, and performance criticality, meeting the SLA for the database
requires a set of skills generally regarded as something of a black art. And of course, even the best of
DBAs can make mistakes — and those mistakes can cause poor performance and unscheduled
downtime.
Over the past three years, changes in data collection, compute power, and cloud-based services have
mitigated the challenges outlined previously. We now have various IT services that are somewhat AI
driven, using machine learning, natural language processing, and question-answer technologies to
help build knowledge bases and predictive models that can be used to automate various parts of IT
operations.
Oracle could not offer this service without applying some advanced technology to solve the more
nettlesome tuning and maintenance problems. Adjustments based on rules and patterns are not
sufficient, and the human intervention required to apply such maintenance would make this service too
expensive to represent a significant improvement over the prevailing do-it-yourself model. To scale this
service, Oracle developed automated database tuning and maintenance based on machine learning.
Although machine learning libraries have been around for decades and have been offered as part of
many of the world's statistical packages, including IBM's SPSS, SAS, and so forth, the use of machine
learning by enterprises hasn't been widespread until recently because these algorithms require a lot of
data and a lot of compute power. However, many leading technology firms have been using machine
learning tools over the past few years to improve programs in many areas, such as image recognition
and programmatic advertising, as well as product and story recommendations and now IT and
database operations.
Supervised machine learning begins with examples of training data paired with identifying labels (e.g.,
right or wrong, positive or negative) selected from the categories to be learned. Using these pairs of
example data and labels ("training data"), the system learns parameters of statistical models that it can
then generalize to unlabeled examples of data items that were not seen in the training data ("test
data"). In most cases, the learned models improve over time via a feedback loop that adjusts the
model parameters to better reflect additional sets of training or production data. The performance of a
learned model can be measured by simple prediction accuracy or by the business metric the learned
model is designed to support. Performance depends on the degree to which the training data matches
the real world, the choice of algorithm, the algorithm's parameters, and the quantity of data.
With these tools, the end game is all about making applications smarter by using special libraries
containing self-learning algorithms, which, when unleashed on a data set, can learn and self-program
to solve various types of problems. The combination of high-performance compute resources,
tremendous amounts of data, and these ML/DL libraries is solving problems and challenges without
the need to resort to programming. These ML/DL libraries and technologies are being used for an
ever-wider array of use cases, from image recognition and disease diagnosis to pricing optimization
and product recommendations.
As an example of how this works, Figure 2 shows how Oracle is going about the process of automating
error handling within its Autonomous Database product. Operations log data is used as the source of
information that is then used to train a machine learning model with some level of supervisory insight
being provided by database experts. That model is then moved into production, helping to identify and
handle errors, performing root cause analysis, and providing preventative and corrective
recommendations for action to solve the various problems that emerge as part of the daily operations
of the database.
While this is a relatively simple example, Oracle is working to automate most, if not all, aspects of
database operations with its new Autonomous Database product — and automated error handling is
just one example of where AI and machine learning are being applied in the new product offering.
ilk
Data Center Tenant
Training
OIMID
4 Feedback & —e.
P$ Improvement " =I 1.1
\ 1
T A
Ticking System
Model
,^,Expert Input - Control Plane
c
- • 1, -II_
•• Cloud Ops
TFA
i
Log cleansing, Model generation Real-time log
metadata with expert processing, fault Log Store
creation & scrubbing detection & notification
clustering
Real-time health r--- 1
dashboards for root
cause analysis & Diagnostic Service
push of preventative &
corrective actions I
Maar indicated that 11880.com runs Oracle Database in support of a variety of both transactional and
analytic applications. The transactional applications include the SAP application suite and Siebel
CRM, and the analytics are driven by Oracle Business Intelligence (BI). "We were eager to be a beta
tester for the [Oracle] Autonomous Database, because it will be possible to have heavy cost savings,"
he told us. "The database architects want to fully concentrate on architecting highly sophisticated
databases. They want to design leading-edge technology and are happy that all the DBA stuff is put
out of their hands."
Maar indicated that the company was an early beta tester, and the first two phases of the rollout were
a bit "rough." It had been testing some of the workloads in comparison with a memory-optimized
analytic RDBMS, and the other RDBMS was winning easily in the first two phases. But, in the third
phase, not only did the Oracle Autonomous Database outperform the memory-optimized competitor
but it did so while requiring no human effort at all.
Maar expects the benefits to 11880.com to come both in the form of process cost for running the
databases and in smoother business continuity and better performance. At the time this document was
written, 11880.com had not yet moved the Oracle Autonomous Database into production but the test
results are very promising indeed.
Oracle Autonomous Database delivers major benefits to customers in two forms: labor savings by
automating database tuning and problem detection and reduced downtime due to unplanned outages
and software/system maintenance. It should provide Oracle Database users with a major incentive to
move to the Oracle Cloud. Currently, the Oracle Autonomous Database addresses analytic workloads
such as data warehouses — but we can expect a transaction-centric version later this year.
This product is likely to evolve rapidly over the next couple of years. If it looks like a good fit for
your organization, you should look at getting your feet wet now, but watch how it evolves
before going all in.
As more clients sign-on, Oracle Autonomous Database will learn from their cases, adding a
richer range of situations and responses and making the system "smarter." So, expect results
to improve over time.
Competitors are likely to introduce their own autonomic features, so if your investment is in
another vendor's product, check with that vendor to see what the plan is.
A move to the cloud involves a radical shift in the IT topology and cost structure. Be sure you
anticipate all the issues involved in moving the database and its applications.
There are also personnel implications. Many DBAs with whom we have spoken in the past
have complained about the tuning and maintenance tasks they have to perform because those
tasks keep them from the stack of user requests for business-relevant changes to the
database. Those DBAs will be happy with this technology. Others, however, regard their
detailed knowledge of Oracle Database tuning to be their source of job security. They may not
be so sanguine about adopting this technology.
Communication with others is important. See what they are doing and what lessons there are
to learn. This may affect your timing in moving to Oracle Autonomous Database.
LEARN MORE
Related Research
Cloud Database Adoption Trends (IDC #US42074617, November 2017)
IDC FutureScape: Worldwide Analytics and Information Management 2018 Predictions (IDC
#US42619417, October 2017)
IDC Market Glance: Cognitive/AI Systems, Search, and Content Analytics, 3Q17 (IDC
#US43098517, September 2017)
Market Analysis Perspective: Worldwide Cognitive/AI Software Platforms, 2017 (IDC
#US42072717, September 2017)
IDC's Worldwide Semiannual Cognitive/Artificial Intelligence Systems Spending Guide
Taxonomy, 2017: Update (IDC #US43031517, September 2017)
Is Cognitive/AI Technology Adoption Real? (IDC #DR2017_T5_DS, February 2017)
IDC PlanScape: Implementation of Cognitive Systems (IDC #US41477516, June 2016)
Intelligent Personal Assistant Adoption for Work (IDC #US41477816, June 2016)
"AI and machine learning are increasingly pervasive technologies that are being used now in a wide
range of products and services," says Carl Olofson, research vice president, Data Management
Software at IDC. "Oracle has recently launched its Autonomous Database product that uses AI and
machine learning to automate operations and management of the product. Organizations should
consider Oracle's Autonomous Database cloud offering if they are evaluating cloud database offerings
and are looking for a fully managed database service from a proven database vendor. The future of
autonomous products like Oracle's Autonomous Database is bright, but organizations need to also
assess the impact these types of products will have on their IT staff."
Global Headquarters
5 Speen Street
Framingham, MA 01701
USA
508.872.8200
Twitter: @IDC
idc-community.com
www.idc.com
Copyright Notice
This IDC research document was published as part of an IDC continuous intelligence service, providing written
research, analyst interactions, telebriefings, and conferences. Visit www.idc.com to learn more about IDC
subscription and consulting services. To view a list of IDC offices worldwide, visit www.idc.com/offices. Please
contact the IDC Hotline at 800.343.4952, ext. 7988 (or +1.508.988.7988) or [email protected] for information on
applying the price of this document toward the purchase of an IDC service or for information on additional copies
or web rights.
Copyright 2018 IDC. Reproduction is forbidden unless authorized. All rights reserved.