Although Larry started his presentation last week with a small faux-pas and members of the audience had to wait for the correct slides, waiting is definitely not something customers can expect from the Oracle Cloud with the new 18c Database. Everything will be better, faster, and most importantly “self-driving”.

Highlights of the launch featured the much-awaited Autonomous Data Warehouse, which he claims will be as “revolutionary as the internet”, as well as improved performance, manageability, and Security.

Basic Principles of Machine Learning

Machine Learning has come a long way since Alan Turing, the Nearest Neighbor algorithm and NetTalk, and we are moving all too quickly into a domain where science-fiction and reality are starting to collide. Machine Learning is a sub-set of artificial intelligence where computer algorithms are used to autonomously learn from data and information. In the past this has been used to teach computers patterns by which they can establish a set of rules use for predicting the next logical outcome in a sequence.

The process of sequencing has been great for regular occurrences or normal, predictable events such as using picture-scans to build rules for facial recognition programs, but has been historically less suited to highly irregular and anomalous phenomenon such as stock trading or determining emotional responses. When machine-learning is combined with a set of predefined rules by which further data can be ingested and augmented to the systems set of principles, when it can truly learn for itself and respond accordingly, we have an autonomous machine or system.

Machine Learning Applied to DB Management

Whether these principles can be applied to database management and SQL optimization is yet to be seen, in these early stages. But Ellison promises that with the recent progress in ML technologies, we can expect to see significant progress in automotive autonomy (self-driving cars), improved facial recognition, and automatic cyber security. The 18c release revolves around Oracle taking advantage of this progress in ML and implementing principles and algorithms to its databases where possible. 18c Databases will automatically recognize and defend themselves against cyber attacks (Automated Cyber-Security), complete all resource provisioning autonomously, and patch or update themselves as needed (Autonomous Database).

The important thing to understand in all this is that SQL tuning is the biggest challenge the database science world has faced in the last two decades. Whether it will be a simple fit as with facial recognition or much more complex as with stock market analysis is yet to be seen. Replacing traditional human DBAs with machines would have radical impact on labor costs surrounding database management, and in turn would change the way a sizeable portion of the IT crowd concerned with DB Management would function. It is not so much all the DBAs would be out of a job in the next year, but rather that their time would be spent less on the grudge work of maintaining databases, but more on the high-knowledge activities that truly generate value for an enterprise.

Machine Learning Progress

Most of the improvements in autonomy can be attributed to better and richer pattern definition and improved anomaly detection. This makes sense when we consider that Oracle has access to usage pattern data from a very large customer base, for the last 40 years and in that time has gained a lot of experience in developing and building databases. Combined with what they must have learned from customers’ cloud usage in the last few years, this lays a good foundation for data ingestion, rich pattern recognition and improved machine learning algorithms in the Oracle house.

From this collection and processing of large amounts of systems data from multiple sources, Oracle’s new ML capabilities promise to be able to not only better classify normal query patterns and automatically tune the Database for optimal performance (that would be nice!), but also detect and connect anomalous events and attacks, and have the database automatically defend and patch itself in response. Both of which are quite a big ask, and definitely more ambitious than any other current projects we see nearing the market.

If this is the case, automation engineers and cyber security specialists may have a lot less, or at least very different work to do soon – but as stated by Ellison himself at the launch, it is doubtful that these new features will replace current security policies and human involvement entirely. Our take: at least in the early stages a significant level of human interaction will still be required as a significant degree of judgment is involved when formulating appropriate responses or making reactive decisions.

Furthermore, new attacks will always be created, and not only the database but also the people behind the attacks will continue to get smarter everyday. As of now we still need human supervision and intervention, but as human interaction recedes, perhaps we will see a war of the algorithms, computing back and forth until one machine breaks the other, sometime in the very near future. Hey, Siri can already have a meaningful conversation with you, her talking back and arguing with you would just be the next step.

Oracle Cloud 18c Autonomous Database

Seeing as the 18c Autonomous Database is heralded as the game changer of the decade, let’s take a look under the hood and see what some of the things that are promised, will actually deliver. Technical characteristics of Oracle’s 18c Autonomous Database will include 4x faster in-memory OLTP access, 5x faster RAC for high contention OLTP, 2x faster streaming ingest for IoT workloads, zero impact grid infrastructure patching, automatic pluggable database (PDB) provisioning in the cloud, NVRAM read row-store, column-store, and SQL, as well as 2x faster in-memory store, 100x faster in-memory analytics for external data, 100x faster approximate query processing and Active Data Guard for data Warehouse loads.

Okay, so apart from significantly improved performance especially in the analytics and in-memory domains, the most valuable new addition included above is Pluggable Databases with copy-on-write Snapshot cloning capabilities inherited from Exadata. This will make Oracle significantly more attractive to Test/Dev refresh use cases and enhance the rate of agility and innovation for customers on the platform.

Truly rolling updates with no-downtime patching are a pretty big deal. Familiar BYOL concepts allow customers to make use of all licensed features in autonomous DB at no additional costs, and an additional 30% discount on PaaS is offered when monthly dollar volume commitments are made.

Furthermore, the new database will come in fault-tolerant RAC configurations and has Exadata-like storage (with predicate pushing capabilities and hardware acceleration). This means Exadata level performance all for the price of traditional Oracle DBs. And although currently Exadata is only available in large quarter, half or whole-rack configurations, 18c is going to be available with 1-2 core increments, making it truly elastic. However, it is worth noting that a lot of the “autonomous” features are not really part of the database itself, but rather built into the toolset which surrounds it.

The Autonomous DB will be available in 4 flavors within the next few months; starting with the most commonly used Autonomous Data Warehouse Cloud (Available Now), the Autonomous OLTP Database Cloud, Autonomous Express Database Cloud, and Autonomous NoSQL Database Cloud being released in June. A series of autonomous services is said to follow sometime in the summer as well.

In practical terms, an autonomous DB would mean a lot less human intervention in performance tuning which would result in a lot less compute and storage consumption, due to automatic data compression, fewer human errors and malice, as well as reduced labor costs. But equally as important, a fully autonomous database with dynamic hardware allocation and decoupled storage and compute would be able to provide true elasticity and make manual provisioning a thing of the past. If all of the above hold true, this would be pretty fantastic and represent a big step in Ellison’s journey toward an autonomous cloud.

Global Expansion and Availability

In terms of availability, Oracle already announced in October last year that 12 new datacenter regions were planned for 2018, including five new data centers in Asia, and the rest in North America, Europe and the Middle East. The location details provided by Oracle as of yet remained sparse but the list includes China, India, Japan, Singapore, South Korea, the Netherlands and Switzerland, along with two unspecified Canadian locations, and two sites in the US for Department of Defense workloads. From what we can tell the current two failover regions in the US remain Phoenix and Ashburn, SLA guarantees are still at 4-nines, and planned downtime has improved slightly to under 2.5 minutes per month. We hope that Oracle still has some surprises left in their strategy for expansion in this area, as enterprises may hesitate to adopt some of the 18c features if resources aren’t accessible and offered in availability zones close to them.

Conclusion

Lots of new services and features have been released with the 18c release, some of which promise to change the world. If Larry keeps what he promises this will change the game entirely and bring on a new age of cloud computing which in turn will revolutionize large parts of the IT sector. Yet only time (and extensive testing from our side) will tell where the new boundaries and limits of the Oracle Cloud lie, and what is truly possible.

While the 18c release represents a large step forwards in the direction of hosted DB development and simplification via automation, one should remember that Oracle is the first to use ML in this context and that a “learning curve” is to be expected (no pun intended). What Oracle is doing is seriously ambitious and has great potential, but also represents a serious challenge, and they are not alone in trying. Amazon won’t just roll over and die, just because they weren’t first to market, and it is likely that Oracle will not be there alone for long. It is thinkable that AWS and GCP have similar technologies in the pipeline, so it won’t be as easy for Oracle to topple Amazon as Larry alluded to in his presentation. Let’s see how it will all work out.

One thing is for sure – Oracle 18c brings many developments in PaaS/DBaaS with it, which in combination with the already fast and high-performant IaaS layer, makes Oracle an ever more attractive option for enterprises looking for cutting-edge cloud solutions. With 18c it looks like Oracle is one step closer to where it needs to be, so that customers can build on a reliable and performant platform, rich in services. In the future we will definitely see more updates on the PaaS side than anywhere else. Development efforts will be focused on creating new high-level services that allow organizations to develop applications in a much faster and easier way, paving the way for a cloud environment, which fosters inexpensive, rapid application innovation.

If the promises are met, then the new world being created is an exciting one and we look forward to working in it!

**DISCLAIMER**
Whilst we are avid technology geeks ourselves and love the nitty-gritty lugs and bolts, kernel profiling and digging through stack traces, we also recognize the need for a higher-level, more digestible approach to understanding the cloud computing landscape. From this origin and perceived need the AVM Consulting Business Blog series has a slightly different tone, aimed at business or management professionals and decision makers. We hope that this series of cloud business blogs will provide valuable information and new insights into the otherwise highly technical and rapidly changing cloud environment. Lastly, it is important to note that the views expressed in these blogs merely represent the opinions, perspectives, and point of view of AVM Consulting, and although some of the findings are based on facts, the meat of the content is purely subjective and open to interpretation. This is what we think, do what you will with this information.CATEGORIES