Database Superpowers – Artificial Intelligence and Machine Learning

In March 2018 Oracle announced their machine learning – self driving, securing and repairing cloud database. In August they launched following keynotes and have set predictions up to 2025. But what is an autonomous database and how does Artificial Intelligence come into play? At &You, we understand the need for a well-managed database and get very excited when we discover new and innovative ways companies can streamline their data systems.

Before we delve into ‘the what’, let’s look at ‘the who’? Oracle Corporation, formerly Software Development Laboratories (1977–79), Relational Software Inc. (1979–82), and Oracle Systems Corporation (1982–95), was founded by Larry Ellison, Bob Miner and Ed Oates. Oracle specialises in database technologies and software, and are statistically the most popular (Feb 2018 Statista)/(Aug 2018 Software Testing Help) providers of database management systems. During the 2018 March Oracle keynotes, Larry Ellison introduced the company’s next step;

“…we think this new autonomous database is maybe the most important thing Oracle’s ever done in terms of data management—and we’re the #1 data-management company on the planet right now.”

Larry Ellison

So what is an Autonomous Database? Oracle describe it as ‘a cloud database that uses machine learning to eliminate the human labour associated with database tuning, security, backups, updates, and other routine management tasks traditionally performed by database administrators (DBAs).’ Oracle focuses on the ROI for this specific product, but what if you were a first time investor looking into artificial intelligence and data management?

Artificial Intelligence and Machine Learning is still largely shrouded in mystery…

…and to some, a thing of fantasy. It’s most commonly seen in the films, set years in the future and generally focuses on replacing the human race. Films including I, Robot, Terminator, The Matrix, to name a few. If we were to set aside the fear factor and acknowledge that it isn’t as advanced as Hollywood has depicted, AI is very cool and very much an asset of the 21st century. In theory terms Artificial Intelligence is the psychology behind what a machine learns and a key super power for the world of databases.

But what do we mean by Machine Learning? Well, it can be broken down into several types – Supervised, Unsupervised, Semi Supervised and Reinforcement. Oracle has adopted multiple applications from their bank of automated software and have designed their own version of a Self-Supervised application. Self-Supervised Machine Learning adopts and combines the best bits of the industrys’ key algorithms, to produce a super-efficient artificial intelligence.

Learns from the past to predict the future. Here is a brief look at two types of Machine Learning algorithms and how they work:

Supervised requires a large amount of training using pre-existing datasets, manual labelling and classification, to formulate predictions on data not yet known. For example, if someone wanted to predict the peak times at a car park, this algorithm would use historical data to find out when the carpark is busiest and map out future instances.

Unsupervised doesn’t use data labelling as it draws its information to achieve a different goal. It looks at the structure; to find patterns and relationships, which is perfect if you are looking to find correlating groups. For example, if you had a gallery of photos of all your friends, there’s nothing to tell the algorithm who each person is but they need to be separated into albums based on different factors e.g. location, year, etc.

In both instances the machine is fed, the goal is set, but how the goal is achieved varies with each type of Machine Learning. In all instances they work with a high volume of data, working in a minimal amount of time, resulting in more accurate output in comparison to a human doing the job.

Every database design requires humans to manage it

Whichever type you’re looking at there is one contradiction to these clever pieces of tech – the necessity for human intervention. This is what Oracle has focussed on in its autonomous database; to remove the ‘donkey work’ a DBA (database administrator) must do to fine-tune and manage the system. This is perhaps when we enter the fictitious world, where there’s an uproar in human society over jobs being taken by machines. This certainly throws a lot of questions into the growth of human vs machine roles but we’re solely looking at the ‘what’ not the consequences of the ‘what’.

Oracle in the case of the autonomous database focusses on a dual management approach. There’s no getting around it, we need to know: what we want to achieve (goal) and what do we do with the results (implementation). These algorithms need us to tell them what to do and we need to know what we can do with the information we’re given. Otherwise it’s all for nothing.

We should also consider, DBAs are usually working on a number of databases (Oracle estimate 50 plus) and don’t work 24/7. Having the addition of autonomous software would be a great solution to spreading a company’s resources and pushing the boundaries of what can be achieved.

If knowledge is power and time is money, these AI design databases are the present not just the future!