One year on from the first iteration of “Autonomous” being preached by Larry, I am left wondering if the Oracle Autonomous Database really is the self-driving car that Oracle profess it to be, which will lead us all to enlightenment?
Rewind to Oracle OpenWorld 2017 and the marquee announcements were not the usual hardware of software feature updates that we were used to; it was all about Cloud and AWS bashing. The cloud announcement to take any notice of was the Oracle Autonomous Data Warehouse (ADW), which also turned into an AWS bashing (again) and how (if you were at the keynote you’d believe me) it would solve the database equivalent of world peace! You would marvel as the Autonomous cloud would do such things as automatically scale, automatically patch, auto tune and, through “AI and Machine Learning” would automatically prevent any type of cyber-attack. Wow! This new age of Autonomous technology would start with Data Warehouse in the cloud and over time would evolve into the Autonomous Transaction Processing Database (ATP) for OLTP workloads.
This year at OOW18, Larry Ellison announced the next generation of the Oracle cloud with its autonomous being once again the differentiator, this time the difference is that Oracle have brought autonomous to its core PaaS capabilities and how autonomous is going to be integrated into its fusion SaaS products and that’s a big ‘Wow’!
But, before we get too excited, what really is Autonomous? What really is Oracle’s understanding of what Autonomous is and how does it differ from the myriad of Database tools that we have today to deliver business efficiency, prevent downtime, create agility through the speed of change…all, essentially, creating the perception of autonomy. After all, many Oracle customers out there can already build a version of “Autonomous” using Engineered Systems infrastructure, configured in a Maximum Availability Architecture and having good operational procedures in place to enforce patching/updates!
Is the Oracle marketing machine blazing a trail with an attempt at redefining what autonomous really means?
The metaphor used by Oracle to articulate the value of Autonomous is idea of the self-driving car. Whilst self-driving cars are no longer the stuff of fiction and they are being tested across the globe in “real world” tests, they are nowhere near close to mainstream adoption. Why? Because the self-driving vehicle manufacturers don’t administer, build or control the road infrastructure, or the other workloads cars. The manufacturers cannot, yet, guarantee that self-driven cars will not crash with other workloads cars.
So, I do believe that Oracle’s version of autonomy will undoubtedly work in a self-contained, self-developed and self-policed road network where we have removed the all the risks of crashes downtime and security threats. But where is that panacea? Software-as-a-Service! Ah, I get it now. “Autonomous” technology from Oracle really will be of value IF you are consuming an application deployed upon the complete Oracle stack of apps down to disk combined with “AI and Machine Learning”. And why wouldn’t it? Oracle are in control of the development of all tiers of this stack, therefore (to continue this car metaphor) they built the network highway, wrote the rules code and developed built the Apps and Databases vehicles to drive on it (remember, they’re all Oracle-only vehicles!).
So what happens in the world of Platform and Infrastructure as-a-service, where customers are still in control, to a certain degree, over the configuration of their Cloud Services and indeed if any non-oracle technology is architected in? What happens when an organisation who wants all the benefits of Autonomous Databases but these lovely little autonomous robots have to co-exist in a customer’s eco-system of mixed workloads and varied configurations? Jenga anyone? What happens when this “Autonomous” technology is available on-premise?
I think it is for this very reason that Oracle is starting with the Autonomous Database in the Oracle Cloud and with a predictable, if not pre-defined, set of workloads. It would be a little too radical for Oracle, even by their standards, to tell its customers “fill your boots” with Autonomous if they had no control over what hardware and software it will be deployed against. I bet even HAL could only work on a certain a certain platform! 🙈
We know it will come to on-premise infrastructure at some point, and this journey will start with Oracle’s “Cloud at Customer” offering, but there are so many other considerations to factor in other than using soundbites like “AI and Machine Learning” to make be believe that this will one day be as ubiquitous as the heteronomy Oracle Database we all know and use today.
I do see great value in Autonomous Databases for customers who are developing applications in the cloud and, specifically, for any ISV’s developing APEX apps I see this as a great platform to build, deploy and run “SaaS” like offerings for your customers. APEX is a feature of the Database, therefore I wonder what “Autonomous” features could creep into APEX over time? That’s a thought for another day.
So in conclusion and back to the Autonomous car; today’s society is not quite ready for self-driven cars to drive past our schools, but we don’t mind been driven down the motorway with our hands resting on the steering wheel! I would argue that the answer to Autonomous Databases is just the same as the adoption of automatous cars; how confident are you to take your hands off the wheel and commit to a level where we feel that technology can take over critical choices on our behalf and where the wrong balance would have fatal consequences.
Author: Martin Guthrie
Job Title: Technical Account Director
Bio: Skilled in Oracle Core Technology, Engineered Systems and Cloud. Martin is an experienced Oracle Solution Architect with a demonstrated 20 plus years of helping customers identify and solve their Oracle infrastructure issues.