Autonomous - the word of the moment. From the increasingly ubiquitous autonomous car (now more than just Tesla), through chatbots, to personal digital assistants, and even autonomous household technology, it seems the world is going autonomy crazy.
Into this Autonomous world steps the Oracle Autonomous Data Warehouse Cloud service, the first of a series of offerings of full self-driving, self-managing, self-healing, self-securing systems. No Install. No DBA interaction. No patching. Just click a few buttons, load your data in and off you go.
But surely we’ve been here before? Indeed, a quick Google search brings up the following examples of white papers by Oracle with a reference to the database being self-managing all the way back to 2003:
Ah, but the world has moved on. Computing power and machine learning (ML) has entered its first golden age, powered by advances in storage capacity (the more data in the model, the better the model typical is), GPU power and Cloud. Most recently it’s been cloud computing that has enabled organisations of all size to run ML models on large scale clusters, without the need to build and fill data centres full of servers and all the Capex and Opex cost and issues associated.
It’s surprising seeing how much cynicism - and sometimes downright negative sentiment - exists in the community (perhaps it’s just those vocal on social media with an axe to grind) around the launch of Oracle Autonomous Database.
My argument is not that people will be removed from the administration lifecycle, just that where they add value inside that cycle will change. After all, if Tesla can use ML to teach a car to drive and SpaceX can programme its Falcon9 rocket to fly and land itself, what is the magnitude of difficulty of those tasks in terms of input data compared to many DBA functions? I’m not talking about deep technical troubleshooting, but many of the tasks we still see people doing; such as: install and database builds, performing generic space management tasks, even application of OS and database patches.
The same is true of service providers and partners (including the one I work for and write this as an employee of), with the days of providing “core DBA” services now having a finite horizon. Our job as IT architects, technicians and service providers will be to help our customers exploit the value of new Platform-as-a-Service (PaaS) and autonomous services to reduce cost and complexity, both of acquisition and on-going management, as well as learning to derive value and exploit the data with these new platforms. After all, as CPU power has exponentially increased it’s not as if we haven’t found ways to consume our new-found resource capabilities.
Increasingly I suspect we will see the real value of cloud for most customers in the adoption of Platform Services (PaaS) into which the Oracle Autonomous Database falls squarely. Indeed, it’s my view that autonomous is a key function of any true PaaS. Infrastructure-as-a-Service (IaaS) - whilst an enabler of cost efficiencies - is probably similar in vain to the initial wave of virtualization where cost optimization drives an initial surge, followed by the real value becoming evident through additional services such as snapshotting, rapid provisioning and automated scaling. IaaS is now simply “table stakes” (the new wording of the moment it seems!) for any cloud provider, but a race in which hardware technology exploitation and economies of scale are reaching the point of diminishing marginal returns. Vendors will now look to differentiate based on the unique Intellectual Property (IP) they can bring to the game; namely the software under their control. Whether this is AWS offering out services such as Dynamo (originally designed for their own usage) or Oracle with a rich heritage of producing enterprise software, it is these unique elements offered as a full managed PaaS solution that will allow vendors to offer additional value to clients and distinguish themselves from their competition. After all, if you had Oracle Intellectual Property - what would you do?
I also believe that Agile and DevOps practices are already having impact on “traditional” RDBMS usage and driving last numbers of PaaS use cases that will start to be moved into production over the next period. This is probably especially true where IT organizations are slow to provision components like databases (we’ve all been in companies where it’s “2 weeks” to get a database stood up). Now, I’m not arguing this should ever be the case, database creation should be easy and automated, especially with technologies like Oracle Multi Tenant option available, just that IT often made this the case through internal inefficiencies; lack of resources (both hardware and biological), licencing constraints or change in management issues. Service providers in particular need to look at themselves here, if organisations are unable to quickly request and begin to consume new services due to contractual issues I would suggest their customers will rapidly outgrow them!
It’s likewise easy to see how cloud vendors are focusing on the DevOps containerized deployments, putting the emphasis firmly on the DevOps side of things to acquire new clients and projects. Most of these DevOps technologies don’t necessarily put the enterprise grade persistence we in the Oracle community have been used to, at the core of their technologies (they weren’t initially designed to address that need), and it suits a cloud vendor to resolve the issue of persistence in containers via driving consumption of their PaaS database services.
I’m pretty sure some of you reading this will feel I’ve “drunk the Oracle Kool-aid”, but I’d say you’re missing the point. Autonomous Database 2018 might not be the holy grail, all singing all dancing version, but the one in 2019 and 2020 when usage data has allowed the ML model to be refined? What about 2021, 2022 etc.? The robots are coming. It doesn’t mean we’re all out of a job, it just means that jobs are going to change. The IT industry itself was seen as replacing workers but has created many more jobs than it’s ever taken away. My hope is that people see this as the start of something more interesting, taking us away from keeping the lights on and into a place where we can point at things we’ve done, improvements we’ve made and the difference our roles create.
So, am I just being contentious? What are your thoughts? Am I seeing things over simplistically? Or is IT about to go through the automation revolution that has already been seen in so many other sectors? Let me know your thoughts in the comments section below or tweeting us @dataintensity.
I will also be speaking on Oracle’s upcoming live webinar ‘Create a data warehouse in seconds’ on Thursday 24th May 2018 at 11:00AM Central European Summer Time. You can find out more and register for the webinar here.
Finally, if you want to try out the Autonomous Data Warehouse Cloud service please get in touch via email at firstname.lastname@example.org - we’re always happy to accommodate a good old Proof Of Concept!