In writing my last couple of books, Silicon Collar and SAP Nation, I have observed how hypocritical the IT industry has been about automation. AI, robotics, drones are good for customers but not in reducing tech labor? The worst offenders are outsourcers with their huge labor pools. Software vendors have been skittish about cannibalizing labor at these partners – for SAP Nation, the economic models I built showed a majority of costs in labor. Beyond economics, machines do dull, dirty, dangerous tasks much better than humans and we hate to admit it but tech does have a lot of repetitive work which would be better automated.
So, it is good to see Oracle announce at OpenWorld an autonomous database (version 18c) and semi-autonomous cyber security services.
Larry Ellison was at his professorial best as he explained concepts of machine learning and the promise of training computers with masses of data and gobs of computing power. The data center and IT/telecom infrastructure generates so much log data as Larry showed in one of his slides (click to enlarge)
And when you look at the trail from end user to network node to application accessed to database, you visualize the “big data” of potential security failures every minute in every organization.
Let's leverage this data to teach computers to provision, upgrade, tune, backup, patch themselves with minimal downtime and much more efficiently than humans can.
He used a carrot and stick approach.The carrot comes in the Oracle offer to run databases cheaper than even at Amazon and to do it with better SLAs. He demoed several workloads simulating customers in financial services, market research, insurance and retail. In each Oracle was faster and cheaper than Amazon’s performance. As he said early and often with sarcasm – you can only have it if you are willing to pay less for it
The stick came from repeated mentions of the shocking state of unpatched systems in IT world and the massive breach at Equifax and others like the Federal Office of Personnel Management. Humans do not do a good job patching systems, especially when they are mandated to reduce downtime. Why are we fighting the bad guys’ computer power with just our people power?
Makes a lot of sense, but the autonomy will only be available in the Oracle Public Cloud or the Cloud at Customer (a managed service from Oracle). I can see their reasoning – if Oracle is going to provide a SLA of 99.995% up-time it wants to control tightly the environment.
Personally, I would like to see the autonomous database also available on-premise or as an appliance (minus, of course, the SLA and the cloud pricing model). Here’s 3 reasons why:
a) Customers are extremely comfortable with Oracle’s database prowess, much less Oracle’s cloud infrastructure competency. In the June 2017 Gartner IaaS Magic Quadrant, Oracle was in the bottom right as a Visionary – way behind Amazon, Microsoft, Google, IBM and Alibaba. In 2-3 years, the comfort level will be much higher, so why not let customers get familiar with 18c in the meantime in their shops? Oracle could justify the customer investment by showing the savings in DBA labor and that of managed services, rather than what came across as Amazon bashing as Larry ran version after version of his demo workloads and showed savings with Oracle that way.
b) Global companies are already struggling with changing data residence requirements around the world. A database in the cloud will require much more due diligence on their part.
c) Many customers are loathe to increase the lock-in large vendors like Oracle, SAP and IBM already have. SAP is trying to sell them databases, Oracle is now trying to sell them infrastructure services in addition. May seem trivial from a vendor point of view, but it is not from the CIO's because of bad behavior of large vendors with the lock-in they already have.
So, Oracle Autonomy is a giant leap forward for the industry. But it may need an even more giant leap of faith for customers. At least, in the near future.