Robotic Process Automation (RPA) offers the opportunity for significant costs-savings across an organisation’s work force. The take-up has increased as robots, mimicking defined human roles, have extended to fully-automated end-to-end processes.

Businesses like Automation Anywhere, UiPath and, in the UK, Blue Prism, demonstrate the explosion in the market place with the AI office market predicted to be US£48.5bn by 2022 (Forrester).

What’s RPA from Oracle?

The work output of software robots is extensive and, of course, 24/7 so the business case is obvious.  But how do the introduction of software robots, with user interfaces, or, more ambitiously, full RPA impact on Oracle licensing? Certainly AI and robotics are at the forefront of Oracle’s current vision with CEO Mark Hurd declaring earlier this year that more than half of enterprise data will be managed autonomously by 2020.

The challenges with Oracle (as ever) are its opaque licensing terms, developed iteratively over 40 years and never quite tracking to contemporary usage. This results in ambiguities in definitions and terms compounded by Oracle often maintaining that it is the only arbiter of its licensing rules.

Result

Oracle audits inevitably throw out unexpected results with high demands for license fees and back-support even where the IT function was convinced that the business was fully licensed.

Oracle has over 50 license metrics for its database technology, applications and middleware. None however refer to robots or reference robotic process automation.

This is the moment when alarm bells should start ringing. The world’s largest brewer, Diageo, was caught up in an uncomfortable GB£60m High Court claim from SAP last year: a central issue was whether SAP could demand license fees when there was no applicable license metric for Diageo’s usage of iPads by third party sales reps to access (indirectly) SAP programs. This uncertainty could readily be paralleled with Oracle licensing.

Oracle database technology at enterprise level is largely licensed on a processor-basis or as ‘named user plus’.

Processor licensing is subject to an overlay imposed by Oracle with its core-factor table and partitioning policy. Neither are legally binding but are essential for Oracle to persuade users as to the veracity of their software license reviews in virtualised environments.

Its demands are (notoriously) based on ‘all processors where the Oracle programs are installed and/or running’ unless partitioned using Oracle’s own partitioning option.

There are many instances of chilling demands being precipitated by virtualisation either on-premise or in a shared services environment. These can be challenged but there needs to be implacable legal argument and absolute confidence to ensure that Oracle back down where processors that might conceivably be available for use are never in fact hosts of the relevant programs.

Advantage of RPA

One advantage of RPA is that processes can be scheduled at times of low demand and accordingly the numbers of virtual machines accessed (or accessible) can be tightly restricted. A close examination of this – and either use of hard partitioning or effective DRS affinity rules – can reduce licensing fees substantially.

The principal alternative metric for the database technology is Named User Plus. This is defined as ‘an individual authorized by you to use the programs which are installed on a single server or multiple servers, regardless of whether the individual is actively using the programs at any given time. A non human operated device will be counted as a named user plus in addition to all individuals authorized to use the programs, if such devices can access the programs’.

It’s critical to note therefore that such ‘user’ can be a device (not defined) and so could be a robot accessing via a user interface. When licensing on an NUP basis, regard must be had to user minimums per processor.

For Oracle’s E-Business suite applications, the most common metric is that of the Application User.  But here, the definition is as ‘an individual authorized by you to use the applicable licensed application programs …’. So, non-humans / devices are not accommodated nor can they be licensed under this metric. One needs to go further downstream to identify which individuals are controlling the robots and so need to be licensed.

Oracle has been vociferous in promoting AI and robot usage – but largely in the context of its own cloud offerings. What is never clear is the licensing impact where the use is on-premise or via an outsourcing partner or hosted on AWS.  It is here that any business case for automated usage can be derailed if the consequential licensing costs are ignored – at least, until the Oracle license management division initiates an audit.

Given the massive disparity between discounts readily available to corporate and public-sector customers when making new purchases as compared to in an audit situation, any organisation should confront these issues early. Prospective licensing costs – and whether some astute configuration could reduce these – must be central to any decision to automate.

Robin Fry

Cerno Professional Services