Maintaining your big data analytics software

When a company buys or subscribes to an analytics solution, keeping the work current can be a challenge. But maintaining that software is key to helping it enjoy a long, useful life.

For example, if you use analytics to assess financial portfolio risk factors brought on by climate change, political unrest, etc., risks will continue to rise and decline. Analytics software has to be adjusted to account for these variations. When these adjustments are made, it can affect the other systems and business operations that engage with the software.

This is where IT (and the company) should have a soft maintenance process defined for big data and analytics, just as it has a software maintenance function in place for transactional systems and systems of record.

Unfortunately, how to maintain big data analytics software in production is still a work in progress for many companies.

In some cases, maintenance of these programs isn’t emphasized because organizations already feel they have it covered in the process of developing analytics. They know that as they’re testing their software, they’re going to keep changing data and algorithms until they reach a point where the analytics are producing the insights the company expects. This is also when you place the software into production to support decision making and other business processes.

But what if the analytics and the big data running in production can no longer provide adequate insights because of changes in business or the world? This is where “old fashioned” software maintenance enters in—and where IT has to play an important role.

When IT should perform maintenance on analytics software

There are several situations that should activate IT’s involvement in software maintenance for big data and analytics.

System integration and downstream processes

Big data analytics software is like any other type of software. If you make a change in it and the software interacts with other systems, the systems that it touches must also be changed where required and retested before a revision goes into production. This is the work of an IT software maintenance group—even if most of the company’s analytics are developed by a separate data science group.

Security and governance changes

If a change in big data analytics software triggers potential changes to surrounding software governance and security checks, these impacts have to be vetted by regulatory and IT compliance teams and the IT security group first for possible impact to systems and business operations. Any changes that surface from this research should be administered by the IT software maintenance team.

Business process integration

If analytics programs are integrated into business processes (e.g., a loan decisioning software that a lending department underwriter uses to determine the credit worthiness of a loan applicant), IT should be engaged with end users (and potentially the data science team) to ensure the new system revisions continue to work with existing business processes. If there is a change that impacts an existing business process, the work becomes modifying either the software or the business process so the two are aligned before any new change goes live.

Who does the IT software support work?

An IT software maintenance team should be defined to support any software code changes and testing to existing systems that the company runs. IT is likely to feel the most impact in big data analytics support from the extra work this will probably place on IT business analysts. These analysts are the ones most likely to interface with end users and the data science and IT maintenance teams to assure that change management occurs for the  analytics software, the existing software and systems that the company runs, and with the business processes the company has in place.

Two steps to take now

1. A software support maintenance team should be designated for big data analytics, if it hasn’t already.

2. Plan for additional staff in the business analyst group, as this group is best positioned to serve as project leads and go-betweens for IT, business users, and data scientists.