Updated: Jan 28
If you are past 35 years old, you have been taught that the proliferation of applications that overlap or, worse, do the same thing is a sure fire way to increase costs. So naturally, it is a rule of thumb that IT executives go by - decommissioning as they modernize. Rightly so.
But judgment needs to come into play. Case in point: I was at a client's a few years back. Because of a CIO's decision to make a stand, the organization changed its ETL platform from DataStage to Informatica. Over three years, they would save money. But the cost of migrating the existing ETL's was not part of the calculation (how worse can this decision get, you ask; yeah). That cost overtook the savings eventually.
But what struck me was the opportunity cost of placing good people on a technology migration project that brought nothing except instability to the business! Think about this for a moment.
Of course, saving one of the two recurring licenses of software and not having to handle multiple skill sets offsets that a little, but we should question whether this is an acceptable cost, for a while at least, in order to capture other benefits.
When in a situation of rapid change and innovation, I claim that it is.
Technical debt is not new. And when looking at data analytics systems, we must think DATA first. It is different than with applications. An ETL tool is just that. The real business solution is the data that has been prepared and serves the business. The tool is probably the smallest cost component of the full solution in this case.
Here is a new approach for organizations looking to modernize their data architecture and how they work with data so they can deliver value faster.
First, don't rebuild what already works. Leave it there. Maintain it. Put your good people on the NEW projects like modernizing your data environment, with the new technologies. At some point, maybe a source system will change that will require significant rework; migrate the system then, when it adds value. Perhaps that data source will go away and be replaced entirely at some point.
Second, focus on facilitating tasks staff cannot do well within the confines of existing business intelligence, analytics, and data preparation systems.
Third, in the same vein of thinking, don't move everything to the cloud. If it works on-prem, keep it there for now for the same reasons as above. Only move when it adds a lot of value. If you are an agile organization, your product backlog will tell you when it makes sense to do it, beautiful, no?
People are thirsty to be able to explore, analyze, and visualize new and diverse types of data. They are frustrated by limitations of old tools and how long IT takes to deliver, so self-service, not just access but data prep as well, can crowdsource some of the efforts and let some of the users do more on their own.
This is where the efforts should be focused as opposed to migrations for the sake of saving a few dollars. The value that will be derived is exponential. It is a simple change that, with the right communication and expectation settings, will raise customer satisfaction and optimism in your organization's journey to becoming data-driven.