In my previous articles, I’ve looked at how company teams can start using BI and then progress their maturity and approach further. This time, I’ll cover the sticky issue of migration.
What is the challenge?
No company sticks with the same technology forever. Even the most useful pieces of software will be updated, upgraded and changed to keep pace with business demands and customer requirements.
However, BI implementations are in a category of their own when it comes to change. For many companies, their BI services are fundamental to their success and their ability to manage operations. Making changes to these critical systems can seem like a recipe for heartache, even when the value from doing so can be huge.
However, change is coming in the BI market. Traditional tools from the likes of Oracle, IBM and SAP BusinessObjects have performed sterling work within enterprises, but they can represent a significant ongoing cost, as well. While the reports that they produce are important to the business, the platforms are inflexible and have failed to keep up with evolving business user needs for speed and agility.
While these products work, they are often not receiving any updates beyond those for security purposes. Rather than providing platforms that can support new centralised and decentralised uses of data within the business, these implementations have ossified under the weight of how businesses rely on them. Rather than changing and keeping pace, they have become frozen in time.
For CIOs in this situation, the challenge is stark. While these systems work – and businesses don’t want to throw away their significant investments in them – they can’t support the new analytic workloads that users require. At the same time, new use cases for big data and self-service analytic support are springing up across the business, and the rapid growth in data volumes is not suited to these existing platforms either. The mantra of being “data-driven” encourages all companies to make more use of their data, but this is not possible with legacy BI tools in place.
Planning a migration – fast or slow?
How and when companies migrate their data and analytics implementations requires some careful thought. Legacy BI tools play a vital role within business operations, yet they don’t meet the needs for agility that new services can provide.
Shifting to a new BI strategy can take one of two forms. The first option is an augmentation approach. This normally involves running multiple BI tools in parallel with each other, so that the existing BI implementation produces its current reports, while any new data projects are implemented on the new BI platform. This approach is less risky for the IT team, but it does involve more expense when running two platforms in parallel. The other consideration is how long the legacy platform will remain in place. It can be expensive to support traditional BI tools in terms of large staffing levels with the appropriate skills on top of the software licenses and maintenance fees.
The alternative here is to “rip and replace” – the emphasis for this is to move over to the new platform sooner rather than later. While there will be an element of running two BI tools alongside each other for a time, this approach involves more preparation for making the move. Modern BI solutions built on multi-tenant cloud architectures can accelerate and simplify the process of replacing the incumbent BI platform. In any case, it’s important to look at how to move data from the existing BI implementation and into the new one.
There are two elements to any migration – the data itself, and the relationships between those data sets. It is possible to look at “lifting” existing data into the Cloud, but with the current volume and variety of data, it’s important to consider the architecture for the data and decide what gets migrated into the Cloud and what remains where it is.
Rather than simply looking at the existing, legacy architecture and then recreating that in the Cloud, it is worth considering how this can be extended further. For example, the existing data’s logical framework can be used to create a semantic layer that is made available to users. This empowers users to interact with the data in business terms and do so in a trusted and consistent way.
This can also encourage more thinking about how the data is used in context. When reports were prepared for very specific purposes, it becomes difficult for other employees to use them. Instead, employees must be able to use data from those central sources in ways that suit them. This can be achieved by matching each department’s or individual’s objectives against the data that they need and then modelling how changes in performance would create better outcomes for the business. The focus here should be on value.
Thinking ahead about data
In the context of a migration, this can provide an opportunity to rethink how reports are used within the business. While traditional reports can be recreated exactly – and there is a comfort factor from doing so – this cloning of existing reports should be avoided. Instead, this is an opportunity to ask management leaders, business departments and individuals how they would all use this data if they could. Following on from any initial deployment, this ability to interpret data in new ways from trusted sources, or combine it with other internal and external data sets, represents the greatest potential opportunity for the future.
This planning is not just about the technology side. It’s important to look at what future results the company is looking to achieve from its use of analytics, as this will guide some of the choices around migration. By looking at how business processes can be affected through use of data – for example, looking at the customer journey from initial interest, to sale completion and payment – each team involved in that process can then evaluate its performance and what they can change to have the most impact.
Companies are producing more data than ever before and employees want to make more use of information. Traditional BI tools served their purpose in the era of centralised analytics, but getting data into the hands of everyone across the business in a way that’s fast, economical and scalable requires a new approach.
When companies look at migrating away from their legacy BI platforms, the speed of that migration has to balance the value that can be delivered to the business against the level of risk with which the CIO is comfortable. However, the need to move is not going to go away.