November 29, 2022
NAB has spent much of the past year standing up a new strategic data platform called 'Ada' that is considered “chapter two” of the bank’s data journey. At the core of Ada is a new data stack comprising technology from Databricks, HVR Fivetran, PowerBI, AWS and Azure. It is understood this is likely to be…

NAB has spent much of the past year standing up a new strategic data platform called ‘Ada’ that is considered “chapter two” of the bank’s data journey.

At the core of Ada is a new data stack comprising technology from Databricks, HVR Fivetran, PowerBI, AWS and Azure.

It is understood this is likely to be further augmented with low-code or no-code tools for analytics and business user-driven insights.

Ada will support all of NAB’s major data workloads, “from real-time analytics and data science to AI-driven recommendation as well as supporting risk management activities and financial crime and regulatory reporting.”

The bank has given itself a timeframe of “approximately five years” to fully transition all its analytics operations to run off the Ada stack.

This aligns with the bank’s new technology strategy, which has data as one of eight strategic pillars.

The strategic objective is to make “data like electricity” – always-on and consumable for teams across the bank.

“The majority of Australians assume that electricity is available when they need it and that it will be delivered safely and at the right standard,” NAB chief data and analytics officer Christian Nelissen said.

“Data at NAB should be the same.”

Ada first broke cover a fortnight ago via a series of LinkedIn posts from NAB executives that referenced an internal launch event.

The new platform will enable NAB to simplify its data architecture by unifying analytics, data science and machine learning into the one stack.

“You can think about [Ada] as ‘chapter two’,” chief technology and operations officer Patrick Wright told iTnews.

“A lot of the technology that we’re using to build Ada wasn’t available, or at least certainly wasn’t scalable and tested, five years ago. 

“Our belief now is that it’s time for us to step into the next generation of capabilities.”

NAB’s existing data lake – the NAB Data Hub or NDH, which is based on an AWS stack – will, “over time, “be deprecated into Ada”, Wright said.

It’s not just NDH that will ultimately be folded into Ada; Nelissen told iTnews that so far, the bank has “delivered the first three data sources onto the platform”, and has one live use case.

“From here, our plan is to deliver all of the core data sources in the bank onto the platform and we have started decommissioning our legacy platforms,” he said in a statement to iTnews.

“The others will follow once Ada is complete.”

Asked why the bank had chosen to re-architect, rather than remain on an AWS-made stack, Wright said it was important not to think about the data plarform – or indeed, any technology platform at NAB – as static.

“When we think about technology, it will always change,” he said.

 When I started in this space, large companies would buy a particular platform and expect it to last for 20 or 25 years. 

“In today’s modern world, the half-life of technology is 18 months, and so if you’re not willing to step into and continuously update and adjust your thinking about how you approach not only these problems but frankly the technologies that underpin solving them, I think you’re going to find yourself outdated quite quickly. 

“So we see [Ada] as a continuous evolution, and I’m certain in four or five years, we’ll look at Ada and wonder what would be the next generation or the next version of that, and obviously we’ll step into it when the time comes.”

The use of Databricks as the “lakehouse” – part data lake, part data warehouse – technology in Ada brought several advantages for the bank, according to Nelissen.

“As an example, Databricks has a component called Unity Catalogue which enables us to integrate the metadata (in particular, flags for personal information, sensitive personal information and jurisdiction) with data access controls rather than layering those controls on top of a data store, with the result that we have greater assurance around who has access to what data,” he said.

“Databricks also gives us easy access to lineage for our critical data elements rather than needing to build those separately.

“From a security perspective, there are inherent data controls built into the platform [with] built-in compliance to standards such as PCI DSS. 

“The platform also allows an in-built method to abstract all PII [personally-identifiable information] from any user, very simply.” 

NAB’s group chief digital, data and analytics officer Angela Mentis, said Ada would help the bank to realise its ambition “to make digital human”.

“We need to power those experiences with data,” she said.

Source