EVP and Head of Platforms, Nordea
Data to Decisions
Nordea is a Financial Services company with over 10 million personal customers and more than half a million corporate customers in the Nordic region. Nordea is currently represented in 16 countries throughout the world, operating through a number of full-service branches, subsidiaries and representative offices. Nordea is one of the 28 globally systemically important banks (GSIB) that exist worldwide.
Nordea manages a complex and distributed data matrix across multiple business lines and jurisdictions. Increase in regulatory demand, velocity of output and consistency of data combined with a need to decrease opex, capex and time to market apply continued pressure on data management activities. Managing this data through manual processes is time consuming, labor intensive and doesn’t offer any economies of scale.
Historically, there was no single repository Nordea could look at that contained information about the specifics on what data was in each system and what was the provenance of that data. However, data regulations like the BCBS 239, sometimes referred to as the “Principles for effective risk data aggregation and risk reporting,” Nordea were raising the bar on data management processes.
The problem was that with the existing legacy systems, which used a combination of mainframe, relational, ETL and classic data warehousing technologies, responding to the demands of regulators in a timely fashion was both a challenge and cost prohibitive.
The Nordea strategy is to move to a combination of Hadoop/Spark as the new data store and analytics environment in conjunction with implementing a smart data catalog and a native Hadoop business intelligence solution. Additionally, traditional data warehouses often require schema changes which make them slow and expensive to implement as new data sources become available. By moving to a Hadoop architecture, Nordea will be able to respond quickly to the addition of new data sources and new requests from regulators thanks to the flexibility provided by schema on read within the Hadoop world
Selection of technology for the new solution was driven by a desire for Hadoop/spark native solutions across the architecture stack. Competitive vendors in data cataloging, wrangling, protection and business intelligence Nordea re all evaluated on their ability to run natively on Hadoop plus their ability to integrate with each other as part of a best in class deployment.
Treating data as a strategic asset was enabled through a simple mantra: Cheaper, Faster, Better & Secure.
Cheaper: Move from proprietary hardware and storage to commodity storage targeting north of 70% savings.
Faster: Increase the velocity of change and reduce time to market for data insight combined with real time data sources and insights.
Better: Automate much of the development and operational processes by automating much of the work done by humans today. But Nordea didn’t want to just replace like for like. Nordea wanted to build a platform that guaranteed the quality of the data that it produced, increased accuracy and that delivered full traceability will built into the system. Nordea need to do far more with the data Nordea have whether it is for managing exposure and risk, or for leveraging it for the good of the balance sheet and the customer.
Secure: Lastly, Nordea needed to guarantee that the right people have the right access to the data in a secure manner. There is an inherent risk in creating a centralized Hadoop store in that it creates a “honey pot”. Nordea needed to deploy our platform in such a way that Nordea could dial up or down security and access as appropriate.
Nordea now spend much less time looking for data and more time actually using it. The time it takes for Nordea to respond to regulatory requests is much more rapid than it was in the past.
- Nordea's annual spend on OpEx in this space is EUR20M, CapEx EUR30M and rising
- Time to find data for use in regulatory reporting before implementation: Months
- Time to find data for use in regulatory reporting after: Days
- Time to integrate/wrangle data for use in regulatory reporting before:15 days
- Time to integrate/wrangle data for use in regulatory reporting after: 1 day
- Liquidity processing time from 2 days down to 180 minutes
- Data ingest end to end went from 16 weeks to 2 weeks
- 70% storage savings per terabyte of data expected over the next few years.
- 73% decrease in platform costs is projected. High cost of manual processes for finding data, profiling it, building transformation logic and data duplication has been replaced by a low cost process of data sourcing, discovery, fingerprinting, profiling, streamlining of delivery, centralization and simplification of transformation and reduction in data duplication.
The overall solution stack is:
- Cloudera for Hadoop distribution
- Waterline Data for data discovery, fingerprinting, tagging and cataloging of data.
- Trifacta for data wrangling
- Privitar for data protection (integrated with the automated tagging from Waterline Data)
- Arcadia data for business intelligence directly into the Hadoop data lake
Nordea has taken an approach of implementing a best of bread set of technologies integrated via API sets.
Nordea is now able to work much interactively with data. The old way of doing things, waiting to find the data needed for a project, waiting for a data engineer to implement ETL to integrate and manipulate data to make it useful, etc has been turned on its head. Nordea are delivering true self service to its internal users up and down the technology stack ranging from data discovery, fingerprinting, cataloging, wrangling and business intelligence.
This means Nordea gets productive value out of new and old data sources much more quickly. They are no longer restricted by the sort of “big-bang” approach to data management. Once the have a data set, they can immediately start to work on it and get on with the quick delivery of business value.
A second area of disruption is the use of investment in a data management platform that provides both better data governance AND more agile analytics. Those two concepts are usually at odds with each other. But in the Nordea environment, better data governance means better documentation of data and data lineage through automated discovery and tagging of data with the ability to make that data quickly and securely available for use. Because much of this process is automated, it also leads to making data easier to find and use resulting in better time to analytic insight as well as the ability to enable a greater level of business self-service.
Nordea’s implementation addresses both “defensive” aspects of data management, making it easier & faster to respond to regulatory requests, while also making it easier to utilize data in support of “offensive” uses for self-service analytics & digital transformation. The combination of data governance for risk reduction & self-service for greater business agility from a single strategic investment is quite unusual.