Open Banking, which came about as a result of the result of the PSD2 rules around opening customer data, promises customers a better service experience. The concept has spread and is now a global trend, albeit in different guises. But as night follows day, Open Banking, with its emphasis on open source technologies, brings challenges.
As financial firms position themselves for success in this rapidly evolving environment, open source technologies play a key role in industry organizations. According to Axel Winter, former global head of enterprise architecture at Standard Chartered Bank in Singapore and currently CTO at Thai business conglomerate Central Group. He says:
Open source is something [banks have] used – unknowingly – for a very long time and today I guess at least 30 percent of their software would be open source. Indirectly, there might be more, through vendors. A lot of their development tools and frameworks end up being open source and that in turn makes it an enabler for open banking.
One might be forgiven for thinking that open source software naturally matches the move to Open Banking interfaces. However, decisions around the delivery of these new models have been driven by pragmatism, according to Martin Percival, a principal solutions architect at Red Hat.
Open source code is constantly seen at the forefront of innovation in the software world and has long since left behind the accusations from more proprietary vendors that it is in some way inferior to those offerings. With the growth of big data, cloud computing, the Internet of Things and Artificial Intelligence, business leaders have realized that open source software should be an integral part of future IT planning, and Open Banking is exactly the kind of innovation catalyst that demands new thinking.
Achieving the reference architecture for Open Banking standards is a multidimensional challenge, which includes handling not just vast data volumes but also variety and the need to draw insights from fast-moving data streams by combining them with years of historical data.
For the data layer, financial services institutions have used open source platforms like Hadoop which ingests and works with all types of information as well as performing multiple types of processing on a given data set. This, in turn, enables fast creation, testing, and deployment of closed-loop analytics.
Financial services is the largest market segment at analytics firm SAS, who has been working with banks and insurance companies for almost all of its 42-year existence. According to the firm’s financial services marketing head David Wallace, there has been an increase in adoption of open source in the sector, Hadoop being the data repository of choice. He says:
We have definitely seen a rise in use of open source and have been working with the Hadoop distributions for the last few years. We can store and extract data from Hadoop and our analytics platform software executes within the Hadoop ecosystem to provide more scalability to financial services clients.
According to Wallace, SAS embraced open source analytics software and extended that to be able to deliver a high-scale execution environment, while also ensuring the governance framework that financial services firms typically need to address regulatory demands.
Open source definitely has a place in financial services. But tools in that sector have to be constantly validated for efficacy and also monitored to make sure that they are delivering what the original developers were expecting – such as documenting data lineage, models, the decay of these models – within the framework of a regulatory environment.
Open source is not without challenges says Wallace:
One of the limitations that we see is lack of scalability in many of the open source capabilities. That’s because they’re written so that when they execute, they’re expecting a certain amount of data to be available but the scale of data that needs to be processed by large financial services firms, where they’re ingesting millions of transactions every day, is huge.
According to Wallace, another challenge that incumbents face is related to the complexity in integrating systems required to cope with all stages of analytics, from the ingest stage through to cleansing and visualization all the way through the scoring and application of models to put data into production:
I think that open source has generally done a good job, but [innovation has been limited to] single machines for the most part, so they’re limited to the number of CPUs, the number of clock ticks that are ultimately available on an individual machine. And because of the volumes of data involved and the complexity of the calculations, we found that doing in-memory distributed computing across sets of machines is a much better computational pattern for being able to scale to the needs of a financial institution.
While some suppliers imply that open source tools need to be bundled into proprietary software to meet companies’ data-related demands for scale, Martin James, regional vice president for Northern Europe at data management vendor DataStax suggests a mixed approach:
We see banks facing a big challenge around deploying systems that can cope with the potential increase in transactions that will come with open banking, and these new technologies are essential here. Without this level of insight, banks won’t be able to deliver the right levels of customer experience in the new, ‘right now’ economy.
But there’s no single stack available – open source or proprietary – that supports Open Banking. Instead, decision-makers have to select the right mix of technologies that can support their organization’s vision of Open Banking. In that context, open source technologies are positioned as instrumental in delivering better customer support and experience, according to James:
Open source tools can be very good for driving innovation which can be used to complement the existing technologies within banks. But open source on its own is not going to solve the problem of scaling up.
James points out:
Deploying at enterprise scale requires more support and service than open source projects tend to be able to supply on their own. So it’s about deploying the best possible, enterprise-ready technologies in the most appropriate ways to meet those production-level requirements.
However, demands prompted by a blended systems architecture present skills challenges. That’s because such projects require specific skills that are not always available in the market, as some of the open source projects don’t have large communities around them. Based on his own experience dealing with banks, James comments on the usual hurdles users run into:
We see companies looking at public cloud deployments and managed cloud services to fill those skills gaps, so they can make use of talent to run their infrastructure more efficiently.
We have also seen organizations struggling to understand which code release and associated patches should be implemented. An active community can drive a lot of irregularly scheduled codes change – that can be difficult for organizations to understand how to fit into their own change cycles.
Fintechs, on the other hand, don’t have difficulty finding people skilled in working on open source projects. One example is Swedish personal finance management startup Tink. The head of engineering at the company, Andreas Lundgren, says:
Almost everyone joining us has a big interest in the open source world and its developments. Keeping up with developments and knowing how to assemble them into the product you are building is a really important part of becoming a great developer.
According to Central Group’s Winter, IT leaders at incumbent organizations should embrace and participate in open source – even if it’s only to gain a deeper understanding of the model and how the community can support them. Winter says:
Participating and contributing to open source would be a sign of a strong and mature IT department. If [banks] lackskills in open technologies, they aren’t able to staff a proper IT department and sustain themselves in the digital area in the first place. No Hadoop, Spark, Big Data, AI, analytics skills = no bank.
As with so many technology shifts, IT organizations are wary of change. This is particularly acute in the banking sector which is focused on risk management. For those firms, open source is still perceived as a kind of Wild West environment. Some firms view open source software as inadequate to deal with security, compliance, and scalability. That’s despite the fact that tech giants like Google, Facebook, and Amazon are both heavy users and publishers of open source technologies that scale at orders of magnitude than any single bank is likely to require.
Established vendors often use open source technology for their own products, but also provide a proprietary front-end environment, so no open source development needs to take place outside of their tooling. This reinforces the idea that there is not much point in investing in open source skills, though the new realities of Open Banking is changing those perceptions.
Open source evolution in an Open Banking context holds the promise of competitive advantages, but incumbents need to support skills development and community growth. Managed cloud offerings could be an example of an initiative that would help free IT bandwidth to improve their open source skills. But I wonder to what extent this is a real priority? Winter could be right in saying that buying tools off-the-shelf would still get banks open source in the end, but would also result in less engaged and knowledgeable teams.
Image credit - © kebox - Fotolia.com