- About Us
- Infosys Knowledge Institute
Cloud computing has fueled rapid growth and adoption of innovative digital business models. From Uber to Spotify and Netflix to Slack, thousands of digital leaders have built their businesses on cloud software. This is because it enables flexible, scalable, cost-efficient and continuously upgradable capabilities. But for banks, adopting cloud technologies can be a real challenge — not least because mainframes continue to be at the core of their business, given their reliability, almost impregnable nature and ability to manage large volumes of complex tasks. In fact, over 90% of the world’s top banks still rely on mainframes.1 This is despite the fact that mainframes are archaic in nature and present a host of issues.
1. Legacy technology
According to an Infosys Knowledge Institute survey, legacy systems currently rank as the third most commonly cited barrier to digital transformation (named by 42% of respondents), but financial services executives expect that they could become the most serious barrier in 2019.2
Millions are spent every year to maintain these archaic systems,3 which are often incompatible with modern technology. These systems act as a barrier to better catering to the needs of the digital customer. The reliance on legacy technology also stops banks from taking advantage of Agile and DevOps ways of working, increased automation, and analytics. According to Gartner estimates, banks need to triple their digital business innovation budgets to modernize legacy applications through 2020.4
Experts say that banks spend 80% of their IT budgets on legacy technology maintenance, and a Tier 1 bank could spend up to $300 million a year to update existing software to meet regulatory requirements.4 But cloud can act as a workaround in reducing costs.6 For example, the Federal Home Loan Bank of Chicago reduced infrastructure costs by 30% by moving all of its internal production workloads to the cloud.7
2. Data management
With data growing by 2,500 petabytes a day,8 data management is increasingly becoming a concern for banks,9 not least because of regulatory and security requirements. Regulators require banks to provide detailed reports, additional information on stress testing and information at a granular level.10 Banks continue to capture data in multiple forms (customer personal information, transaction history, journey maps, market data), and this huge volume of data is now stored in data warehouses and lakes.
Banks store data in various locations, and it’s often accessed by different users, creating a siloed approach and multiple layers of data duplication.11 Singular data warehouses and lakes are formed, giving rise to further issues of storage, meeting regulatory requirements, authenticity, incompatible formats and higher time to compute. These have an indirect bearing on run-the-bank costs.
The data collected by banks must also be analyzed and deployed to facilitate better decisions.12 However, due to legacy technology, banks are unable to leverage access to this data for analytics and insights and improve customer experiences.
3. Manual data migration
Currently, when a bank decides to move data to a new system — whether it’s in the cloud or on-premises — a new project team is set up and works in a silo. Data is migrated from the data lake into the new system, or a new connection is created between them. Because of the silos in the bank, other teams using the data may be unaware of this migration or new connection.
In many instances, multiple teams want the same data and follow a similar process without getting connected to each other. This leads to duplication of data, with multiple copies created in the data lake and transported to different systems either in the cloud or internally.
Banks’ manual and siloed approach of ingesting and migrating data is time-consuming. It is also counterproductive, as in this digital age, customers’ demand for instant, real-time, and personalized products and services has become the new normal.13
4. Lower number of legacy coders
Legacy systems are built on outdated languages such as COBOL. For instance, the U.S. financial sector has over 200 billion lines of COBOL code currently active, and COBOL powers over 90% of ATMs.14 Despite COBOL’s widespread use, today’s coders prefer to use new languages that are compatible with artificial intelligence, machine learning or cloud computing. Few people want to learn a language that can talk only to legacy technologies, and coders familiar with COBOL can be well into their 50s or 60s,15 presenting significant skills shortage challenges in maintaining older technologies.16 Although upgrades are available, they are still not fit enough to compete or converge with systems of this digital era.
5. Security concerns
Customer data in any form is sensitive for banks. The major cloud vendors, such as Google and Microsoft, have outstanding security expertise, and all are certified compliant with federal data governance standards. However, for years banks have delayed and tried to avoid the issue of modernizing their infrastructure. While they agree that cloud infrastructure offers them the ability to better cater to the needs of the digital customer, they are reluctant to adopt it until they are convinced of the safety and security of their data. They do not have a clear strategy that will help them quickly adopt cloud.
6. Agile and DevOps ways of working
Legacy systems are expensive to maintain and delay a product’s time to market.17 They are also not ideally suited to Agile programming methods, instead relying on waterfall methods that can slow down the production of software and result in less timely feature releases.
Banks that have made the shift to Agile and DevOps ways of working have benefited. In 2012, J.P. Morgan followed a quarterly software-release cycle, and coordination between development and operations was minimal. These quarterly releases heightened risks, were cumbersome and time-consuming, and increased delivery costs. The financial institution decided to embrace Agile and DevOps practices. The software-releases cycle changed from quarterly to 100 releases in 2015, 200 in 2016 and over 400 in 2017.18
Capital One’s move from waterfall to Agile software development helped cut the time to build new application infrastructure by 99%. DevOps’ automation and continuous integration of new code helped speed the bank’s development cycles, and releases occurred with increased frequency and higher reliability.19
7. Cultural shift
Banks need to undergo a perception change. They must be ready to build a culture that is cost-conscious, customer-conscious and efficiency-conscious.20 This is easier said than done, as many systems, processes and people have grown with the bank. A Boston Consulting Group assessment revealed that among companies that underwent a digital transformation, the number of profitable enterprises was five times higher among those that focused on a cultural shift compared with those that did not.21
Moving data and applications to the cloud can save banks money. Some say it could cut IT costs by as much as 75%.22 A large global bank that Infosys partnered with to solve this problem estimated that it could save 50% of costs overall by adopting cloud. Some areas in the bank expect to save 90% of their costs in the transition. To achieve this, the bank is working with Infosys to build a multi-cloud data management system that interfaces with Google, Amazon and Microsoft clouds to migrate its data.23
Cloud is pivotal in architecting the bank of the future. Banks can benefit from the move to the cloud not only by reducing costs, but also by capitalizing on cloud’s computing capabilities, its ability to scale IT solutions and its reliability. In fact, moving to the cloud can make banks more like the fintech upstarts with which they increasingly compete. Founded in 2015, fintech company Monzo has an infrastructure that is capable of serving 1.7 million customers supported by only 10 people on its infrastructure and reliability team. The bank has 400 core banking microservices on the cloud, which help it deliver value to customers in the form of offerings such as instant balance inquiry and real-time statements.24
The bank that Infosys partnered with to build the open source data management platform started off with the classic challenges faced by many of its peers. Project teams worked in silos and used various tools and software. For moving data, Ab Initio was used; for tagging, Collibra; and for scheduling, Control-M. There were nearly 20 moving parts, with licenses attached to each.
The bank generates and moves terabytes of data each day. Each process of data ingestion took eight to 12 weeks to deliver and was cumbersome, time-consuming and counterproductive.
To solve this problem, Infosys built an open source, petabyte-scale multi-cloud data ingestion and management platform. It is a metadata-driven data management ecosystem that has been designed to meet an organization’s current and future data delivery requirements. The platform allows businesses or functions within banks to move data from a source to a destination in a defined format at an agreed frequency.
The platform was built to solve a host of banking issues, beginning with data management. It provides a central way to monitor all banking data and enable it to be ingested in the cloud without duplication. The platform has also enabled the implementation of a Trusted Source Framework which helps with data lineage. This allows users to track data usage, understand who has made the last change, how the data has been tagged and better manage the single view of the available data.
The data management platform first targets the ingestion problem — taking data and moving it to cloud or on-premises, ingesting it on cloud systems such as Hadoop, Google Cloud Platform (GCP), Amazon Web Services (AWS) and Microsoft Azure.
Second, it focuses on the data management issue. The platform enables automated ingestion of data versus the manual and siloed approach previously followed by banks. It also provides a platform and interface to ingest data centrally.
The platform can replicate data from on-premises to a multi-cloud environment, while supporting batch and near-real-time movement. It was built with the purpose of enabling guaranteed data delivery at scale. The data management platform carries out various functions, removing the need to use multiple types of software for each function and saving licensing costs.
The platform acquires, ingests and transforms data in the following stages:
Banking transactional data is stored on multiple databases and interchange formats built on legacy mainframe technology. Structured and unstructured data is stored in big data Hadoop platforms, Oracle databases, DB2, etc.
The platform interfaces with the databases to acquire the stored data.
Once the raw data is pulled in from on-premises, it needs to be stored in a format that is durable and can be easily accessed. The platform’s architecture ingests this data in various phases:
3. Publish, Transform and Manage
Stored data is transformed into actionable information, and the results are converted into a format that is easy to draw insights from.
With Infosys’ data management platform, banks can now migrate volumes of data back and forth quickly at scale and deploy artificial intelligence and machine learning to analyze data, provide better insights and make better decisions. As a result, banks can begin to act truly as digital-first companies, much like the fintech competitors they increasingly face in the market. As an open source solution, Infosys’ data management platform will benefit from the contributions of the open source community and other banks that choose to test and use it. We hope that in the future, it will become a standard platform that enables traditional large banks to engage with the cloud at petabyte scale.