How in-memory computing can maximize the performance of modern payments, IoT and blockchain applications

By Nikita Ivanov | 1 November 2017

In-Memory Computing for Financial Services eBook, Part 2,” published by GridGain Systems, takes a detailed look at how in-memory computing platforms such as GridGain and Apache® Ignite™ can help financial services firms leverage in-memory computing to take advantage of modern payments systems, the Internet of Things (IoT), and blockchain. In-memory computing platforms provide blazing fast speed, massive scalability and maintain data in RAM across a distributed computing environment to provide blazing fast speed and massive scalability compared to systems built on disk-based databases.

Modern Payments Systems

According to, fewer than 25 percent of U.S. in-store purchases in 2017 are expected to be cash-based, and worldwide, the mobile payments market is projected to reach a trillion dollars by 2019.

To compete in this rapidly evolving arena, payment services providers must ensure an excellent customer experience by processing payment transactions in real-time. They must do so even as they continue to scale their business and implement sophisticated analytics for fraud prevention, compliance, and revenue-generating insights.

To satisfy all these demands, payment services providers must have an IT infrastructure capable of maintaining the highest possible performance levels, which is easily scalable, and provides 24/7 availability at scale. They must also have the flexibility to offer payment options from a wide range of devices and applications, adapt to evolving payment scenarios, and cooperate with other vendors, banks and fintech firms.

To achieve this, the underlying technology infrastructure should be highly performant. Ideally, it should be based on open source software and built with commodity hardware, ensuring maximum interoperability and scalability. For payment situations requiring real-time analysis of large datasets, such as currency conversion rates for international payments, the technology must also be able to handle big data analytics in an environment which supports hybrid transactional/analytical processing (HTAP).


In principle, by transmitting information simultaneously to all interested parties, the decentralized, self-perpetuating blockchain infrastructure can streamline many types of transactions and save substantial amounts of time and money for those needing access to the transaction data.

Most larger banks, along with many technology companies serving the financial services industry, are building platforms to support the use of blockchain. In a 2016 Deutsche Bank survey of 200 participants in the global financial industry, a stunning 87 percent of respondents expected blockchain to have a major impact on the securities services market, with widespread adoption within the next three to six years.

Bitcoin, for example, is a flourishing, fully electronic “cryptocurrency” based on the distributed-ledger blockchain technology. Whenever a Bitcoin transaction occurs, such as a purchase, a secure transaction message goes from the buyer to the seller. Only the message sender (buyer) and receiver (seller) can view the contents of the encrypted message (the transaction details), but a record of the transaction (source, destination, and bitcoins spent) is added to the blockchain (the distributed ledger) that is maintained by each blockchain subscriber.

As a fully electronic currency, however, Bitcoin is vulnerable to cybertheft, hacking, authentication issues, performance issues and system failures. Because of this, the adoption of blockchain will likely require significant technology upgrades. Financial services firms will need to store the incoming blocks and translate and validate the information. They will need to maintain tables of security identifiers, validation information, cross-references, and so on. They will need real-time hybrid transactional/analytical processing, or HTAP, capabilities. And they will need all of this on a massive scale.

The Internet of Things

Most organisations are now familiar with the potential of the Internet of Things (IoT) across a range of industries. Across the financial services landscape, potential IoT use cases are rapidly moving toward reality, and IoT devices are already collecting data that needs to be analyzed in real time and stored for historical analysis. For example, banks are installing location-aware ATMs that pre-load account information from approaching customers who are identified by their cell phones.

However, to support IoT initiatives, financial services firms must have a highly reliable, highly available platform capable of fast and scalable back-end storage along with tremendous computational and analytical prowess. The platform must also be able to stream data collection in near real time while providing interoperability and the best possible security.

Meeting these challenges with in-memory computing

In-memory computing platforms provide parallel, distributed processing across a pool of RAM deployed on a computing cluster. This model enables much faster transaction processing and analytics for data-intensive applications compared to applications built on disk-based databases, along with ability to easily scale out by adding new nodes to the cluster. Until recently, the high cost of RAM meant in-memory computing was economically feasible only for very high-value applications. But today, a steady drop in the price of RAM has made in-memory computing platforms economical for a much wider range of use cases. Gartner projects that the in-memory technology market will grow at a compound 22 percent per year to reach $13 billion by the end of 2020.

For modern payment systems, an in-memory computing platform can provide response times that are 1,000 to 1,000,000 times faster than traditional approaches based on disk-based databases. They can maintain this performance and low latency as the system scales. Built with open source software and commodity servers, such a platform offers the flexibility and interoperability to adapt rapidly to evolving payment scenarios.

For Bitcoin infrastructures, an in-memory computing platform provides the performance and scale necessary to scale and secure the underlying blockchain distributed ledger technology. The scalability of the computing cluster, which automatically utilizes the RAM and CPU power of each new server added to the cluster, means firms can provide the required level of performance and security at scale.

For IoT applications, an in-memory computing platform, perhaps more than any other technology, enables organizations to cope with the ever-growing, ever-accelerating stream of real-time data. With in-memory computing, organisations can avoid the cycle that rapid growth often entails with systems built on relational databases: increased latency, new hardware expenditures, increased latency, etc. In addition, an in-memory computing platform built using open source software and commodity servers offers long-term cost savings and significantly higher ROI.

In-memory computing is a mature technology which provides high performance, massive scalability, high availability and cost savings across a range of industries and use cases. Financial services firms should explore how in-memory computing can help them thrive in today’s highly competitive, always evolving and growing, information-centric environment.