Building Resilient Architecture for FedNow Using VMware GemFire

August 31, 2023 Nainesh Jhaveri

Recently the Federal Reserve has introduced the FedNow Service, which is a real-time payment service introduced in the United States. It aims to provide individuals and businesses with the ability to make instant, 24x7 payments and transfers between banks and financial institutions. By offering faster and more accessible payment options, FedNow seeks to enhance the efficiency and convenience of financial transactions, fostering innovation in the payments ecosystem and meeting the evolving needs of consumers and businesses in the digital age.

A diagram of a fed-now serviceDescription automatically generatedThe FedNow transaction flow diagram

One of the primary advantages of the FedNow Service is its ability to clear and settle transactions instantly, allowing financial institutions to enable their customers to send and receive money in a matter of seconds, with 24x7 availability.

There are several benefits of the FedNow Service, including 

  • Attracting and retaining customers 
  • Lowering costs through increased efficiency (e.g., automation and operations) 
  • Reducing interbank settlement risk 
  • Growing revenue (e.g., transaction fees for merchants, as well as fees for new product offerings)

Looking at the success of the Unified Payments Interface (UPI) system in India, we can positively predict that FedNow adoption in the U.S. may grow quickly and exponentially. There are use cases such as mortgage payments, that happen instantly without having the need to generate a banker’s cheque (cashier’s check). For example, simple transactions such as discounts at your dentist for procedures like root canals also happen quickly. Credit card companies, however, charge upwards of two to four percent per transaction for the merchant, and payments are delayed by at least a couple of days. Sometimes the same dentist would be willing to pass that discount over to the customer in order to receive money instantly and maybe save a percent or two in the process. Many other use cases such as gig workers, restaurants, and shop owners prefer instant payments over credit cards or simple bill payments. Therefore, there is a high likelihood that Automated Clearing House (ACH) transactions will eventually be replaced.

FedNow Service is now live, and one of the largest banks has architected their system by leveraging VMware GemFire to satisfy instant payment requirements of the Federal Reserve Board (FRB), both from a resiliency standpoint, as well as in meeting the low-latency and high-availability (24x7x365) needs.

Challenges

Let’s dig into a few challenges that are generally faced by financial institutions and how GemFire can be leveraged to tackle such challenges and keep systems up and running at all times.

  • Low latency – Need to process payments within milliseconds
  • High availability/disaster recovery – Services must be up and running 24x7 
  • Resiliency – Ability to serve traffic from different availability zones (AZs) without any user interruptions
  • Risk mitigation – Need to watch for and mitigate any financial risks (e.g., fraud, anti-money laundering, specially designated national checks, suspicious activity reports, etc.)
  • Net zero emissions – Maintenance of  the company’s goals. 

 Sample architecture for processing instant payments by leveraging VMware GemFire

Low latency 

GemFire is an in-memory data grid and is well known for its ultra-low latency on individual read/writes. In parallel, it supports very high frequency concurrent read/writes. In this context, by storing the payment record in the GemFire in-memory cache, all the downstream systems can be processed quickly and work in parallel to register a “yes/no” decision directly in the GemFire cache for this specific transaction/payment.

High availability/disaster recovery

If the server hosting the primary copy of this payment record goes down, GemFire upgrades the server with the secondary copy as the primary copy for that key. Although this might cause a temporary loss of redundancy (i.e., from three copies down to two copies), there would be no loss of data and no user interruption either. 

Whenever there are not enough secondary copies to satisfy redundancy, the GemFire system works to recover redundancy by assigning another member as secondary and copying the data to it. 

Let’s take this one level deeper. What if the entire data center goes down? That is not a problem because GemFire offers WAN replication as shown in the following image.
 

A diagram of a computerDescription automatically generatedGemFire multi-site architecture

GemFire can share data with another site for disaster recovery needs and continuously replicate data between GemFire clusters located across multiple sites.

Therefore, the overall benefits of using GemFire for your DR needs are

  • Global deployments with fast local data access
  • Data is always current everywhere 
  • Satisfies disaster recovery in seconds 
  • Allows for planned upgrades and outages 
  • Solves the regulatory compliance needs of having service live 24x7
  • GemFire also offers rolling upgrades to keep the grid always running
  • Secure data transfer 24x7 leveraging encryption using your organization’s X.509 certificates and latest SSL/TLS standards

Resiliency 

As shown in the previous image, you’ll see how GemFire allows the payment record to be replicated in three different GemFire nodes running in three AZs. GemFire will always place redundant copies of the data in a different AZ. With redundancy turned on, even if one member fails, operations continue with no interruption of service using a different server in a different AZ.

Risk mitigation

ACH transactions take longer to complete, but they have a number of checks and validations that we cannot skip. So in this context, there is a need to complete all the checks that took about two days in just a few seconds. To solve this, downstream systems need to employ the GemFire in-memory cache compute facilities. By keeping the last 3–6 months of data active in the GemFire cache, the system does not need to wait for any spinning disks or queries to execute. For example, to complete a fraud or AML, or to do a complete SAR check, the system can render a yes/no decision for a specific payment in a matter of milliseconds, by keeping the last 3–6 months of data in the GemFire cache.

Net zero emission

GemFire has a persistence layer that makes it very reliable. In fact, many companies leverage it as a system of record. By storing data close to the application tier, one can eliminate thousands of roundtrips to backend databases and also eliminate dozens of copies made in downstream systems, thereby reducing the number of servers and storage devices attached to various downstream systems and reducing emissions.

In conclusion, VMware GemFire is perfectly suited for FedNow Service transactions due to its robust, high-performance data management capabilities. As a distributed in-memory data grid, GemFire ensures low-latency access to real-time data, aligning perfectly with the instantaneous nature of FedNow payments. Its scalability and fault-tolerance features enable seamless handling of high transaction volumes, guaranteeing reliable and uninterrupted services. The data replication and data partitioning mechanisms of GemFire ensure data consistency and availability across multiple nodes, providing a resilient platform for critical payment systems. With GemFire, the FedNow initiative can achieve the speed, reliability, and scalability required to support real-time payments effectively and securely.

About the Author

Nainesh Jhaveri

Nainesh has over 25 years of experience in the IT industry and is passionate about building products, services, and solutions that are secure, operational, Sensible, and most importantly resilient. At VMware, Nainesh is leading the Tanzu GemFire product line, ensuring the delivery of world class software leveraged by large enterprises for their franchise-critical applications.

Follow on Linkedin More Content by Nainesh Jhaveri
Previous
Open Source vs. Enterprise Edition of Bitnami Application Catalog: How to Make the Right Choice
Open Source vs. Enterprise Edition of Bitnami Application Catalog: How to Make the Right Choice

The widely popular Bitnami-packaged content is available in a standard open source version as well as an en...

Next
The Double-Edged Sword of Modern Software Delivery
The Double-Edged Sword of Modern Software Delivery

Kubernetes offers undeniable benefits, but without standardizing Kubernetes adoption across your organizati...