Community
Payment Switch is a complex fabric in which functionality and technicality are intricately woven to provide a solid platform to deliver ever increasing different type of payment services with high throughput & resilience.
From 10000 Feet view, it might seem that “PS” is a generic platform used to transform and route different kind-off payment messages for authorization. But as we climb down the chain we can see different payment services, multiple components and complex technical components and frameworks to support payment services.
In the above image, I tried to capture few properties/components of a switch. Layers at higher level are dependent on the ones at lower level.
Payment Services: Gone are the days when switch was only supposed to process Credit Card (CC)/ Debit Card (DC) authorization transactions.
Supporting Services: To support different payment services, additional services are required.
Moving further down, service components are defined. Payment Services consume service components directly or via Supporting services. Above mentioned abstract layering is just an “VIEW” of different type of services/components.
The second aspect of switch, i.e., technology is equally important otherwise it becomes next to impossible to provide payment services in a consistent manner.
From architecture and design view point Payment Switch has to be high performance system besides highly resilient and available. This aspect makes PS technically challenging.
To support transactions with TPS of 10000 and more and provide high-availability at different level of system is not a simple task. Lot of architecture design patterns, integration methodologies are available to solve the NFR (non-functional requirements) of switch.
Load and Performance: High end switch has to handle simultaneous tens of thousands of ATMs and POS device and they have to ensure zero or less degradation in performance. Earlier ATMs, used to maintain connection with switch for long durations, which is again a huge bottleneck.
With large number of transactions, performance becomes a major sufferer: from application server to Databases - less memory, less CPU and less disk space.
Solution comes primarily in 2 variations:
High Availability: Being a “financially” sensitive system, any downtime (crash) of switch results in financial loss. HA needs to be maintained from multiple perspectives: Inside Data Centre, Primary DC -BackUp DC, Primary DC – Primary DC.
Factors like latency (both inside DC & between DCs), scheduled/non-scheduled maintenance time, patch deployment, self-healing etc. matters lot in production environments.
Resilience: The expected (both functional and technical) behavior of switch under varied conditions & zero transaction loss in any case. Especially, behavior under crash & no response from components (component is live but not responding).
Tendency towards conversion for Spaghetti Architecture: Mostly payment systems have to integrate with multiple external systems for data posting, validation, authentication, data reporting etc. Its results in supporting various formats, transmission protocols, communication protocols and most importantly multiple invocation points inside transaction life-cycle.
Finally, the architecture converts into Spaghetti architecture, which in long run is a BIG NO and DANGEROUS.
In this blog, I tried to bring some aspects of Payment Switch from both functional and technology perspective. In the next blog, I will discuss the solution….Till then.
This content is provided by an external author without editing by Finextra. It expresses the views and opinions of the author.
Kunal Jhunjhunwala Founder at airpay payment services
22 November
Shiv Nanda Content Strategist at https://www.financialexpress.com/
David Smith Information Analyst at ManpowerGroup
20 November
Konstantin Rabin Head of Marketing at Kontomatik
19 November
Welcome to Finextra. We use cookies to help us to deliver our services. You may change your preferences at our Cookie Centre.
Please read our Privacy Policy.