Listen to this story
|
According to the National Payments Corporation of India, UPI recorded a staggering 12.02 billion transactions in December 2023 – its highest to date. The country has become an undisputed global leader riding on the sheer volume of daily and monthly payments made as UPI transactions grew 147% catapulting from INR 92 crore in 2017 to INR 8,375 crore in 2023.
One of the major players contributing to this growth is the real-time NoSQL database company Aerospike.
“We’ve found that Aerospike can effectively handle the scale associated with such high transaction volumes. It’s exciting not just because we’re addressing a technical challenge we’re designed for, but also because we’re actively participating in the rapid growth observed in India’s payment space,” Srini V Srinivasan, co-founder and chief technology officer, Aerospike, told AIM in a recent conversation.
Founded over 14 years ago by Brian Bulkowski and Srini, the company’s primary customers are from adtech, telco, payments, financial services, gaming, e-commerce, and cybersecurity firms. Some prominent names associated with Aerospike are Yahoo, NFL, Airtel, Flipkart, PayPal, Adobe, Nokia, Dream11, are some of them.
The company has its headquarters in Bengaluru and Mountain View, US, and Srini believes that its entry into India wasn’t just strategic – it was timely. Aerospike entered the Indian market during the early stages of internet companies, Snapdeal and InMobi being among its first customers, and later with companies like Airtel and Flipkart.
“India has proven to be an incredible market for us, driven by the dynamic growth of the internet, the telecom sector, and the government’s initiatives in financial services. The emphasis on scalable systems aligns perfectly with Aerospike’s strengths. The robust competition among companies and the government’s focus on initiatives like universal financial inclusion and UPI have created a conducive environment for our solutions,” added Srini.
The company initially focused on providing solutions for adtech, banking, financial services, and payments, where identity resolution plays a crucial role. Recognising the increasing complexity in identity resolution, it expanded its offerings to include the document model (JSON) and an SQL interface via Starburst.
“Our focus has been addressing a crucial challenge faced by large-scale, consumer-oriented internet companies, which revolves around maintaining real-time systems while delivering a consistently high-quality user experience,” said Srini.
What Makes Aerospike a Customer Favourite
There are several factors that make Aerospike the preferred choice for these customers. “Firstly, we can leverage SSDs (Solid State Drive) for real-time data, reducing server footprint and offering a more sustainable and cost-effective solution, with transaction costs 5 to 10 times lower than other in-memory systems,” said Srini.
Secondly, its distributed database system capabilities are robust in the face of failures. This means that even if a node fails or a network issue occurs, the system continues to operate, providing a consistent and positive user experience. Lastly, it also provides a true database that is strongly consistent, maximising performance while ensuring this level of consistency support for applications.
Moving ahead, its customers are more interested in improving efficiency in an environment-friendly manner.
“When we started the company, we prioritised building highly efficient systems at scale, with a significantly reduced footprint and cost. This efficiency extends to easy power management. Lately, our focus has shifted to sustainability, particularly in terms of minimising power usage per user and per transaction while reducing our carbon footprint,” added Srini.
Another important point which Srini noted is that clients are actively tapping into the generative AI space and Aerospike is helping it with its support for graph and the upcoming vector database work.
Fuelling GenAI Applications
Being a high-performance database system, the company doesn’t directly build AI/ML applications. Instead, its customers employ Aerospike for various classical AI use cases, such as fraud detection, risk analysis, cyber threat analysis, and recommendation engines.
Over the years, it has enhanced its platform with features supporting predictive AI, including robust indexing, secondary indexing capabilities with scans, and parallel access to the database. These features facilitate efficient training of data models and quick data processing, making Aerospike a reliable database for storing real-time activities at the edge.
However, with customer-driven demand, the company ventured into graph databases with Aerospike Graph, based on Gremlin query language, putting established players like Neo4j and TigerGraph under immense pressure.
The trend among customers developing custom identity solutions using Apache TinkerPop with Aerospike as the storage layer prompted Aerospike to officially implement a graph solution. After evaluating the potential and witnessing customers scale to petabytes while maintaining low access times, the company introduced Aerospike Graph, inspired by a large customer’s success in the payment systems business.
“Moving forward, we are actively participating in this wave of generative AI innovation by adding support for vector and graph features. The vector database serves as a relationship management system, enabling enterprises to keep their data while utilising Aerospike to generate vectors,” said Srini.
The company plans to integrate tools like LangChain and other cloud-based AI/ML modelling environments to allow users to seamlessly build LLM-based applications.
Read more: Why Graph Databases Remain Untapped