Aerospike is a real-time database built for infinite scale, speed, and savings. Pulse 2.0 interviewed Aerospike CEO Subbu Iyer to learn more about the company.
Subbu Iyer’s Background
What is Subbu Iyer’s background? Iyer said:
“For the last 30 years, I’ve been leading teams at innovative enterprise technology companies. Scaling businesses to reach their full potential has always been a passion for me. Prior to joining Aerospike as CEO, I gained invaluable experience in executive leadership roles at companies like Oracle, Veritas, Hewlett Packard Enterprise, Dell EMC, and Riverbed Technology.”
Formation Of Aerospike
How did the idea for Aerospike come together? Iyer shared:
“Aerospike was founded to solve the scaling problems that face modern databases. When our co-founder Srini Srinivasan was senior director of engineering at Yahoo, responsible for mobile services, he continually encountered the limits that hold back traditional databases. Most importantly, as overall throughput and database sizes grew, performance plunged, service outages became more frequent, and costs soared.”
“Srini’s vision was to create a solution that can process extremely high read-write transaction throughput (millions of operations per second) on enormous amounts of data (terabytes to petabytes) and not go broke doing it. He felt that Aerospike could build a system that was 10x faster but also 10x cheaper by exploiting the parallel random access read capabilities of SSDs whose speeds and densities were accelerating, while prices were plummeting. With that as a foundational principle, we created a patented Hybrid Memory Architecture (HMA) technology that enables enterprises to use the best storage options for use by real-time database applications — and all of it is optimized for performance at scale. The Hybrid Memory Architecture combined with Aerospike’s robust distributed design with support for geo-distributed transactions provides a system that keeps data safe and provides high availability along with high performance.”
“Initially, potential customers didn’t believe Aerospike’s database performance at scale was possible. They were used to decades-old tradeoffs that were just a way of life with traditional databases, such as getting high performance but only on a small amount of data processed on expensive hardware. But many early customers quickly realized what we already knew — that Aerospike handled real-time applications better than most other database systems.”
“Fast forward to today, and Aerospike customers are at the forefront of their respective industries. Adobe, AppsFlyer, Barclays, Flipkart, Myntra, PayPal, Riskified, Criteo, DBS Bank, Experian, Snap, and Sony Interactive Entertainment are just a sample of the enterprises using Aerospike at the heart of their mission-critical real-time applications, with use cases ranging from AI and ML to AdTech, recommendations, customer 360, and fraud detection.”
Core Products
What are the company’s core products and features? Iyer explained:
“Aerospike is a real-time multi-model database supporting NoSQL data models (key-value and document) as well as graph and vector data models. In each of these cases, Aerospike offers better performance at a massive scale than other databases. It does this while operating on a fraction of the infrastructure required by other databases, enabling customers to reduce server footprints and costs by as much as 80%, which, in turn, results in dramatically lower carbon emissions.”
“As a multi-model database, Aerospike serves all the most popular database implementations — key-value, document, graph, and now vector search — within a single platform. Combining data models in a single platform enables creation of richer solutions that can manipulate these different types of data within the same application. This means that companies can have the maximum flexibility across data models, clients, and microservices. They can seamlessly scale applications, customers, and data without being restricted by the limitations of other traditional systems.”
Challenges Faced
What challenges have Iyer and the team faced in building the company? Iyer acknowledged:
“AI is disrupting all industries. It demands a lot of data for accuracy. This exacerbates the traditional tradeoffs between performance, scalability, and costs when dealing with massive amounts of data and high-transaction volumes. Legacy data architectures and databases weren’t designed for the firehose of data AI demands. Even newer systems struggle with being affordable as they scale.”
“Accuracy in the face of this deluge of data requires specific context for effective decision-making. This matters because by understanding and applying the nuances of specific, in-the-moment contexts, AI systems can deliver the highly personalized experiences that bring the greatest value to users.”
“With large language models (LLMs), this specificity is achieved through retrieval augmented generation (RAG), which incorporates detailed context into queries to produce relevant and customized responses.”
“Embeddings, which are mathematical representations — or vectors — also play a pivotal role in capturing and utilizing context by encoding diverse aspects of data. Embeddings provide a rich, contextual backdrop that augments the semantic understanding of LLMs, leading to more precise and contextually relevant outcomes.”
“Aerospike eliminates many obstacles that have roadblocked the industry for years and which must be overcome to innovate and grow in the AI era. Our multi-model database allows enterprises to affordably scale to handle tens of millions of transactions per second. And, our new highly performant vector solution inherits the strengths of our core database to provide the only enterprise-grade vector solution that can handle billions-to-trillions of records with consistent millisecond accuracy.”
Evolution Of Aerospike’s Technology
How has the company’s technology evolved since launching? Iyer noted:
“Aerospike has been deployed in companies throughout the world for over 13 years, achieving many patents, benchmarks, and awards. More than 250 companies at the forefront of their respective industries use Aerospike at the heart of real-time applications ranging from AI and ML to AdTech, recommendations, customer 360, and fraud detection.”
Significant Milestones
What have been some of the company’s most significant milestones? Iyer cited:
“In 2023, the company released Aerospike Graph, which enables millisecond multi-hop graph queries with exceptional throughput, even across billions of vertices each with thousands of edges. In April, the company unveiled Aerospike Vector—the only enterprise-grade vector database solution to ensure consistent accuracy at scale while reducing infrastructure requirements.”
“Additionally, Aerospike has been recognized numerous times for our innovation and the successes that customers have realized by making Aerospike a part of their technology stack. Aerospike was recently chosen as the winner in the In-memory Solution of the Year category for the Data Breakthrough Awards and was named a winner in the 2024 Business Sustainability Awards.”
Funding
When asking Iyer about the company’s funding details, he revealed:
“Aerospike recently closed $114 million in growth capital investment from Sumeru Equity Partners and existing investor Alsop Louie Partners. This sizable investment was a great validation of Aerospike’s product strength and business performance.”
Differentiation From The Competition
What differentiates the company from its competition? Iyer affirmed:
“As noted earlier, Aerospike makes it possible to perform millions of transactions at sub-millisecond speed, even on massive amounts (e.g. petabytes) of data. And even better, Aerospike can do it on a fraction of the servers that traditional solutions require. Not only does this provide incredible cost savings, but it also highlights Aerospike’s sustainability. With an ability to run on up to 80% fewer servers than other databases, Aerospike dramatically lowers carbon emissions.”
Future Company Goals
What are some of the company’s future goals? Iyer pointed out:
“Aerospike has been aggressively building upon its current capabilities and customer traction as a data platform for AI, and this funding will allow us to accelerate our go-to-market capabilities, talent, and research and development to keep up with significant growth and the massive demand for data in AI.”
Additional Thoughts
Any other topics to discuss? Iyer concluded:
“In the context of real-time AI applications, certain drivers lead to an explosion of data. For example, hyper-personalization is moving away from predictions and decisions being made on statistically aggregated data and offline models trained on historical data to decisions predicated on a particular user’s attributes and behavior history, including their most recent actions. This emphasis on using both relatively static profile data and temporal behavioral data leads to both a requirement to use vastly larger data sets, as well as the need to ingest and update these at a far larger scale.”
