Aerospike, Inc., unveiled new expression indexes that boost query performance while significantly reducing memory use and operational costs for machine learning and generative and agentic AI.
Most NoSQL databases scan each individual record, even if most are irrelevant to a query. This slows application performance, wastes resources, and skyrockets operational costs. To even attempt expression-like indexing, most data and application teams are forced to create extra fields (or what Aerospike calls “bins”) and develop client-side logic to compute and filter values. This leads to duplicated logic across multiple clients, which poses a risk to consistency.
Aerospike is one of the first real-time NoSQL databases to offer expression indexes for any application that needs to read smaller, more relevant slices of information in a massive database. Also known as function-based indexes, expression indexes push conditional logic into the index itself. This lets the Aerospike database use expressions to index only relevant computed data. Memory usage drops proportionately to the number of excluded records, creating fewer records to scan, smaller indexes, lower memory use, faster queries, and simpler application logic.
“As organizations race to scale up ML and data infrastructure for a competitive advantage, they are slamming into a shortage of data center space, performance barriers, and rising complexity and costs,” said Srini Srinivasan, CTO and founder, Aerospike. “Aerospike has always been at the forefront of maximum performance and scale with the lowest footprint and cost. With expression indexes, we continue to eliminate any data infrastructure tradeoffs in performance, scale, complexity, and cost.”
Expression indexes are ideal for developers, architects, and DevOps teams managing large-scale data operations, especially in sectors like ad tech, financial services, and e-commerce.
They’re available immediately in Aerospike Database 8.1.
Learn more in this Quickstart Start guide.
