Real-time recommendations at any scale with Aerospike Vector Search

Vector embedding models are revolutionizing fraud detection, product recommendations, and personalized digital advertising. With Aerospike Vector Search (AVS), you can unlock real-time search across your data at any scale.

What is Aerospike Vector Search?

Existing real-time search and retrieval solutions are high-cost, brittle, and inefficient. Worse, they frequently deliver low-quality results, limiting their effectiveness at making better recommendations, detecting fraud, or serving real-time personalized advertising. 

Aerospike Vector Search changes the game. It empowers developers to leverage the latest in machine learning models to build search indexes directly on their real-time key-value data, delivering fast, accurate, and cost-effective results. 

Key features of Aerospike Vector Search

We’re excited to extend Aerospike's limitless storage capabilities with Aerospike Vector Search's limitless possibilities. 

Flexible storage for any (search) use case

Aerospike offers the highest throughput and lowest latency storage for both your vector indexes and associated metadata, enabling a variety of storage options to best fit your SLA and budget. Using Aerospike to store both gives you the flexibility to make optimal cost and performance trade-offs for your use case. 

• Small indexes: Stored in memory for lightning-fast performance. 

• Large indexes: Hybrid memory delivers near real-time results at any scale at a low total cost.

Learn what level of performance to expect in our AVS scaling guide. 

Extend your data however you want

Aerospike Vector Search enables developers to add vectors to existing records, eliminating the need for separate,  search-specific systems.

• Avoid duplicating data.

• Build unlimited indexes tailored to your needs. 

Learn more details about our data model. 

Self-healing live indexes

Aerospike Vector Search enables accuracy and reliability with durable, self-healing indexing that updates in real time. 

• Scale your cluster independently for indexing.  

• Maintain query throughput and performance without interruptions. 

This healing approach allows for the scale-out of ingestion capabilities and can be tuned for each specific index.

Fits seamlessly into your AI stack

AVS integrates with popular frameworks and cloud providers, offering flexibility for your development needs:   

• Use our Langchain extension for building RAG applications or our 

• AWS Bedrock sample embedding example to build out your enterprise-ready data pipeline with ease. 

Enterprise-ready

AVS comes ready with all the necessary tools and features required for production-ready deployment:

• RBAC (Role-Based Access Control):  Creating a large index can have costly implications. Separate the index management capabilities of your data pipeline from search applications by leveraging unique admin and developer roles. (Docs)

• CLI management tool: Search admins need to know how to manage indexes at a glance. Troubleshoot your cluster, browse your data, and manage your indexes, all from a simple-to-use and intuitive CLI.  (Docs | Releases)

• Grafana monitoring dashboards: At a glance, information is not always enough to tune and troubleshoot your search system. AVS includes monitoring dashboards to gain nuanced details on cache behavior, query latencies, and indexing performance. (Docs | Releases)

• Perfect for container orchestration: AVS nodes are staples, which makes them perfect for container orchestration platforms like Kubernetes. Search admins can focus on automating scaling of the data pipeline and search application SLAs and count on Aerospike's high-performance, low-cost storage capabilities. (Docs)

Get started today! 

Kickstart your journey with our sample apps, Python notebooks, and deployment scripts. Get familiar with the AVS Python client and learn how to create vector embeddings using open-source models. If you’d like to explore the full implementation yourself, check out the code here.