AWS Database Blog
Amazon ElastiCache re:Invent 2025 recap
AWS re:Invent 2025 was another incredible event! Tens of thousands of builders came together to learn about the latest AWS innovations. For the ElastiCache team, it was a valuable opportunity to meet several customers and hear directly from them about what’s working well and what challenges they’re facing – insights that directly shape our roadmap to better serve them. At the same time, we got to showcase new ElastiCache capabilities that help customers unlock even greater performance and value.
The key themes this year were upgrading to Valkey from Redis, caching strategies for generative AI and agentic AI workloads, and building cost-efficient, high-performance applications with ElastiCache Serverless. In case you missed some of these sessions, or you wanted to get caught up on why customers like Expedia, Scopely, Adobe, and Amazon.com are building on ElastiCache, you can read this helpful summary of some of the ElastiCache highlights from re:Invent 2025.
Leadership Session: Databases made effortless so agents and developers can change the world
From startups to global enterprises, AWS Databases has been the trusted foundation powering innovative applications for nearly two decades. Join G2 Krishnamoorthy, Vice President of AWS Database Services, as he presents our mission to transform how you manage your data. Discover our vision where database management becomes effortless, delivering robust performance, seamless scale, and comprehensive security. Learn how databases integrate with AI agents and tools to reduce operational complexity while supporting a new generation of AI-native applications to advance your business capabilities. Hear from Vercel and Robinhood as they build new opportunities with AWS.
- Full session video
For ElastiCache specifically, this session covers semantic caching with ElastiCache for Valkey, which achieves the lowest latency vector search with highest throughput and best price-performance at 95% recall rate among popular vector databases on AWS for both agentic AI (such as conversational memory) and generative AI (such as Retrieval Augmented Generation) use cases. The session also covers how frameworks like Mem0, LMCache, and LangGraph use ElastiCache for Valkey to achieve millisecond retrieval performance for short-term agentic memory.
- ElastiCache segment
To learn more about semantic caching with ElastiCache, see Lower cost and latency for AI using Amazon ElastiCache as a semantic cache with Amazon Bedrock.
Scaling MONOPOLY GO! to millions with DynamoDB & ElastiCache
Scopely shared their incredible journey of scaling MONOPOLY GO! to serve millions of concurrent users with minimal resources. Learn how they used Amazon DynamoDB and ElastiCache Serverless to power MONOPOLY GO!’s rapid growth, handling an incredible 2.1 million writes per second at peak traffic. Discover the architectural patterns and optimization strategies that enabled seamless matchmaking, friend discovery, and real-time gameplay for millions of concurrent users. Explore practical insights on managing unpredictable traffic spikes, reducing operational complexity, and achieving cost savings through serverless technologies.
- Full session video
Better, faster, cheaper: How Valkey is revolutionizing caching
Discover how the powerful combination of open source Valkey and Amazon ElastiCache is transforming the landscape of distributed caching. Explore recent innovations such as probabilistic data structures like bloom filters, a memory efficient dictionary that reduces memory usage by up to 40%, and a multi-threaded architecture that supports instantaneous scaling by up to 230%. Learn about the infrastructure that enables ElastiCache to seamlessly patch hundreds of thousands of customers while maintaining high availability, even for extreme workloads like Prime Day’s trillion requests per minute.
- Full session video
Disagree in Commits: The Performance Improvements That Cut Costs by a Third
When paths diverge in open source, innovation often follows. Join Corey Quinn and Madelyn Olson as they dive into how Amazon ElastiCache for Valkey cut prices by up to 33% while simultaneously improving both performance and reliability through community-driven development. You’ll discover how open governance enables optimization decisions that benefit all users, why performance improvements and stability aren’t mutually exclusive, and what it really takes to create a seamless migration path at scale. Whether you’re currently running Redis, evaluating alternatives, or simply interested in how open-source economics can benefit cloud customers, you’ll leave with practical insights on reducing infrastructure costs without compromising on reliability.
- Full session video
Optimize agentic AI apps with semantic caching in Amazon ElastiCache
Multi-agent AI systems now orchestrate complex workflows requiring frequent foundation model calls. Learn how semantic caching with Amazon ElastiCache for Valkey reduces latencies from single-digit seconds to single-digit milliseconds while cutting foundation model costs for production workloads. By implementing semantic caching in agentic architectures like RAG-powered assistants and autonomous agents, you can create performant and cost-effective production-scale agentic AI systems.
- Full session video
Build serverless chatbots using Amazon ElastiCache & Aurora PostgreSQL
Learn how to build a serverless-first chatbot application by leveraging Amazon Aurora with pgvector as a vector store and Amazon ElastiCache Serverless for Valkey to manage chat history, semantic caching, and agent memory. Discover how to use these services to create contextually aware, scalable chatbots with agentic capabilities that retain conversational memory across interactions. Explore architectural patterns, implementation strategies, and real-world use cases. Gain insights into optimizing query latency, performance, and cost-effectiveness. This session equips you with practical knowledge to build modern, serverless-powered chatbot applications using generative AI.
- Full session video
Deep dive into AWS Database Savings Plans
Database Savings Plans are a flexible pricing model that reduces your database costs by up to 35% when you commit to a consistent amount of usage over a 1-year term. This savings plan automatically applies to eligible serverless and provisioned instance usage regardless of engine, instance family, size, deployment option, or AWS Region with no upfront payment. For example, with Database Savings Plans, you can switch between instance types such as Aurora db.r7g and db.r8g instances, shift a workload from EU (Ireland) to US (Ohio), or migrate from ElastiCache for Valkey instances to ElastiCache Serverless for Valkey and still benefit from the discounted rates. The session covers usage pattern analysis, cost optimization strategies, and how to maximize savings while maintaining workload flexibility.
For ElastiCache users, Database Savings Plans apply to both ElastiCache Serverless for Valkey (30% discount) and ElastiCache for Valkey instances (20% discount).
- Full session video
Optimizing for high performance with Amazon ElastiCache Serverless
Learn how to build highly scalable, high-performance applications with Amazon ElastiCache Serverless. This session shows how ElastiCache Serverless works behind the scenes and provides metrics to help you identify and debug application performance issues. Discover how to build effective applications with ElastiCache Serverless and use the design of the service to your advantage.
- Full session video
Advanced data modeling for Amazon ElastiCache
This session delves into the intricacies of Amazon ElastiCache data modeling using purpose-built Valkey data types to optimize application performance and scalability. Explore the use of strings, sets, sorted sets, hashes, bitmaps, and geospatial indexes to model complex relationships and solve use cases such as caching, session store, feature store, real-time analytics, geospatial applications, and rate limiters.
- Full session video
Video sampling & search using ElastiCache & multimodal embeddings
You can use multimodal embeddings for a wide range of video analysis capabilities, from content understanding and summarization to classification and recommendations. Learn how to use smart sampling and video search, generating multimodal embeddings in Amazon Bedrock and using vector search powered by Amazon ElastiCache. See live coding examples to discover how to build delightful video experiences for your users.
- Full session video
Key highlights
Three key highlights from this year’s ElastiCache sessions:
- Serverless at scale: Customers are increasingly upgrading their workloads to ElastiCache Serverless. MONOPOLY GO! achieved 2.1M writes/second with ElastiCache Serverless, demonstrating how serverless can handle workloads of virtually any scale.
- Valkey delivers: By upgrading to ElastiCache for Valkey, customers can achieve up to 40% memory reduction, 230% scaling improvements, and 33% reduced pricing. Customers can upgrade from ElastiCache for Redis to ElastiCache for Valkey with virtually no downtime.
- Purpose built semantic caching: ElastiCache delivers lowest latency vector search with highest throughput and best price-performance at 95% recall rate among popular vector databases on AWS. With vector search, customers can reduce latencies from single-digit seconds to single-digit milliseconds while cutting foundation model costs for production workloads.
- Valkey as a memory layer for agents: Newly launched connectors allow customers to use ElastiCache for Valkey for short-term agentic memory through frameworks such as Mem0, LMCache, and LangGraph.
- Database Savings Plans: Customers can use Database Savings Plans to reduce database costs by up to 35% with a 1-year commitment and no upfront payment required. Database Savings Plans are available for ElastiCache for Valkey.
Getting started
To learn more about Amazon ElastiCache and get started, visit the ElastiCache product page. For Valkey specific guidance, check out What is Valkey?
We’re excited about the innovations showcased at re:Invent 2025 and look forward to seeing what you build with ElastiCache. Thank you for joining us at re:Invent 2025, and we can’t wait to see you at re:Invent 2026!