In-memory datastores like Valkey are particularly well-suited for applications and functions that depend on extremely low latency. These can include:

• Caching: Valkey can act as a high-speed cache layer in front of slower, disk-based databases. Frequently accessed data can be stored in Valkey, significantly reducing the load on the primary database and accelerating response times.
• Message queues: Valkey's Pub/Sub capabilities make it an excellent choice as a message broker. It can be used to facilitate real-time communication between different services or microservices, supporting use cases like chat applications and notifications.
• Real-time analytics: For applications that need to process and analyze data in real time, such as parcel trackers, navigation apps, or gaming leaderboards, Valkey’s speed makes it an ideal choice.
• Session stores: Managing user sessions in web applications, such as e-commerce sites or social media platforms, requires quick access to user preferences, login status, and shopping cart contents. Valkey provides a fast and reliable way to store and retrieve this session data.
• Machine learning: Valkey’s low latency and vector search capabilities also make it a powerful database for applications that use machine learning. You can store and rapidly search vector embeddings—the numerical representations of data like text or images—to perform complex similarity searches. This can be used to suggest similar products, articles, or media to users in real time.

Valkey was initially forked from Redis 7.2, meaning it started from the same codebase and shares some fundamental characteristics and functionality. Like Valkey, Redis is an in-memory key-value datastore that offers high throughput and supports a range of data structures. Both are known for high availability and scalability.

The core difference is that Valkey is open source, released under a permissive BSD license, so developers are free to use, modify, and make contributions to the software.

In 2024, Valkey was created in response to Redis Inc. switching to a more restrictive ‘source available’ license. To maintain an open-source alternative for the community, several core Redis contributors came together and launched Valkey based on the final open-source version of Redis.

Valkey is driven by a collaborative, community-led development approach, where the roadmap and new features are decided by the contributing members of the Linux Foundation project with contributions from major cloud vendors, including Google Cloud.

Since Valkey 8.0, major enhancements have been made to the Valkey engine, focusing on vector search enhancements, improved cluster management, and multi-threaded command execution—the ability to process commands in parallel across multiple CPU cores.

For those who want more granular control, customization, or specific deployment configurations, Valkey can be manually deployed onto Google Cloud’s compute infrastructure. This method offers you complete oversight of the environment.

• Compute Engine (VMs): Launch Google Compute Engine instances (VMs) and install the open-source Valkey server directly. This provides full autonomy over configuration, scaling parameters, and the underlying operating system.
• Google Kubernetes Engine (GKE): Valkey can be readily containerized using Docker and deployed as a Pod or a stateful application on a GKE cluster. This can be a well-suited approach for applications already utilizing Kubernetes, offering robust and advanced orchestration capabilities.

Memorystore for Valkey is a fully managed service that provides a high-performance, highly available, and scalable Valkey service in the cloud.

It provides the benefits of an in-memory database without the operational and financial burden of managing the underlying infrastructure.

Google Cloud takes care of the administrative workload, including patching, security, and scaling, so you can focus on building applications with real-time data.

• Fully managed: Google handles the operational overhead of managing Valkey instances, including provisioning, patching, and failover.
• High performance: Leveraging Google Cloud's infrastructure, Memorystore for Valkey delivers the low latency and high throughput required for demanding real-time applications—with sub-millisecond data access.
• High availability: Memorystore automatically replicates data across multiple zones, providing up to a 99.99% SLA.
• Scalable: Memorystore offers seamless scaling of Valkey instances without downtime to meet the changing demands of your application.
• Secure: Memorystore is protected with robust security features, including virtual private cloud (VPC) connectivity, 24/7 monitoring, and identity and access management (IAM).
• Perfect for AI applications: Memorystore for Valkey supports approximate nearest neighbor (ANN) vector search and exact nearest neighbor (KNN) vector search, making it an ideal low-latency datastore for generative AI.