Title

AWS re:Invent 2022 - Samsung SmartThings powers home automation with Amazon MemoryDB (DAT215)

Summary

Samsung SmartThings uses Amazon MemoryDB to power their home automation and IoT platform.
SmartThings was founded in 2012 and acquired by Samsung in 2014, becoming a key part of Samsung's IoT strategy.
The platform has evolved from a monolithic application to over 200 microservices, supporting thousands of devices and processing billions of events daily.
SmartThings faced challenges with their shard-based architecture and self-hosted RabbitMQ, which were difficult to scale and maintain.
Industry initiatives like Matter and Samsung's Hub Everywhere are expected to significantly increase the number of IoT devices and hubs, necessitating a scalable and reliable architecture.
SmartThings adopted a cell-based architecture and replaced RabbitMQ with Amazon MemoryDB for Redis, which offers fast performance, scalability, reliability, and ease of maintenance.
MemoryDB for Redis is used as a primary database and for event streaming, with data structures like Redis streams, hashes, and sorted sets.
The new architecture involves breaking shards into smaller cells, allowing for controlled growth and easier traffic management.
MemoryDB for Redis ensures data durability and high availability, with multi-AZ data durability and automatic failovers.
The session included demonstrations of the new platform and architecture, highlighting the performance and reliability improvements.

The transition from a shard-based architecture to a cell-based one allows SmartThings to manage growth effectively and reduce the impact of failures on users.
By using Amazon MemoryDB for Redis, SmartThings can handle a massive increase in traffic and device connections expected from industry initiatives without compromising performance.
MemoryDB for Redis's durability and managed service features relieve SmartThings developers from infrastructure maintenance, allowing them to focus on feature development.
The use of Redis streams as a replacement for RabbitMQ messaging layer indicates a shift towards more modern, scalable, and managed event-driven architectures.
The session highlighted the importance of planning for future growth and industry changes when designing IoT platforms, ensuring that the architecture can adapt to new standards and user demands.
The partnership with AWS and the use of AWS services like MemoryDB for Redis align with SmartThings' goals for cost efficiency and energy consciousness, as they can scale resources according to demand.
The insights from the session can be valuable for other companies facing similar scaling challenges in the IoT space, demonstrating the benefits of leveraging cloud services and modern database technologies.