AWS Local Zones
An infrastructure extension that places AWS compute, storage, and database services closer to major metropolitan areas, delivering single-digit millisecond latency to end users
Overview
AWS Local Zones extend AWS Region infrastructure closer to major metropolitan areas. By running AWS services such as EC2, EBS, ECS, EKS, RDS, and ElastiCache in locations physically closer to end users, they deliver single-digit millisecond latency. Since Local Zones work by extending a parent Region's VPC into Local Zone subnets, they integrate naturally into existing architectures. As of 2024, Local Zones are deployed in over 30 cities worldwide.
The Latency Problem Local Zones Solve
When serving users in Osaka from the Tokyo Region (ap-northeast-1), the network round-trip time is typically 10-20 milliseconds. While this delay is acceptable for most applications, real-time gaming, live video editing, financial trading, and AR/VR applications demand single-digit millisecond latency. Placing a Local Zone near users can reduce round-trip time to 1-5 milliseconds. A Local Zone is created as a subnet within the parent Region's VPC, so it can communicate with other resources in the VPC via private IP. The typical architecture places latency-sensitive frontend processing (web servers, API servers, caches) in the Local Zone while keeping backend processing (databases, batch jobs) in the parent Region. Enabling a Local Zone requires only opting in from your account settings - no additional contracts are needed.
Available Services and Design Constraints
The services available in a Local Zone are a subset of the parent Region. Key available services include EC2 (limited instance types), EBS (gp2, gp3, io1), ECS, EKS, ALB, RDS (select engines), and ElastiCache. S3 cannot be placed in a Local Zone; you use the parent Region's S3 instead. Available instance types vary by Local Zone - some offer the latest GPU instances while others provide only general-purpose instances. A key design constraint is that a Local Zone consists of a single data center, so it does not provide AZ-level redundancy. For high availability, you need to design a failover architecture combining multiple Local Zones or the parent Region's AZs. The recommended pattern is configuring Auto Scaling groups to span both Local Zone subnets and parent Region subnets, falling back to the parent Region if the Local Zone fails. For a deeper understanding of edge computing architectures, books on edge computing (Amazon) are a great resource. Data transfer charges between a Local Zone and its parent Region are equivalent to inter-AZ transfer rates.
Media and Entertainment Industry Use Cases
One of the industries where Local Zones see the most adoption is media and entertainment. The Los Angeles Local Zone is designed for Hollywood film and television production studios, enabling low-latency VFX rendering, video encoding, and real-time color grading on GPU instances (g4dn, g5). Studios can connect their on-premises environments to the Local Zone via Direct Connect, transferring large video assets at near-local speeds while leveraging cloud scalability in a hybrid configuration. In the gaming industry, multiplayer game matchmaking servers and game session servers are placed in Local Zones to deliver low-latency gaming experiences to players. Combined with GameLift, dynamic server placement that routes sessions to the nearest Local Zone based on player geographic location is also possible. In Japan, the Osaka Local Zone is available and can be used to improve latency for users in the Kansai region.