AWS just announced several new Local Zones, expanding their edge network significantly. The stated goal is single-digit millisecond latency for latency-sensitive applications. While the marketing is predictably vague, I'm interested in the concrete implications for data workloads, particularly around hybrid architectures and real-time analytics.
From a data infrastructure perspective, Local Zones sit between a full AWS Region and a simple Outpost. They offer core services like EC2, EBS, VPC, and crucially for us, **Amazon RDS**. The new ones (like Delhi, Helsinki, Barcelona) target specific metropolitan areas lacking a full region.
My immediate questions for benchmarking would be:
* **Database Latency:** For an application in, say, Denver, what is the measurable latency difference for OLTP queries hitting an RDS instance in us-west-2 (Oregon) versus a Denver Local Zone? We need real `SELECT` query round-trip times, not just ping.
* **Cost Premium:** Local Zone resources carry a price premium over standard regions. Is this a 10%, 25%, or 50% uplift for comparable RDS instance types? The TCO for a low-latency analytics serving layer needs this data.
* **Service Limitations:** Key services like Redshift, Athena, or managed Kafka (MSK) are not listed in the Local Zone service tables. This forces a split architecture: hot data in RDS/EC2 at the edge, cold data and heavy ETL in the parent region.
This feels like a potential game-changer for specific use cases: real-time dashboard front-ends needing sub-100ms response times, or hybrid setups where on-prem data must join with cloud data with minimal lag. But it's likely niche for pure cloud-native analytics stacks, as the limited service catalog and cost increase are significant constraints.
Has anyone run performance tests or cost comparisons for data workloads on existing Local Zones (like LA or Boston)? I'm particularly interested in seeing:
- Network latency benchmarks from common monitoring tools.
- A breakdown of RDS instance costs in a Local Zone vs. its parent region.
- Any practical experience with data synchronization challenges between Local Zone databases and central data warehouses.