AWS Weekly Roundup: AgentCore Runtime V2, S3 Tables Hive Migration, and DynamoDB Bulk Put
·2026.11.10 02:55
Key point
The new runtime uses snapshots to keep cold starts at 2 seconds for images up to 2 GB, while S3 Tables now migrates Hive data to Iceberg in three steps.
Details
- Amazon Bedrock AgentCore Runtime V2 reduces P75 cold starts to approximately 2 seconds for container images between 200 MB and 2 GB, compared to 5.4 to 30 seconds for the original runtime.
- Amazon S3 Tables introduces managed migration for Apache Hive tables to Apache Iceberg, allowing production workloads to remain writable until cutover with no charge for the conversion.
- Amazon DynamoDB supports bulk put items from S3, enabling asynchronous loading of line-delimited JSON or Amazon Ion files into existing tables with dedicated system capacity.
- AWS Interconnect launches a last-mile preview with Equinix, offering 1 Gbps private connections at no charge in San Jose and Northern Virginia with a 99.99% availability SLA.
AgentCore Runtime V2
- Memory management pages in on demand and releases memory when sessions go cold, reducing costs for idle agents.
- Opt-in requires setting platformVersion to V2; containers must report healthy on /ping within 120 seconds.
- Available in 10 Regions, with committed baseline pricing expected soon.
Data and Networking Updates
- S3 Tables migration validates sources and produces compatibility reports, supporting queries via Amazon Athena or any Iceberg REST API engine.
- DynamoDB bulk loads use an Accelerated tier for deadline-bound jobs, with illustrative times under 1 hour for 100 million 1 KB items.
- AWS Interconnect provides quad-redundant capacity with BGP and VLANs configured by AWS, using MACsec by default.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.