AWS — OffNet Newsroom — Sunday, September 13, 2026

InfoQ 2026-09-12 generaldevops

AWS Lambda SnapStart Now Supports Container Images Up to 10 GB

AWS has extended Lambda SnapStart support to container image functions, removing the previous 250 MB zip archive limit. This allows teams to retain full dependency headroom and source code without stripping whitespace or docstrings to fit size constraints. The update enables sub-second startup times for larger container-based workloads that were previously compromised by storage limits.

SnapStart now applies to container images with up to 10 GB storage, replacing the 250 MB zip limit.
Developers can keep full dependencies and source code without aggressive stripping to meet size caps.
Sub-second cold starts are now viable for larger container functions without sacrificing code integrity.

Read at source →

AWS Database Blog 2026-09-11 awsdatabase ↺ since 09-12

AWS Zero-ETL: Run DuckDB SQL Analytics on DynamoDB via Iceberg on S3

AWS introduces a zero-ETL integration that replicates Amazon DynamoDB tables into Apache Iceberg format on Amazon S3. An AWS Lambda function running DuckDB processes these Iceberg tables, exposing SQL query capabilities through an IAM-authorized Lambda function URL. This enables ad hoc analytics on DynamoDB data without building custom ETL pipelines.

Eliminates custom ETL by auto-replicating DynamoDB to Apache Iceberg on S3.
Leverages DuckDB for high-performance ad hoc SQL queries on NoSQL data.
Access is secured via IAM-authorized Lambda function URLs.
Enables analytics workloads without moving data out of the AWS ecosystem.

Read at source →

OpenAI News 2026-09-11 llmaiagents ↺ since 09-12

OpenAI scales Habitat storage to 1B users and 22M req/s

OpenAI has transformed its internal Python library, Habitat, into a globally distributed storage platform capable of handling over one billion ChatGPT users. The system is currently processing approximately 22 million requests per second to support this massive scale. This evolution represents a significant architectural shift from a localized library to a global infrastructure component.

Habitat evolved from a Python library to a global distributed storage system
Platform now serves over 1 billion ChatGPT users globally
System throughput reaches approximately 22 million requests per second
Demonstrates OpenAI's internal infrastructure scaling capabilities

Read at source →

InfoQ 2026-09-11 generaldevops ↺ since 09-12

Form3’s Triple-Active Multi-Cloud Strategy: When It Works and When It Doesn't

Form3 engineers detail their migration from a single-cloud model to a triple-active architecture spanning the UK, Europe, and US. The approach relies on CockroachDB for distributed data consistency and NATS for messaging, supported by custom Kubernetes operators to manage infrastructure complexity. The presentation highlights the operational trade-offs of maintaining cross-cloud networking and meeting distinct regional disaster recovery requirements.

Triple-active setups require rigorous cross-cloud networking and latency management.
CockroachDB enables data distribution across distinct geographic regions.
Custom Kubernetes operators are essential for managing multi-cloud complexity.
Regional DR expectations vary significantly across UK, EU, and US markets.
Multi-cloud adds significant operational overhead compared to single-cloud.

Read at source →

AWS What's New 2026-09-10 awsdatabase ↺ since 09-11

Amazon Redshift RG instances now available in Europe (Zurich)

AWS has extended the availability of Amazon Redshift RG instances, powered by Graviton processors, to the Europe (Zurich) Region. These new instances offer up to 2.4x better performance than RA3 generations at a 30% lower cost per vCPU. The system utilizes a vectorized query engine to process Apache Iceberg and Parquet data directly on cluster nodes.

RG instances in Zurich deliver up to 2.4x speedup over RA3 with 30% lower cost per vCPU.
Unified engine processes data warehouse and data lake formats like Iceberg and Parquet.
Available in rg.large, rg.xlarge, rg.4xlarge, and rg.12xlarge sizes.
Existing RA3 clusters can migrate to RG using Snapshot & Restore or resize operations.

Read at source →

AWS What's New 2026-09-09 awsdatabase ↺ since 09-10

AWS Lambda extends timeout to 90 minutes on Managed Instances

AWS Lambda now allows 90-minute execution limits for asynchronous and event source mapping invocations on Lambda Managed Instances, a sixfold increase from the prior 15-minute cap. This change enables direct execution of long-running tasks like media transcoding, Monte Carlo simulations, and AI inference without architectural workarounds. The update specifically targets data-intensive workloads that previously required complex splitting or migration strategies.

Timeout limit increased from 15 to 90 minutes on Lambda Managed Instances only.
Applies to both asynchronous invocations and event source mappings.
Eliminates need for complex splitting or migration for long-running batch jobs.
Supports heavy workloads like AI inference and financial simulations natively.
No re-architecture needed for existing applications requiring extended execution time.

Read at source →

InfoQ 2026-09-11 generaldevops ↺ since 09-12

Terraform AWS Provider v6.62.0 Adds New Capabilities and Improves State Management

Terraform AWS Provider v6.62.0 introduces support for emerging AWS infrastructure features while refining how the tool interprets and manages existing state. This release addresses growing complexity in AWS environments by enhancing provider stability and coverage. The update ensures practitioners can leverage the latest cloud services without manual workarounds.

Update to v6.62.0 to access newly supported AWS resources and services.
Benefit from improved state analysis for better drift detection and management.
Simplify IaC workflows as AWS infrastructure complexity increases.
Ensure compatibility with latest AWS API changes via provider updates.

Read at source →