From InfoQ · 40 stories
AWS Releases Open-Source File Mover Express for S3 Transfers
AWS launched File Mover Express, an open-source tool under Apache 2.0, for transferring large files between local systems and Amazon S3. It supports parallel transfers, automatic performance tuning, and resume capabilities, addressing challenges faced by media teams in moving terabytes of assets.
Spring Releases First Release Candidates for Boot 4.2, Framework 7.1, Data 2026.1, and More
Spring has released the first release candidates for multiple projects, including Spring Boot 4.2.0, Spring Framework 7.1.0, Spring Data 2026.1.0, and Spring Cloud 2026.0.0. These updates introduce bug fixes, dependency upgrades, and new features, with Spring Cloud Bus now integrated into Spring Cloud Stream.
Fortanix Integrates Confidential Containers with Attestation and Key Management
Fortanix has added support for Confidential Containers to its Confidential AI platform, linking Kubernetes workload attestation to encryption-key release. This integration allows workloads meeting security policies to decrypt models and sensitive data within hardware-based trusted execution environments.
LocalStack Azure Emulator Enters Public Preview, Lacks Security Enforcement
LocalStack launched its Azure emulator into public preview, offering a free, self-serve tool for local Azure development and testing. While it supports various Azure services and templates, the emulator currently stores but does not enforce many security and networking configurations, impacting its fidelity for certain testing scenarios.
DoorDash Reduces Grocery Reorder Latency to 2 Seconds Using AI Precomputation
DoorDash decreased the latency of its AI-powered grocery reorder flow from 16 seconds to 2 seconds by precomputing recurring purchase needs. This change doubled shopping list views and increased the order rate by approximately 34%, utilizing batch large language model inference and deterministic validation.
Netflix details GPU reliability platform for large-scale AI training and inference
Netflix developed a reliability platform for its large-scale GPU training and inference workloads, addressing hardware failures in distributed AI jobs. The platform uses layered health checks, automated remediation, and fault-tolerant training to maintain job continuity. This approach resulted in over 99.95% GPU-node availability and increased job success from 98% to over 99.5%.
Netflix Titus Adds Single-Tenant VM Isolation for Containers for Performance-Critical Workloads
Netflix has introduced a single-tenant mode for its Titus container platform, allowing specific tasks to run in isolation on dedicated EC2 instances. This change addresses performance interference issues in multi-tenant environments for stateful, high-throughput streaming workloads, while maintaining the same API and deployment path for application teams.
Microsoft Retires Azure Deployment Environments and Dev Box with No Direct Replacements
Microsoft is discontinuing Azure Deployment Environments by February 2027 and Microsoft Dev Box by September 2028. Azure Deployment Environments will have no direct successor, with Microsoft advising customers to use infrastructure-as-code tools, while Dev Box users are directed to Windows 365, which is not an equivalent service. This move removes a self-service developer environment offering, shifting the responsibility for catalog and portal management back to customers or third-party solutions.
Google OAuth Platform Adopts DPoP to Secure Refresh Tokens Against Theft and Replay Attacks
Google's OAuth Platform now supports Demonstrating Proof-of-Possession (DPoP) for refresh tokens, a security mechanism that cryptographically binds tokens to a client's asymmetric key pair. This change mitigates token theft and replay attacks by preventing stolen tokens from being replayed without the corresponding private key. The adoption of DPoP addresses the vulnerability of long-lived refresh tokens, which are a primary target for persistent online impersonation.
Datadog Releases Scout AI Model for SRE Production Problem Investigations
Datadog introduced Scout, an AI model designed to investigate production problems within its Bits Investigation SRE product. Scout aims to provide initial analysis of alerts and telemetry to identify root causes, reducing the cost and time associated with SRE investigations by acting as a first responder.
Airbnb Reduces Recommendation Staleness from Days to Seconds Using Chronon Updates
Airbnb updated its open-source Chronon feature platform with Push Mode and Near Real Time (NRT) Model Transform, reducing the staleness of recommendation embeddings from nearly two days to seconds. This change improved offline NDCG by 1.67% and increased uncancelled bookings by approximately 0.33% in A/B tests. The update allows guest activity to be reflected in recommendations almost immediately, improving the relevance of search results.
Chrome Trials Email Verification API to Eliminate Email-Based Verification Steps
Google has launched an origin trial in Chrome 150 for the Email Verification API, allowing websites to verify user email addresses without sending verification emails. This API aims to improve user experience by removing the need for users to switch contexts to their inbox during sign-up or login processes.
Cloudflare releases Quicksilver as Workers KV Instant, offering faster reads with new limits
Cloudflare has made its internal Quicksilver key-value store available to customers as Workers KV Instant, a new mode for Workers KV. This new mode offers significantly faster read times, with 99th percentile reads resolving in 1.62 ms, compared to 287 ms for classic KV. However, it introduces new limitations on storage size, key-value pairs, write rates, and higher costs for writes and storage.
Uber Introduces Normalized Compute Units for Consistent CPU Allocation Across Heterogeneous Fleet
Uber implemented Normalized Compute Units (NCUs) within its Odin orchestration platform to standardize CPU allocation for stateful workloads across its diverse server infrastructure. This system accounts for varying processor performance, ensuring consistent compute capacity regardless of the underlying hardware and preventing resource waste or under-provisioning.
AWS uses AI agents to triage Linux kernel CVEs for EC2 infrastructure
AWS engineers detailed their process for handling Linux kernel CVEs in EC2, which involves an automated filter and an AI agent. This system helps manage the increased volume of kernel CVEs since the kernel project became a CVE Numbering Authority.
OpenTofu 1.13 Improves Plan-Time Resolution of Unknown Values and Adds Linting
OpenTofu released version 1.13, introducing new functions like `convert` and `assumenotnull` to help module authors resolve expressions during the planning phase, particularly for `count` and `for_each` with unknown inputs. This version also adds experimental built-in linting rules and an opt-in Symbol Libraries feature, enhancing the platform's ability to handle complex configurations and improve code quality.
DigitalOcean Launches Learning and Data Services for AI Agent Improvement
DigitalOcean introduced Learning Services and Agent Data Services in public preview, enabling developers to build, evaluate, and continuously improve AI agents. These services allow agents to learn from interactions, identify weaknesses, test improvements, and utilize company data within DigitalOcean's AI-native cloud platform.
MCP C# SDK 2.0 Released, Making HTTP Stateless and Adding Multi Round-Trip Requests
Microsoft released version 2.0 of the official MCP C# SDK, implementing the 2026-07-28 revision of the Model Context Protocol (MCP) specification. This update makes Streamable HTTP servers stateless by default, standardizes HTTP headers for routing, and introduces Multi Round-Trip Requests (MRTR) for interactive tools, significantly changing how MCP operates over HTTP.
PingCAP Suffers Data Breach, Source Code and Customer Records Stolen
PingCAP confirmed unauthorized access to its internal systems, resulting in the theft of source code and customer-related data. The cyber extortion group FulcrumSec claimed responsibility, alleging they stole 10.6 TB of data including TiDB Cloud control-plane records, customer SQL statements, and API keys. This breach impacts PingCAP customers whose data, including credentials, may have been exposed.
Cloudflare Reduces Memory Footprint by 100 TB with Consistent Hashing Optimization
Cloudflare optimized its Rust implementation of consistent hashing in pingora-ketama, reducing the memory footprint of its internal load-balancing service by over 100 TB. This was achieved by compacting hash-ring entries and statistically analyzing hash distribution to decrease the number of hashes per backend by approximately 90% while maintaining load-balancing accuracy.
CSIRO Develops Serverless Beacon on AWS for Genomic Variant Querying
CSIRO developed Serverless Beacon (sBeacon), a serverless architecture on AWS that queries genomic variant data directly from Amazon S3 without requiring data copies. This system uses AWS Lambda, Amazon Athena, and Amazon DynamoDB to implement the GA4GH Beacon standard, enabling efficient genomic queries across large datasets while keeping data in its original compressed format.
NgRx v22 Released with Resource Extensions and Dynamic Deep Signals for Angular
NgRx, an Angular state management framework, released version 22, introducing resource extensions for customizing Angular's resource API and dynamic deep signals for union types. These updates address limitations in error handling and state modeling, providing more flexible and robust state management within Angular applications.
Vercel Labs Open-Sources fx, an Embeddable Zig Coding Agent
Vercel Labs released fx, an experimental coding agent and command-line tool written in Zig, designed for scripting and embedding rather than full-screen terminal use. It aims to provide a lightweight engine for developers building agent infrastructure, offering structured output for automation and integration with editors and JavaScript.
Docker Harbor Integrates Sandboxes for Isolated AI Agent Evaluation in MicroVMs
Docker integrated its Sandboxes with Harbor, an open-source AI agent evaluation framework, to run each evaluation task in an isolated microVM. This change improves performance and reproducibility by preventing agents from interfering with tests or each other, addressing challenges in evaluating untrusted AI-written code at scale.
Netflix Develops Workload Attestation for AWS EMR Spark Jobs to Bridge IAM and Internal PKI
Netflix created a workload attestation mechanism that allows Apache Spark jobs on Amazon EMR to exchange their AWS execution identity for an internal workload identity. This system connects AWS IAM with Netflix's Metatron PKI, enabling managed compute workloads to authenticate to internal services without manual credential injection.
Cilium 1.20 Expands Kubernetes Gateway API Support and Opens eBPF Datapath
Cilium 1.20 has been released, enhancing Kubernetes Gateway API support with features like ExternalAuth, ListenerSets, TCPRoute, and UDPRoute. This update allows for broader traffic management, including non-HTTP services and external authorization, and opens the eBPF datapath for third-party plugins, reducing the need for upstream contributions or forks for custom networking behaviors.
TornadoVM 7 Enables GPU-Native Workloads for Java Applications
TornadoVM, an open-source framework for accelerating Java programs on GPUs, released version 7. This update introduces architectural changes, including the use of the Foreign Function & Memory (FFM) API and removal of JVMCI dependency, alongside deeper integration with Nvidia's CUDA ecosystem and Apple's MXL library. This allows Java developers to offload data-parallel computations to GPUs, potentially improving performance for cloud and machine learning workloads.
Pinterest Implements Event Time Watermarks for Petabyte-Scale Data Partition Finalization
Pinterest introduced an event time-based partition finalization mechanism for its database ingestion framework, utilizing Apache Flink and Apache Iceberg. This allows downstream jobs to reliably determine when streaming data partitions are complete, addressing late-arriving data in continuous Change Data Capture (CDC) ingestion. The system supports petabyte-scale data from MySQL, TiDB, and KVStore, making online changes available within 15 minutes to an hour.
JEP 543 Finalizes Structured Concurrency in JDK 28 After Seven Preview Rounds
JEP 543, Structured Concurrency, has been finalized and integrated into JDK 28, making the StructuredTaskScope API a stable feature. This concludes the longest preview cycle for a Project Loom feature, providing a standardized way to manage groups of related subtasks as a single unit of work.
Buildkite Introduces Pipeline-Scoped Secrets and New Agentic CI Tools
Buildkite has released pipeline-scoped secrets and new Model Context Protocol (MCP) tools for CI/CD pipelines. These updates aim to improve secret management, pipeline validation, and build diagnostics within the platform.
NestJS 12 Released with ESM Packages, Standard Schema Validation, and Observability SDK
NestJS, a TypeScript framework for Node.js, released version 12, which transitions core packages to ESM, adds first-class Standard Schema validation, and introduces an official observability SDK. This update modernizes the framework's tooling and improves developer experience for validation and monitoring.
Google Replaces Pytype with Meta's Pyrefly for Python Type Checking
Google has switched from its internal Python type checker, Pytype, to Pyrefly, an open-source tool developed by Meta. This change was made due to Pyrefly's superior speed, accuracy, and lower infrastructure costs, addressing Pytype's challenges in keeping up with Python's evolving typing system.
AWS European Sovereign Cloud to Operate Without Global Backbone for Several Hours
AWS will temporarily disconnect its European Sovereign Cloud from the AWS Global Network backbone for several hours on October 24. This exercise aims to demonstrate the region's operational independence and align with EU cloud sovereignty frameworks, though it does not address legal sovereignty concerns.
Amazon Acquires DuckLabs, Integrates DuckDB into Amazon Aurora PostgreSQL
Amazon has acquired DuckLabs, the company behind the open-source analytical database DuckDB. DuckDB's creators will continue to lead the project within AWS, and the database will remain open-source under the MIT license, with the DuckDB Foundation maintaining supervision.
Google Open-Sources EnvHarness for Dynamic AI Agent Training Environments
Google researchers open-sourced EnvHarness, a framework that dynamically adapts agent-training environments to an agent's weaknesses without altering the agent or environment. This framework aims to improve learning effectiveness and efficiency by allowing agents to continuously learn from evolving environments.
Zod 4.5 and 4.6 Introduce In-Process Schema Compilation and Memory Reduction
Zod, a TypeScript schema validation library, released versions 4.5 and 4.6, adding in-process schema compilation for faster parsing and significantly reducing memory footprint. These updates improve performance and memory efficiency for schema validation in TypeScript applications.
Uber Automates Service Mesh Dependency Analysis for Failure Propagation
Uber developed an automated system to map service dependencies and classify failure propagation within its service mesh. This system uses RPC middleware to correlate caller and callee failures, identifying dependencies as fail-close, fail-open, or unknown, which helps engineers quickly diagnose the root cause of request failures in a microservices environment.
Helidon 27 Released with Restructured Programming Models and New Modules
Helidon 27 has been released, adopting Oracle's Tip & Tail Model for library development, aligning with JDK 27. This version introduces restructured programming models, including Helidon Core with Imperative and Declarative models, and new Helidon Messaging and Helidon Data JDBC modules. The changes aim to offer both bleeding-edge Java features and a stable baseline for development teams.
AWS releases Strands Decider 2B and Strands Box for AI agent development
AWS introduced Strands Decider 2B, a 2-billion-parameter model for decision-making within AI agent workflows, and Strands Box, a sandbox for controlling agent execution. These open-source projects provide tools for building AI agents that interact with external systems and execute code in a controlled environment.
Instagram Direct Migrates Android UI to Jetpack Compose, Cuts AI Token Costs by 33%
Instagram Direct migrated its Android UI to Jetpack Compose, creating an AI-native architecture. This change reduced UI code volume by 50% and decreased AI agent token costs per session by 33%. The migration focused on designing the codebase to guide AI agents towards producing correct and high-quality code.