Blog category
Guides on Lambda, ECS, Fargate, and EKS — from serverless APIs to container orchestration, autoscaling, and production deployment patterns on AWS.
On 9 Sep 2026 AWS raised LMI async and ESM timeout from 15 to 90 minutes (5,400 s). A 90-minute SQS consumer needs at least 9 hours of visibility. Sync and classic on-demand Lambda stay at 15 minutes.
EKS Kubernetes 1.36 is available with user namespaces GA and mutating admission policies. Since July 2026 you can roll back an in-place control-plane upgrade within 7 days at no extra charge — but only N→N-1, and Fargate or strict PDBs can block you.
On July 23, 2026 AWS GA’d the Durable Execution SDK for .NET (Amazon.Lambda.DurableExecution). Score your C# workflows against our durable-vs-Step-Functions matrix — most integration-heavy flows still belong in ASL.
After Mar–May 2026 LMI updates (32 GB / 16 vCPU, 4,096 FDs, EventBridge scheduled scaling), a B2B analytics API (~40k peak RPM, 9–6 UTC) cut idle capacity cost ~62% by scheduling MinExecutionEnvironments 20→3 off-peak.
For a product SaaS (~12 production nodes, 3 clusters), EKS Auto Mode cut internal K8s ops from 80 to 35 hours/month at similar compute spend — partner MSP quote was $6,500/mo on top for 24/7 paging alone.
AWS shipped ECS Express Mode on November 21, 2025 — three inputs (image + two IAM roles) and Express Mode provisions Fargate, ALB, HTTPS, auto scaling, and a *.ecs.*.on.aws URL. Up to 25 services can share one ALB. No Express Mode surcharge.
As of July 2026, Lambda S3 Files still mounts S3 as POSIX via NFS — roughly S3 Standard storage rates vs EFS for large files — but versioning is required, IAM is s3files:*, capacity providers are incompatible, and a ~60s write-back window breaks naive ports.
A deep technical guide to running PHP, Python, and Node.js applications on Amazon ECS in production — covering Fargate vs EC2, FrankenPHP vs Nginx+FPM, multi-container task patterns, zero-downtime deployments, and observability.
Most auto-scaling configs end up reactive — scaling after load hits, not before. The patterns that anticipate demand instead of chasing it across EC2, ECS, and Lambda, with the cost trade-offs of target tracking, step scaling, and predictive scaling.
SQS is "reliable" only if you understand visibility timeout, dead-letter queues, and the silent failure mode where messages get processed twice. Standard vs FIFO, DLQ tuning, Lambda integration, and the patterns that turn SQS from "queue with default settings" into reliable backbone messaging.
Event-driven sounds elegant in design docs. Then a malformed event ships to a critical bus and you spend a weekend debugging fan-out. EventBridge buses, rules, schema discovery, cross-account patterns, and the failure modes you only learn about in production.
July 2026: EdTech serverless on AWS — exam-day spikes, SQS grading buffers, CloudFront video egress, and a readiness checklist so load tests match Monday 9am reality.