Serverless & Containers

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.

AWS Lambda can now mount S3 buckets as a POSIX file system. At roughly $0.023 per GB-month for large files it is about 13× cheaper than EFS — but a 60-second write-back delay, broken advisory locks, and atomic-rename quirks will break naive ports. Here is when to use it, when to wait, and how to wire it up safely.

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.

EdTech traffic doesn't curve — it spikes at 9am Monday and 7pm Tuesday and the load test never sees the right shape. Serverless architectures for LMS, assessments, video delivery, and AI-powered learning that scale to millions of students without paying for them on weekends.

ECS is "AWS-native containers." EKS is "Kubernetes, but you're still on the hook for everything Kubernetes." A decision guide for picking between ECS and EKS based on team Kubernetes experience, operational complexity, and the cost gap at production scale.

Step Functions is the AWS service most teams under-use until they need it badly. Patterns for sequential pipelines, parallel fan-out, error handling, human approval workflows, and the cost optimisations that keep state-transition bills predictable.

Single-table design is either DynamoDB's superpower or a six-month rewrite waiting to happen. SaaS access patterns, tenant isolation, GSI strategies, and the patterns that make DynamoDB the right serverless database for multi-tenant apps — plus the patterns that get teams in trouble.