tag: Serverless · 15 items
- Platform/SRE — Learn: New GA observability plugin for Lambda durable functions exports execution history to 13 destinations including CloudWatch, S3, and OpenSearch; worth noting when designing telemetry for serverless workflows, but installed at the function-code level so no platform infrastructure change required.
- CI/CD — Skip
- Leader — Learn: New no-cost AWS capability for observability into long-running Lambda workflows; relevant context for teams standardizing on serverless architectures, but no licensing, cost, or vendor-risk decision is triggered.
- Platform/SRE — Learn: Lambda MicroVMs (GA, VM-level isolation with state preservation) is now available in Hong Kong, Malaysia, Thailand, and Spain — relevant if you’re architecting isolated code-execution environments for users in those regions, but no action required unless you have a workload targeting those geos.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: GA expansion of Lambda durable functions to GovCloud adds a new serverless orchestration primitive for multi-step workflows; worth noting for architecture decisions if your platform serves government or regulated-industry tenants in those regions, but no operational change required today.
- CI/CD — Skip
- Leader — Learn: If your org serves government or regulated-industry customers in GovCloud, this broadens available serverless orchestration options for compliant AI and workflow use cases — worth factoring into platform strategy conversations but no decision is pending.
- Platform/SRE — Plan: If your platform runs multi-step or agentic Lambda workflows, the 30,000 operation limit and 1 MB checkpoint open up architectures previously blocked by quota; evaluate whether existing workflows hitting the old 3,000-op ceiling should be redesigned to take advantage.
- CI/CD — Skip
- Leader — Learn: The expanded limits make Lambda durable functions more viable for long-running agentic AI orchestration, worth tracking if your org is evaluating serverless vs. container-based workflow engines for AI workloads.
- Platform/SRE — Learn: New GA AWS Lambda primitive supporting direct HTTP handling (Express/Hono, no adapters) with a different billing model; worth tracking for future platform architecture patterns, but no existing infrastructure needs changing and existing Lambda Event Functions require no migration.
- CI/CD — Skip
- Leader — Plan: The active-CPU billing model (vs. idle CPU) could meaningfully reduce Lambda costs for I/O-bound workloads; evaluate this as a golden-path serverless option and assess cost-optimization potential against current Lambda spend this planning cycle.
- Platform/SRE — Learn: Regional expansion of a GA serverless orchestration feature; relevant if you operate workloads in the EU Sovereign Cloud region and are evaluating durable workflows as an alternative to Step Functions or similar.
- CI/CD — Skip
- Leader — Learn: If your org has EU data-residency or sovereignty compliance requirements, this expands the serverless orchestration options available within the sovereign boundary — worth noting for future architecture decisions.
- Platform/SRE — Plan: If your platform uses Azure Functions with PowerShell workers, plan migration to 7.6 this quarter to stay ahead of older runtime EOL cycles; no forced-upgrade date announced yet.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Learn: New GA capability for end-to-end TLS on Azure Functions Flex Consumption; worth noting if your platform uses this hosting model, but no migration deadline or breaking change.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Plan: This GA capability removes a key architectural constraint that pushed long-running workloads off Lambda toward ECS/Batch/Step Functions; evaluate whether existing workarounds in the platform’s serverless stack can be simplified this quarter.
- CI/CD — Skip
- Leader — Learn: The 6x timeout increase changes the serverless vs. dedicated-compute trade-off for batch and AI inference workloads, worth factoring into the next platform strategy review, but no urgent decision is required.
- Platform/SRE — Learn: New AI-assisted diagnostic capability for Lambda and connected services (API Gateway, EventBridge, DynamoDB, etc.) is worth evaluating if your team uses AI coding agents, but it requires no platform-side change today.
- CI/CD — Skip
- Leader — Learn: AWS expanding MCP-based AI agent access to operational data is a directional signal for AI-assisted platform tooling strategy, but no procurement or standards decision is required yet.
- Platform/SRE — Plan: GA capability that reduces container image cold starts to sub-second by snapshotting initialized environments; worth adopting this quarter for latency-sensitive Lambda workloads (ML inference, interactive APIs) running on container images.
- CI/CD — Skip
- Leader — Learn: SnapStart extending to container images changes the cost/performance tradeoff for serverless-based architectures, informing build-vs-buy and platform strategy discussions around cold-start optimization.
- Platform/SRE — Learn: This default-on guardrail stops runaway Lambda recursion via S3/SQS/SNS and sends Health Dashboard alerts; worth knowing if you operate Lambda at scale, and note that intentional recursive patterns now require explicit opt-out via PutFunctionRecursionConfig.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Lambda MicroVMs is now GA in Frankfurt, Stockholm, Mumbai, Singapore, and Sydney, giving platform teams a managed VM-isolation primitive for multi-tenant or AI workload sandboxing without managing hypervisors. No action required — worth evaluating if latency or data-residency in these regions is a current pain point.
- CI/CD — Skip
- Leader — Learn: The regional expansion signals AWS maturing Lambda MicroVMs as a serious compute tier for AI coding assistants and sandboxed execution — worth tracking as a potential building block for internal developer platform strategy, though no decision is needed today.
- Platform/SRE — Learn: Public preview of serverless JavaScript execution at the Azure Front Door edge layer — worth evaluating if you run AFD as your ingress/CDN layer, but pre-GA so no action yet.
- CI/CD — Skip
- Leader — Skip
- Signals: pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Plan: This GA capability removes the per-Region Lambda-managed storage quota for teams running large function/layer footprints; evaluate adopting
S3ObjectStorageMode=REFERENCEthis quarter for deployments approaching the old 75GB ceiling, and note the default limit has already been raised to 300GB for all accounts. - CI/CD — Skip
- Leader — Learn: Cost impact is neutral-to-positive (standard S3 rates replace implicit Lambda storage overhead) with no forced migration, but worth flagging to platform teams running high function counts so they can evaluate whether S3-backed storage fits their existing artifact management posture.