Why does ClickMasters use Fargate instead of Lambda for NHS patient-facing services?

NHS patient-facing cold start requirement: NHS clinical applications must respond within acceptable clinical latency. Lambda cold start (200ms-2 seconds on first invocation after inactivity) is unacceptable for: (1) NHS GP Connect real-time prescriptions (GP waiting room -- 2-second cold start breaks clinical workflow), (2) NHS 111 triage (urgent care -- cold start adds critical seconds), (3) NHS emergency department systems (A&E -- cold start is patient safety risk). ECS Fargate always-warm containers: Fargate tasks are always running -- no cold start, consistent 20-50ms response time. ClickMasters DTAC compliance: Fargate is the standard for NHS clinical applications. Lambda for NHS: acceptable for background processing (ILR report generation, overnight data quality checks, scheduled ESR imports) where latency is not clinically critical.

When should a UK FinTech use Lambda instead of Fargate?

Lambda is right for UK FinTech when: (1) event-driven processing (GoCardless webhook received -- Lambda triggered by SQS -- process payment event without always-on container), (2) scheduled reporting (FCA GABRIEL monthly submission -- Lambda triggered by EventBridge Scheduler), (3) S3 event processing (transaction CSV uploaded -- Lambda triggered by S3 notification -- import and validate), (4) image/document processing (ID document uploaded for KYC -- Lambda + Rekognition), (5) low-traffic internal tools (FCA SMCR log search -- Lambda + DynamoDB -- not IBS, no PS21/3 monitoring required). Lambda is WRONG for FinTech IBS: FCA PS21/3 IBS requires reliable availability monitoring -- Lambda cold starts make CloudWatch Synthetics IBS availability monitoring unreliable. ClickMasters PS21/3 rule: all IBS services on Fargate, non-IBS event processing on Lambda.

Lambda Vs Fargate Vs App Runner Uk Serverless

AWS Lambda vs Fargate vs App Runner -- UK Serverless (2026)

Updated: June 20269 min read

UKGBPComparisonUK GDPRUK Law

Direct Answer

ECS Fargate is ClickMasters default for UK regulated production workloads (NHS patient-facing, FCA IBS, GDS live services): no cold starts (critical for clinical applications), predictable latency, NHS DTAC compliance evidence, and FCA PS21/3 IBS availability. Lambda is right for event-driven background jobs (S3 triggers, SQS consumers, scheduled tasks). App Runner is right for simple containerised APIs with automatic scaling and zero ECS operational overhead. Compute choice affects NHS DTAC Domain 3 evidence, FCA PS21/3 IBS availability, and UK GDPR data residency in AWS eu-west-2.

Factor	ECS Fargate	AWS Lambda	AWS App Runner
NHS DTAC clinical application (no cold start)	BEST -- no cold start, consistent latency	Cold start 200ms-2s -- not acceptable for clinical	Auto-scaling but App Runner has warm containers
FCA PS21/3 IBS availability	BEST -- CloudWatch Synthetics monitoring, multi-AZ	Lambda retries on failure -- IBS availability harder to monitor	App Runner has health checks -- PS21/3 possible
UK GDPR data residency (eu-west-2)	BEST -- VPC private subnets, eu-west-2	eu-west-2 available -- but VPC Lambda adds latency	eu-west-2 available
Cost (light traffic, < 100 req/min)	Higher -- Fargate tasks always running	BEST -- pay per invocation, free tier	Low -- App Runner scales to zero (v2 feature)
Cost (heavy traffic, > 1000 req/min)	Predictable -- flat rate per task	Can be expensive -- per-invocation at scale	Predictable -- per vCPU/memory hour
Operational overhead	Medium -- ECS task definitions, CDK	Low -- just function code + trigger	LOWEST -- container image, no task config
NHS DTAC Domain 3 VPC isolation	BEST -- private subnets, security groups	VPC Lambda available but adds cold start	Limited VPC integration
Event-driven background jobs	Overkill -- use Lambda	BEST -- S3 triggers, SQS, EventBridge, scheduled	Not designed for event-driven
ClickMasters default	Patient-facing NHS, FCA IBS, GDS live services	S3/SQS event processing, scheduled jobs, ML inference	Simple containerised APIs, internal tools