awsServerless compute

AWS Lambda

Runs a function in response to an event without any server to provision. Billed per request and per GB-second of execution, scaling to zero when idle.

Limits

  • Function timeout is capped at 900 seconds (15 minutes) and cannot be increased

  • Memory is configurable from 128 MB to 10,240 MB in 1 MB increments, and CPU is allocated in proportion — at 1,769 MB a function has the equivalent of one vCPU

  • The /tmp directory provides between 512 MB and 10,240 MB of ephemeral storage, configurable in 1 MB increments

  • Asynchronous invocation payloads are limited to 1 MB, while synchronous request and response payloads are 6 MB each

  • Default concurrent executions per Region is 1,000, and each function scales by 1,000 execution environments every 10 seconds

Pricing model

Per request plus GB-seconds of execution time, where GB-seconds are duration multiplied by configured memory. Raising memory therefore raises the per-millisecond rate but often shortens duration, so cost is not monotonic in memory.

Cross-provider equivalents

  • Google Cloud Run

    How it differs: Cloud Run allows a request timeout up to 60 minutes (3,600 seconds) against Lambda's hard 15-minute (900 second) ceiling, so a long-running transcode or batch job that must be split into steps on Lambda often runs unchanged on Cloud Run.