Serverless Functions

Functions triggered by events

A function is code that runs in response to an event: an HTTP request (through API Gateway or a function URL), a file uploaded to storage, a message on a queue, a schedule. The platform starts as many copies as needed, in parallel, and you pay per request and per GB-second of execution.

def handler(event, context):                 # AWS Lambda, Python
    order = event["detail"]
    save(order)
    return {"status": "ok"}

Things to know

• Cold starts: the first request on a new instance loads the runtime and your code; keep packages small and initialise clients outside the handler so warm invocations reuse them.
• Limits: Lambda runs at most 15 minutes per invocation, with up to 10 GB of memory; long jobs belong in containers or step-by-step workflows (Step Functions, Durable Functions, Workflows).
• Concurrency: each concurrent request uses one execution environment; accounts have concurrency quotas, and downstream systems (databases) may not cope with thousands of parallel connections.
• Statelessness: keep state in databases, caches or storage, not in the function.

PRICE_PER_GB_SECOND = 0.0000166667      # example Lambda price (x86), USD
PRICE_PER_MILLION_REQUESTS = 0.20

def monthly_cost(requests, avg_ms, memory_mb):
    gb_seconds = requests * (avg_ms / 1000) * (memory_mb / 1024)
    return gb_seconds * PRICE_PER_GB_SECOND + requests / 1_000_000 * PRICE_PER_MILLION_REQUESTS

for requests in [1_000_000, 50_000_000, 500_000_000]:
    print(f"{requests:>11,} requests/month at 120 ms, 512 MB: ${monthly_cost(requests, 120, 512):9,.2f}")

  1,000,000 requests/month at 120 ms, 512 MB: $     1.20
 50,000,000 requests/month at 120 ms, 512 MB: $    60.00
500,000,000 requests/month at 120 ms, 512 MB: $   600.00

For deep dives, see the Python, Node.js and Java serverless courses.