Lambda Function Configuration (Runtime, Memory, Timeout)

Configuring a Lambda function correctly is the difference between a performant, cost-efficient function and one that times out, runs out of memory, or costs more than necessary. This lesson covers the key configuration options — runtime selection, memory and CPU allocation, timeout settings, concurrency controls, and deployment packaging.

---

Runtime Selection

The runtime determines which programming language and version your function uses. AWS provides managed runtimes that include the language interpreter, SDK, and operating system dependencies.

Available Managed Runtimes

Runtime Identifier	Language	OS	End of Support
nodejs20.x	Node.js 20	Amazon Linux 2023	TBD
nodejs18.x	Node.js 18	Amazon Linux 2	2025-09
python3.12	Python 3.12	Amazon Linux 2023	TBD
python3.11	Python 3.11	Amazon Linux 2	TBD
java21	Java 21 (Corretto)	Amazon Linux 2023	TBD
java17	Java 17 (Corretto)	Amazon Linux 2	TBD
dotnet8	.NET 8	Amazon Linux 2023	TBD
ruby3.3	Ruby 3.3	Amazon Linux 2023	TBD
provided.al2023	Custom runtime	Amazon Linux 2023	TBD

Choosing a Runtime

Consider these factors:

Decision Matrix:

Team expertise ---------> Use the language your team knows best
Cold start tolerance ---> Node.js/Python for low latency; Java/C# for throughput
Ecosystem needs --------> Java for enterprise libraries; Python for data/ML
Max performance --------> Custom runtime with Go or Rust (compiled binaries)

Custom Runtimes

If your language is not natively supported, use the provided.al2023 runtime with a bootstrap file:

#!/bin/sh
# bootstrap — the entry point for custom runtimes
set -euo pipefail

# Initialisation
# ...

# Processing loop
while true; do
  # Get next invocation event from Lambda Runtime API
  HEADERS=$(mktemp)
  EVENT_DATA=$(curl -sS -LD "$HEADERS" \
    "http://$\{AWS_LAMBDA_RUNTIME_API}/2018-06-01/runtime/invocation/next")

  REQUEST_ID=$(grep -Fi Lambda-Runtime-Aws-Request-Id "$HEADERS" | tr -d '[:space:]' | cut -d: -f2)

  # Process the event (call your binary/script)
  RESPONSE=$(./my-handler "$EVENT_DATA")

  # Send response
  curl -sS -X POST \
    "http://$\{AWS_LAMBDA_RUNTIME_API}/2018-06-01/runtime/invocation/$REQUEST_ID/response" \
    -d "$RESPONSE"
done

---

Memory Configuration

Memory is the single most important performance lever for Lambda functions. It controls not just RAM but also proportional CPU, network bandwidth, and disk I/O.

Memory-to-CPU Relationship

Memory        vCPU Allocation       Network Bandwidth
128 MB   ---> Fraction of 1 vCPU    Low
512 MB   ---> ~0.3 vCPU             Moderate
1024 MB  ---> ~0.6 vCPU             Moderate
1769 MB  ---> 1 full vCPU           High
3538 MB  ---> 2 vCPUs               High
5307 MB  ---> 3 vCPUs               Very high
7076 MB  ---> 4 vCPUs               Very high
8845 MB  ---> 5 vCPUs               Very high
10240 MB ---> 6 vCPUs               Maximum

Key insight: At 1,769 MB, you get one full vCPU. Below this, CPU-bound tasks will be throttled. Above this, Lambda allocates multiple vCPUs (but your code must be multi-threaded to benefit).

Memory Range

• Minimum: 128 MB
• Maximum: 10,240 MB (10 GB)
• Increment: 1 MB steps

Right-Sizing Memory with AWS Lambda Power Tuning

The open-source AWS Lambda Power Tuning tool tests your function at different memory settings and plots cost vs. duration:

Duration (ms)
     |
 800 |  *
 600 |     *
 400 |        *  *
 200 |              *  *  *  *  *
     |________________________________
      128  256 512  1024  1769  3008  (MB)

Cost ($)
     |
0.10 |                          *  *
0.08 |                    *  *
0.06 |              *
0.04 |        *
0.03 |     *  *
     |________________________________
      128  256 512  1024  1769  3008  (MB)

The sweet spot is where duration plateaus but cost is still reasonable — typically the "knee" of the curve.

---

Timeout Configuration

The timeout setting determines the maximum time a single invocation can run before Lambda terminates it.

Setting	Value
Minimum	1 second
Maximum	900 seconds (15 minutes)
Default	3 seconds

Setting Timeout Guidelines