DVA-C02 Troubleshooting and Optimization Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Increase the deployment package size.

  Why it's wrong here

  Cold start time scales with the size of the code package because the runtime must download, unpack, and load the code before executing. Larger deployment packages increase initialization time, not decrease it, so this option actually worsens cold starts. To reduce cold starts, keep the package lean by minimizing dependencies and using layers only when necessary.

• ✓

  Increase the memory allocated to the function.

  Why this is correct

  Lambda allocates CPU proportionally to the amount of memory configured, so more memory means more CPU power available during initialization. This speeds up tasks like loading the runtime, unpacking code, and running static initializers, thereby shortening the cold start duration. It is a practical tuning knob, though it increases cost per invocation.

• ✓

  Use Provisioned Concurrency.

  Why this is correct

  This initializes execution environments ahead of time and keeps them warm, so invocations skip the cold start initialization phase. It allocates pre-warmed instances that handle requests immediately, effectively eliminating latency from environment setup and code loading. This is a direct, guaranteed way to reduce cold starts, especially for latency-sensitive workloads.

• ✗

  Place the function in a VPC.

  Why it's wrong here

  This does not reduce cold starts; it typically increases them because Lambda must attach an elastic network interface (ENI) to your VPC, adding several seconds to initialization. The ENI creation happens during the cold start phase, incurring extra overhead. Using a VPC is for secure access to private resources, not for performance, and you can mitigate the latency by using VPC endpoints or a NAT gateway, but those don't eliminate the ENI setup delay.

• ✗

  Reduce the function timeout.

  Why it's wrong here

  The timeout setting caps how long an invocation can run; it has no impact on the initialization phase before your handler executes. Cold start measures the time to create the environment, load code, and run initializers, which happens before the timeout applies. Reducing the timeout might cause faster failures but does not accelerate environment provisioning.

Go deeper

Related to this question