While considering AWS Lambda as a service we should be aware of its ethics. To know more about it let us consider some of the following value points –
1. When you have to access your resources from lambda (such ec2, rds running under VPC), you have to
configure AWS lambda to run under your VPCs. For each container launched by AWS lambda it has 1 private IP attached to it. In case your VPC does not have ENIs/IPs available, container would not launch, and this error is never reported. (note – this is the case for async calls).
If you see many of your lambda requests are not processing/starting, issue could be insufficient ENIs/IPs.
2. Whatever you log in lambda function using context logger or system stream is logged into cloud watch.
3. In case of async call, lambda retries (3 times) for a failed function, such as time out error. After retries it writes that transaction with details in DLQ (Dead Letter Queue), So if you are using lambda asynchronously, make sure to provide DLQ in function configuration and watch out for any data written in that queue.
4. You only have option to choose memory and proportionate CPU is allocated. Higher the memory you provide higher the cost will be.