Processing

Question 1

What is Lambda’s limitation?

Answer 1

Execution:
The memory can be allocated from 128 MB to 10 GB at the increment of 1 MB. The maximum execution time for a lambda function is 15 minutes. Environment variables can be of four KB. This capacity in the function container means that means in the temp folder is from 512MB to 10GB.

Deployment:
Lambda function deployment size cannot be more than 50 MB, for a stop size of uncompressed deployments (code and dependency) would be up to 250 MB. The size of an environment variable is 4KB.

Question 2

What could go wrong while processing ata in Lambda?

Answer 2

Key points to consider include:

1. Large batch sizes may cause Lambda service timeouts, so it’s essential to ensure that processing time doesn’t exceed the Lambda function’s configured limit (15 mins) .
2. There is a payload limit of 6 MB for Lambda, so batches must be split if they exceed this limit.
3. Lambda will automatically retry processing a batch of streaming data until it succeeds or the data expires, which can cause issues if errors are not handled properly and may stall the entire shard. Increasing the number of shards can help address issues with stalling pipelines due to errors and retries.
4. Lambda processes shard data synchronously, so it’s crucial to carefully consider batch size and the amount of data sent through in a single batch to ensure efficient processing.

Question 3

What is the cost model for Lambda?

Answer 3

1. Lambda cost model is affordable.
2. Pricing based on number of requests and memory consumption.
3. Generous free tier: 1 million requests/month and 400,000 GB-seconds of compute time.
4. Beyond free tier: 20 cents/million requests, $0.00001667/GB-second of throughput.
5. No fixed baseline cost; pay only for what you use.

Question 4

What are Lambda features?

Answer 4

1. Lambda offers high availability through replication and redundancy.
2. No scheduled downtime; automatic retries up to three times if code fails.
3. Unlimited scalability, but with a safety throttle of 1000 concurrent executions per region.
4. Option to request AWS to lift the throttle if needed.
5. High-performance guarantees with quick processing times.
6. Code is cached automatically for faster processing.
7. Users must specify a timeout, with a maximum of 900 seconds (15 minutes).
8. Processing should be designed to be completed within the maximum timeout to avoid system failures.