An experiment built on the Rust Tokio async runtime. Minimises SQS costs by long polling for the maximum of 10 messages from an SQS queue. Messages are distributed to a concurrent worker pool. Can be further improved by batching delete message requests.
Produce some messages for consumption:
RUST_LOG=INFO cargo run --bin producer
Consume the messages:
RUST_LOG=INFO cargo run --bin consumer
Configure consumer concurrency and producer message volume in Settings.toml
Use the terraform files in /infra to provision a queue.