Suppose your service needs to process millions of meta‑video reports, but the application consumes excessive memory and eventually crashes. What type of messaging channel would you use to handle this data flow?

For high‑volume meta‑video processing that spikes memory usage, decouple producers and consumers with a messaging system that supports back‑pressure and horizontal scaling. A Kafka topic is ideal: producers write records to the topic, and multiple consumer instances read in parallel, each maintaining its own offset. Kafka’s log compaction and retention policies keep the data durable without consuming excessive memory. Alternatively, AWS SQS or Pub/Sub can be used for simpler workloads. The key is to stream the data rather than load it all into memory, allowing the consumer to process records incrementally and commit progress, preventing memory exhaustion.