Section-Based – Data Ingestion and Transformation – Q6

[Angie Triana]

I don’t understand how EFO can help here when the problem is about ingesting the data and the explanation didn’t make it clear for me, can anyone help me?

“A large e-commerce company is leveraging an Amazon Kinesis stream to ingest real-time data from its website, such as user clicks and page views. A major sale event has been launched recently, which significantly increased the volume of data flowing into the Kinesis stream. The stream is unable to handle the increased data volume, causing data loss and increased latency.

Which of the following actions should be taken to handle the increased data volume and ensure that the Kinesis stream can ingest the data in real-time?”

[Alas]

Hello, Angie!

Thanks for posting.

It’s a great question because it can seem like the scenario is only describing an ingestion problem, but it is actually describing two separate bottlenecks.

Kinesis Data Streams ingestion limits are determined by the number of shards. When traffic increases, the appropriate action is to perform a shard split using SplitShard, which increases the total ingestion capacity because each shard provides a fixed amount of write and read throughput.

However, the scenario also mentions increased latency, which reflects a consumer-side bottleneck. This is where Enhanced Fan-Out (EFO) becomes important. EFO gives each consumer its own dedicated 2 MB/second per shard using HTTP/2, preventing consumers from competing for read throughput and reducing overall delivery latency.

So, while resharding resolves the ingest-side constraints, EFO resolves the consumer-side delays that occur when traffic surges. Both features target different parts of the data flow, which is why both are applicable in this situation.

You can review the official AWS documentation at: https://aws.amazon.com/blogs/aws/kds-enhanced-fanout/

If you have any questions, feel free to message us. We’re happy to help!

Regards,

Ace @ Tutorials Dojo