Most data is produced continuously. For example, mobile apps, logs, IoT sensors, and so on generate data at a furious pace. Recent data is considered highly valuable, if you act on it in time. These insights can diminish in value or perish with the passage of time. A different set of tools for collecting and analyzing real-time data is required for implementing such applications. The fast pace and variable rates (bursts) of incoming data need to be stored durably and processed correctly, in a continuous, fast, and reliable manner. Typical use cases for such processing include time series analytics, feeding real-time metrics, and generating real-time alarms and notifications.

Amazon Kinesis Stream makes it easy to work with real-time streaming data. You can reliably ingest and durably store streaming data at a low cost. Additionally, you can build custom real-time applications to process streaming data. It also provides the ability to scale using a configurable number of shards. Shards give you a certain amount of throughput, for example, a single shard gives you 1,000 writes per second or 1 MB of total ingestion. Increasing the number of shards to 10 shards will give you approximately 10,000 writes per second.