Data is ingested from Producers and is stored for 24 hours (default) up to 7 and even 365 days (additional charges apply) so that it can be later used for processing by other applications.

A Kinesis Data Stream is a set of Shards, each shard contains a sequence of Data Records.

Data Records are composed of a Sequence Number, a Partition Key, and a Data Blob, and they are stored as an immutable sequence of bytes.

Kinesis Datastream is real time ~200ms.

KCL (Kinesis Client Library ) helps you consume and process data from a stream by enumerating shards and instantiating a record processor.

There are two ways to get data from a stream using KCL: Classic mechanism happens by Poll while Enhance Fan Out by Push - consumers can subscribe to a shard and data will then be pushed automatically.

Each shard is processed by exactly 1 KCL worker.

One shard provides a capacity of 1MB/sec data input and 2MB/sec data output and allows 1000 PUT records per second.

Ordering is maintained by shard

Destinations are:

Redshift

DynamoDB

EMR

S3

Firehose

Kinesis Data Stream does NOT have AutoScaling capabilities,

To scale you need to manually add shards ( and pay for them )