Amazon Kinesis
External
- https://aws.amazon.com/kinesis/
- Amazon Kinesis Whitepaper https://d0.awsstatic.com/whitepapers/whitepaper-streaming-data-solutions-on-aws-with-amazon-kinesis.pdf
- https://docs.aws.amazon.com/kinesis/
- Amazon Kinesis Data Streams Developer Guide https://docs.aws.amazon.com/streams/latest/dev/introduction.html
Internal
Subjects
Overview
Kinesis acts as a highly available conduit to stream messages between data producers and data consumers. The Kinesis service can be integrated and exposed externally via the Amazon API Gateway.
Concepts
Stream
A Kinesis data stream is a named set of shards. Streams can be created from the AWS Management Console, with AWS CLI and via the Kinesis Data Stream API.
Stream Name
The name space if defined by the AWS Account and AWS region: for the same account, streams with the same name can exist in different regions.
Data Stream
Associated with Kinesis Streams.
Delivery Stream
Associated with Kinesis Firehose.
Shard
A shard is a uniquely identified sequence of data records in a stream. Shards are identified in a stream by their partition key. They also automatically get a Shard ID, which can be obtained with AWS CLI describe-stream command. Shards have a fixed capacity: they can support up to 5 transaction per second for reads, and a maximum total data read rate of 2MB/sec and up to 1,000 records per seconds for writes, up to a maximum total data write rate of 1 MB per second, including partition keys. To increase the capacity of the stream you can add more shards.
Shard ID
The shard ID is different from the partition key.
Shard Iterator
A shard iterator represents the position of the stream and shard from which the consumer will read.
The choice of shard iterator is important. Different shard iterators may yield different behaviors while receiving records, including not receiving some records.
Shard Iterators:
LATEST
Start reading just after the most recent record in the shard, so that you always read the most recent data in the shard. If a consumer using the "LATEST" shard iterator and a producer as start concurrently at about the same time - which may be the case in testing - receiving the record is prone to race conditions, the record may or may not be received depending on whether the consumer was started before or after the producer.
TRIM_HORIZON
Start reading at the last untrimmed record in the shard in the system, which is the oldest data record in the shard. If all untrimmed records in the stream are read with one TRIM_HORIZON iterator, then a new TRIM_HORIZON is created and used, the same records are received again.
AT_SEQUENCE_NUMBER
Start reading from the position denoted by a specific sequence number, provided in the value StartingSequenceNumber.
AFTER_SEQUENCE_NUMBER
Start reading right after the position denoted by a specific sequence number, provided in the value StartingSequenceNumber.
AT_TIMESTAMP
Start reading from the position denoted by a specific time stamp, provided in the value Timestamp.
Record
A record is a unit of data stored in a stream. A records has a sequence number, a partition key and a data blob. After the data blob is stored in a record, Kinesis does not inspect, interpret or change it in any way.
Data Blob
The data blob is the immutable sequence of bytes constituting the payload of a record. Kinesis Data Streams does not inspect, interpret, or change the data in the blob in any way. A data blob can be up to 1 MB.
Partition Key
The partition key is used to identify different shards in a stream, and allow a data producer to distribute data across shards. The partition key that is associated with each data record is used to determine which shard a given data record belongs to. Partition keys are Unicode strings with a maximum length limit of 256 bytes. An MD5 hash function is used to map partition keys to 128-bit integer values and to map associated data records to shards. When an application puts data into a stream, it must specify a partition key - setting the partition key is mandatory when using the API.
Sequence Number
A sequence number is a unique identifier for records inserted into a shard. The sequence number is assigned by Kinesis after data is written to the stream with client.putRecords or client.putRecord. Sequence numbers increase monotonically, and are specific to individual shards. The sequence number for a record is accessible as a header ("aws_receivedSequenceNumber").
Producer
Producers put data records into streams. The producers can continually push data to Kinesis Data Streams.
Consumer
The consumers may process data in real time. A Kinesis Data Stream consumer can be a custom application running on Amazon EC2 or an Amazon Kinesis Data Firehose delivery stream.
Retention Period
The retention period is the length of time that data records are accessible after they are added to the stream. The default retention period is 24 hours, and it can be increased up to 168 hours (7 days).