-
Useful Links:
-
Exam Tips:
- Designed for huge scale ingestion.
- Multiple consumers, 24 hour rolling window.
- Data ingestion, analytics. monitoring, app clicks.
- Producers:
- Anything which puts data into Kinesis.
- Android/iPhone
- Software
- EC2 instance(s)
- Anything which puts data into Kinesis.
- Consumers:
- Consumes and uses the records from a Kinesis stream.
- EC2 instance(s) running the Kinesis Consumer Library (KCL).
- Lambda functions
- Kinesis Firehose
- Consumes and uses the records from a Kinesis stream.
- Producers -> Stream -> Consumers
- Can scale to any load that is required.
- Streams:
- Data Streams:
- 24 retention period default - up to 7 days at an extra cost.
- Each shard = 1 MB across all producers, 2 MB across all consumers.
- Uneven write pattern, when you don't reach the total performance of a stream.
- Can configure shard level metrics at extra cost.
- Private only vpc? Need to us a Kinesis endpoint to access public resources.
- Data Streams:
- Data Analytics:
- Use cases:
- Time-series analytics
- Realtime dashboards
- Realtime metrics
- Complex manipulation of data in realtime.
- Allows SQL queries against realtime data streams.
- Create dashboards or alerts.
- Destinations:
- S3
- Redshift
- ElasticSearch
- Splunk
- Firehose (makes it near realtime)
- Understand:
- Reference table.
- In application input and output.
- Different products that can be used as inputs.
- You can define an error stream.
- Use cases: