Book Image

Scalable Data Streaming with Amazon Kinesis

By : Tarik Makota, Brian Maguire, Danny Gagne, Rajeev Chakrabarti
Book Image

Scalable Data Streaming with Amazon Kinesis

By: Tarik Makota, Brian Maguire, Danny Gagne, Rajeev Chakrabarti

Overview of this book

Amazon Kinesis is a collection of secure, serverless, durable, and highly available purpose-built data streaming services. This data streaming service provides APIs and client SDKs that enable you to produce and consume data at scale. Scalable Data Streaming with Amazon Kinesis begins with a quick overview of the core concepts of data streams, along with the essentials of the AWS Kinesis landscape. You'll then explore the requirements of the use case shown through the book to help you get started and cover the key pain points encountered in the data stream life cycle. As you advance, you'll get to grips with the architectural components of Kinesis, understand how they are configured to build data pipelines, and delve into the applications that connect to them for consumption and processing. You'll also build a Kinesis data pipeline from scratch and learn how to implement and apply practical solutions. Moving on, you'll learn how to configure Kinesis on a cloud platform. Finally, you’ll learn how other AWS services can be integrated into Kinesis. These services include Redshift, Dynamo Database, AWS S3, Elastic Search, and third-party applications such as Splunk. By the end of this AWS book, you’ll be able to build and deploy your own Kinesis data pipelines with Kinesis Data Streams (KDS), Kinesis Data Firehose (KFH), Kinesis Video Streams (KVS), and Kinesis Data Analytics (KDA).
Table of Contents (13 chapters)
1
Section 1: Introduction to Data Streaming and Amazon Kinesis
5
Section 2: Deep Dive into Kinesis
10
Section 3: Integrations

Monitoring KDA applications

CloudWatch is a service used to monitor applications in AWS. In addition to CloudWatch, we can also use the Flink dashboard when we use KDA for Flink CloudWatch groups metrics in what's known as a namespace, and for KDA, that namespace is AWS/KinesisAnalytics. When it comes to KDA Flink applications, we can choose to have KDA emit metrics at the Application, Task, or Operator level. The two most interesting metrics are KPU, the amount of KPUs our application is consuming, and MillisBehindLatest, which tells us how backed up we are (the difference between the timestamp of the record we are processing and the record in the stream under the LATEST position).

For in-depth guidance on how to use CloudWatch with KDA for Flink, please refer to the Enhanced monitoring and automatic scaling for Apache Flink blog in the Further reading section:

Figure 6.19 – CloudWatch metrics dashboard

Please note that we can also emit our...