Watch this talk here: https://www.confluent.io/online-talks/apache-kafka-architecture-and-fundamentals-explained-on-demand This session explains Apache Kafka’s internal design and architecture. Companies like LinkedIn are now sending more than 1 trillion messages per day to Apache Kafka. Learn about the underlying design in Kafka that leads to such high throughput. This talk provides a comprehensive overview of Kafka architecture and internal functions, including: -Topics, partitions and segments -The commit log and streams -Brokers and broker replication -Producer basics -Consumers, consumer groups and offsets This session is part 2 of 4 in our Fundamentals for Apache Kafka series.

1. 1
Fundamentals for Apache Kafka®
Apache Kafka Architecture & Fundamentals Explained
Joe Desmond, Sr. Technical Trainer, Confluent

2. 2
Session Schedule
● Session 1: Benefits of Stream Processing and Apache Kafka Use Cases
● Session 2: Apache Kafka Architecture & Fundamentals Explained
● Session 3: How Apache Kafka Works
● Session 4: Integrating Apache Kafka into your Environment

3. 3
Learning Objectives
After this module you will be able to:
● Identify the key elements in a Kafka cluster
● Name the essential responsibilities of each key
element
● Explain what a Topic is and describe its relation to
Partitions and Segments

4. 4
The World Produces Data

5. 5
Producers

6. 6
Kafka Brokers

7. 7
Consumers

8. 8
Architecture

9. 9
Decoupling Producers and Consumers
● Producers and Consumers are decoupled
● Slow Consumers do not affect Producers
● Add Consumers without affecting Producers
● Failure of Consumer does not affect System

10. 10
How Kafka
Uses
ZooKeeper

11. 11
ZooKeeper Basics
● Open Source Apache Project
● Distributed Key Value Store
● Maintains configuration information
● Stores ACLs and Secrets
● Enables highly reliable distributed coordination
● Provides distributed synchronization
● Three or five servers form an ensemble

12. 12
Topics
● Topics: Streams of “related” Messages in Kafka
○ Is a Logical Representation
○ Categorizes Messages into Groups
● Developers define Topics
● Producer Topic: N to N Relation
● Unlimited Number of Topics

13. 13
Topics, Partitions, and Segments

14. 14
Topics, Partitions, and Segments

15. 15
The Log

16. 16
Log Structured Data Flow

17. 17
The Stream

18. 18
Data Elements

19. 19
Brokers Manage Partitions
● Messages of Topic spread across Partitions
● Partitions spread across Brokers
● Each Broker handles many Partitions
● Each Partition stored on Broker’s disk
● Partition: 1..n log files
● Each message in Log identified by Offset
● Configurable Retention Policy

20. 20
Broker Basics
● Producer sends Messages to
Brokers
● Brokers receive and store
Messages
● A Kafka Cluster can have many
Brokers
● Each Broker manages multiple
Partitions

21. 21
Broker Replication

22. 22
Producer Basics
● Producers write Data as Messages
● Can be written in any language
○ Native: Java, C/C++, Python, Go,, .NET, JMS
○ More Languages by Community
○ REST Server for any unsupported Language
● Command Line Producer Tool

23. 23
Load Balancing and Semantic Partitioning
● Producers use a Partitioning Strategy to assign each message to a Partition
● Two Purposes:
○ Load Balancing
○ Semantic Partitioning
● Partitioning Strategy specified by Producer
○ Default Strategy: hash(key) % number_of_partitions
○ No Key Round-Robin
● Custom Partitioner possible

24. 24
Consumer Basics
● Consumers pull messages from 1..n topics
● New inflowing messages are automatically retrieved
● Consumer offset
○ Keeps track of the last message read
○ Is stored in special topic
● CLI tools exist to read from cluster

25. 25
Consumer Offset

26. 26
Distributed Consumption

27. 27
Scalable Data Pipeline

28. 28
Q&A
Questions:
● Why do we need an odd number of ZooKeeper nodes?
● How many Kafka brokers can a cluster maximally have?
● How many Kafka brokers do you minimally need for high
availability?
● What is the criteria that two or more consumers form a
consumer group?

29. 29
Continue your Apache Kafka Education!
● Confluent Operations for Apache Kafka
● Confluent Developer Skills for Building Apache Kafka
● Confluent Stream Processing using Apache Kafka Streams
and KSQL
● Confluent Advanced Skills for Optimizing Apache Kafka
For more details, seehttp://confluent.io/training

30. 3030
Certifications
Confluent Certified Developer
for Apache Kafka
(aligns to Confluent Developer Skills
for Building Apache Kafka course)
Confluent Certified
Administrator for Apache
Kafka
(aligns to Confluent Operations Skills
for Apache Kafka)
What you Need to Know
○ Qualifications: 6-to-9 months hands-on
experience
○ Duration: 90 mins
○ Availability: Live, online 24/7
○ Cost: $150
○ Register online:
www.confluent.io/certification

31. 3131
cnfl.io/slack
Stay in touch!
cnfl.io/kafka-trainingcnfl.io/download

32. 32
Thank you for attending!
• Thank you for attending thesession!
• Feedback to: training-admin@confluent.io

33. 33
Copyright ©Confluent, Inc. 2014-2019. Privacy Policy | Terms &Conditions.
Apache, Apache Kafka, Kafka and the Kafka logo are trademarks of
the Apache Software Foundation