Apache Kafka is a real-time streaming data processing platform. Discover everything there is to know to master Kafka. Streaming data processing offers many advantages. This approach makes it possible in particular to set up a Data Engineering architecture more efficiently. However, additional technologies are needed. One such technology is Apache Kafka.

What is Apache Kafka?

Apache Kafka is an open source data streaming platform. It was originally developed internally by LinkedIn as an email queue. However, this largely evolved tool and its use cases have multiplied. This platform is written in Scala and Java. However, it is compatible with a wide variety of programming languages.

Unlike traditional mail queues like RabbitMQ, Kafka retains messages after they have been consumed for a certain period of time. Messages are not deleted immediately upon confirmation of receipt. Also, mail queues are usually designed to expand vertically by adding power to a machine. Kafka, on the other hand, expands horizontally by adding additional nodes to the server cluster.

You should know that Kafka is distributed. This means that its abilities are elastic. Simply add nodes, i.e. servers, to a cluster to expand it. Another peculiarity of Kafka is its low latency. This means that it can support the processing of a lot of real-time data.

The Main Apache Kafka Concepts

To understand how Apache Kafka works, it is necessary to understand several concepts. First of all, an “event” is an atomic piece of data. An event is for example created as soon as a user registers on a system. An event can also be perceived as a message containing data. This message can be processed and saved somewhere if needed. To use the example of registering on a system, the event will be a message containing information such as username, email address or password. Thus, Kafka is a platform for working with event streams.

Events are written by “producers”: another key concept. There are different types of producers: web servers, application components, IoT devices… all write events and pass them to Kafka. For example, a connected thermometer will produce “events” every hour containing information on temperature, humidity or wind speed.

Conversely, the “consumer” or “consumer” is an entity using data events. It receives the data written by the producer and uses it. By way of example, we can cite databases, Data Lakes or analytical applications. An entity can be both a producer and a consumer, like applications or application components.

The producers publish the events on “Kafka topics”. Consumers can subscribe to get access to the data they need. Topics are sequences of events, and each can serve data to multiple consumers. Thus, producers are sometimes called “publishers” and consumers “subscibers”. In reality, Kafka acts as an intermediary between data-generating applications and data-consuming applications. A Kafka cluster has multiple servers called “nodes”.

“Brokers” are software components running on a node. The data is distributed between several brokers of a Kafka cluster, and that is why it is a distributed solution. There are multiple copies of data on a single cluster, and these copies are called “replicas”. This mechanism makes Kafka more stable, error tolerant and reliable. The information is not lost in the event of a problem with a broker. Another will take over. Finally, partitions are used to replicate data between brokers. Each Kafka topic is divided into multiple partitions, and each partition can be placed on a separate node.

What are the use cases of Apache Kafka?

There are many use cases for Apache Kafka. It is used for real-time data processing. Indeed, to function, many modern systems require that data be processed as soon as it is available. For example, in the field of finance, it is essential to immediately block fraudulent transactions. Similarly, for predictive maintenance, data flows from equipment must be continuously monitored to give the alert as soon as problems are detected.

Connected objects also require real-time data processing. In this context, Kafka is very useful since it allows streaming transfer and processing. Originally, Kafka was created by Linkedin for application activity tracking. This is therefore its original use case. Each event occurring in the application can be published to the corresponding Kafka topic.

User clicks, registrations, “likes”, time spent on a page… so many events that can be sent to Kafka topics. “Consumer” applications can subscribe to these topics and process the data for different purposes: monitoring, analysis, reports, news feeds, personalization…

Additionally, Apache Kafka is used for logging and monitoring systems. It is possible to publish logs on Kafka topics, and these logs can be stored on a Kafka cluster for a certain period of time. They can then be aggregated and processed.

It is possible to build pipelines, composed of several producers and consumers where the logs are transformed in a certain way. Then, the logs can be saved on traditional solutions. If a system has a component dedicated to monitoring, this component can read data from Kafka topics. This is what makes this tool useful for real-time monitoring.

What are the advantages of Kafka?

Using Apache Kafka brings several major benefits to businesses. This tool is designed to meet three specific needs: to provide a publish/subscribe messaging model for data distribution and consumption, to enable long-term data storage, and to enable real-time data access and processing. .

It is in these three areas that Kafka excels. Although less versatile than other messaging systems, this solution focuses on distribution and on a publish/subscribe model compatible with streaming processing.

In addition, Apache Kafka shines with its capabilities in data persistence, fault tolerance and repeatability. Data is replicated across the cluster, and elasticity allows data sharing across shards for increased workloads and data volumes. Topics and partitions also simplify data access. Designed as a communication layer for real-time log processing, Apache Kafka is a natural fit for real-time stream processing applications. This tool is therefore ideally suited to applications using a communication infrastructure capable of distributing high volumes of data in real time.

By combining messaging and streaming functionality, Kafka delivers a unique ability to publish, subscribe, store, and process recordings in real time. Storing data persistently on a cluster provides fault tolerance. In addition, this platform allows data to be moved quickly and efficiently in the form of records, messages or streams. It’s the key to interconnectivity, and it’s what allows data to be inspected, transformed and acted upon in real time.

Finally, the Connector API makes it possible to integrate many third-party solutions, other messaging systems or legacy applications through connectors or open source tools. Depending on the needs of the application, different connectors are available.

What are Kafka’s limits?

However, Kafka is not suitable for all situations. This tool is not suitable for processing a small volume of daily messages. It is designed for large volumes. Up to a few thousand messages per day, traditional message queues like RabbitMQ will be more adequate.

Moreover, Kafka does not allow easy transformation of data on the fly. It is necessary to build a pipeline of complex interactions between producers and consumers and to maintain the system in its entirety. This requires a lot of time and effort. It is therefore best to avoid this solution for ETL tasks, especially when real-time processing is required.

Finally, it is not relevant to use Kafka instead of a database. This platform is not suitable for long-term storage. Data can be retained for a specified period, but this period should not be too long. In addition, Kafka keeps copies of the data and storage costs are increased. It is better to opt for a database optimized for data storage, compatible with various query languages and allowing the insertion and retrieval of data.


We, at London Data Consulting (LDC), provide all sorts of Data Solutions. This includes Data Science (AI/ML/NLP), Data Engineer, Data Architecture, Data Analysis, CRM & Leads Generation, Business Intelligence and Cloud solutions (AWS/GCP/Azure).

For more information about our range of services, please visit: https://london-data-consulting.com/services

Interested in working for London Data Consulting, please visit our careers page on https://london-data-consulting.com/careers

Write a Reply or Comment

Your email address will not be published. Required fields are marked *