/plushcap/analysis/datastax/datastax-quatre-raisons-pour-lesquelles-apache-pulsar-est-essentiel-dans-votre-stack-data

Quatre raisons pour lesquelles Apache Pulsar est essentiel dans votre stack data moderne

What's this blog post about?

DataStax, a company known for its distributed database technologies, has announced that it will be building a messaging solution to complement its existing ecosystem of databases. The company has started by evaluating the most popular option, Apache Kafka, and found it lacking in four key areas: geo-replication, scaling, multi-tenancy, and queuing. Apache Pulsar, on the other hand, meets all these requirements to DataStax's satisfaction. Geo-replication is a crucial feature for many applications, as it allows data to be replicated across multiple regions, ensuring local latency and compliance with data sovereignty regulations. While Cassandra supports geo-replication, Kafka does not, which can lead to increased latency for users outside the region where Kafka is deployed. Pulsar, like Cassandra, integrates geo-replication into its architecture, allowing messages to be visible to consumers across regions. Scaling is another important consideration when choosing a messaging solution. In Kafka, adding capacity can slow down the cluster before speeding it up, which may not be ideal for businesses with rapidly changing needs. Pulsar addresses this issue by separating storage and computation, allowing new nodes to be added without affecting existing data or requiring additional work from the cluster. Multi-tenancy is a feature that allows multiple users or organizations to share an infrastructure while remaining isolated from each other. This can help reduce costs by amortizing the cost of shared components across all users and simplifying administration. While Kafka does not support multi-tenancy, Pulsar has it built into its architecture, allowing for easy management of multiple tenants in different regions with features like authentication, authorization, isolation policies, and storage quotas. Finally, queuing is a messaging model that allows messages to be consumed in any order, which can be useful when the order of processing is not important. Pulsar supports both pub/sub (publication/subscription) and queuing models, making it a versatile solution for various use cases. In conclusion, DataStax's decision to build a messaging solution based on Apache Pulsar highlights the advantages it offers over Kafka in terms of geo-replication, scaling, multi-tenancy, and queuing. The company is excited to join the Pulsar community with its acquisition of Kesque, a Pulsar-as-a-service provider, and plans to open-source management and monitoring tools built by the Kesque team in its new Luna Streaming distribution of Pulsar.

Company
DataStax

Date published
April 8, 2021

Author(s)
Yahya JARRAYA

Word count
1203

Language
français

Hacker News points
None found.


By Matt Makai. 2021-2024.