apache-flink mit Homebrew, Nix installieren

Projektlinks

• Homepage
• Repository
• Upstream-Dokumentation
• Paketmanager-Seite

Befehle und Aliase

• flink

Projektgeschichte

Flink grew out of the Stratosphere research project, which began in 2009 around TU Berlin and related European data-management research groups. The project entered the Apache Incubator in April 2014, released 0.6-incubating later that year, and graduated as an Apache top-level project in December 2014.

The ASF top-level announcement framed Flink as a system that deliberately combined streaming/pipelining and batch processing techniques. From early on, the project included its own serialization, memory management, and cost-based query optimization pieces, which made it more than just a wrapper around MapReduce-era execution.

Adoptionsgeschichte

Flink's early adoption story centered on companies experimenting with or adopting it for production data infrastructure. ASF's 2014 announcement quoted ResearchGate, Spotify, and Amadeus discussing production or exploratory use, while noting more than 70 contributors by the time of graduation.

Over time, Flink became a common choice for event-time processing, exactly-once state, streaming SQL, continuous ETL, and event-driven applications. Its ecosystem later externalized many connectors into separate Apache repositories, which matters for packages because the main flink CLI is only the core of a broader connector and deployment stack.

Wie es verwendet wird

The Homebrew package exposes the flink command, which developers use to run and inspect jobs, interact with local clusters, and submit work to configured deployments. Apache's current stable docs describe Flink as designed for common cluster environments, in-memory speed, and scale.

Starting with Flink 2.0, Apache's docs say the supported configuration file is config.yaml in the conf directory. That makes $FLINK_HOME/conf/config.yaml the package-level config location for the current formula.

Warum Paket-Nerds sich dafür interessieren

Flink is package-nerd material because it is both a developer CLI and a distributed runtime. A local install needs Java, shell scripts, a config directory, plugin/connectors, logs, and often a clear boundary between the package's $FLINK_HOME and the user's deployment configuration.

It is also one of the major Java data engines where versioning matters deeply: jobs, connectors, state formats, savepoints, and cluster versions are coupled enough that packaging the latest major line and a versioned 1.x line is practically useful.

Zeitleiste

• 2009: The Stratosphere research project begins.
• 2014: Flink enters the Apache Incubator.
• 2014: Apache Flink 0.6-incubating is released.
• 2014: Flink graduates as an Apache top-level project.
• 2026: Current docs describe config.yaml as the Flink 2.x configuration file.

Related projects

• Stratosphere is Flink's research-project predecessor.
• Apache Kafka, Hadoop YARN, Kubernetes, Apache Paimon, Flink CDC, and the externalized Flink connector repositories are common adjacent projects.

Quellen

• https://github.com/apache/flink
• https://incubator.apache.org/projects/flink.html
• https://news.apache.org/foundation/entry/the_apache_software_foundation_announces69
• https://nightlies.apache.org/flink/flink-docs-stable/docs/deployment/config/