Apache Flink is an open source framework and distributed processing engine for stateful computations over unbounded and bounded data streams. Sound like a mouthful? Read on for a comprehensive overviews of this powerful software solution, and a look at how companies use Flink to expand the way they process data.
What is Apache Flink?
Flink is an open source framework and distributed, fault tolerant, stream processing engine built by the Apache Flink Community, a subset of the Apache Software Foundation. Flink, which is now at version 1.11.0, is operated by a team of roughly 25 committers and is maintained by more than 340 contributors around the world.
The name Flink derives from the German word flink which means fast or agile (hence the logo, which is a red squirrel — a common sight in Berlin, where Apache Flink was partially created). Flink sprung from Stratosphere, a research project conducted by several European universities between 2010 and 2014.
Flink is part of a new class of systems that enable rapid data streaming, along with Apache Spark, Apache Storm, Apache Flume, and Apache Kafka. The open source tool is helping countless businesses transition away from batch processing in use cases where it makes sense to do so. Flink is now widely used in many leading applications, which we will explain further in this post.
With Flink — which is written in Java and Scala — companies can receive event-at-a-time processing and dataflow programming, using data parallelism and pipelining.
Up next, let’s take a deep dive and explore what you can do with this powerful open source program.
What Can Apache Flink Do?
Here are some of the ways that organizations use Apache Flink today.
1. Facilitate simultaneous streaming and batch processing
As creators Fabian Hueske and Aljoscha Krettek explain in a DZone post, Flink is built around the idea of “streaming first, with batch as a special case of streaming.” This, in turn, reduces the complexity of data infrastructure.
“As the original creators of Flink, we have always believed that it is possible to have a runtime that is state-of-the-art for stream processing and batch processing use cases simultaneously; a runtime that is streaming-first, but can exploit just the right amount of special properties of bounded streams to be as fast for batch use cases as dedicated batch processors,” Hueske and Krettek write.
This is arguably the best feature of Flink. Its network stack can support low-latency and high-throughput streaming data transfers along with high-throughput batch shuffles — all from a single platform.
This can drastically simplify operations, helping organizations save time and money along the way.
2. Process millions of records per minute
Since Flink uses an event-at-a-time processing schematic, it can process millions of events per minute/second.
Here’s how it works: Flink consumes an event from the source, processes it, and sends it to a sink. Then it goes on to process the next event immediately; it doesn’t wait while aggregating a batch of events.
With this functionality, Flink can process tons of events with ultra-low latency. As a result, you can to increase the throughput of your applications while having the ability to scale your systems to multiple machines.
3. Power applications at scale
One of the top reasons why developers use Flink is because it can run stateful streaming applications that can support just about any workload that you feed it. Applications are parallelized into thousands of tasks, distributed and concurrently executed in a cluster, allowing applications to use virtually any amount of memory, CPU, disk, and network IO.
One user, WalmartLabs Software Engineer Khartik Khare, says he has given Flink jobs with more than 10 million RPM, with no more than 20 cores.
Flink can also scale effectively by minimizing garbage collection and data limiting transfers across network nodes. In addition, Flink uses buffering and credit-based flow control for handling backpressure.
Add it all up, and Flink helps ensure powerful applications deliver modern user experiences at scale.
4. Utilize in-memory performance
Flink produces ultra-low processing latencies by utilizing local and in-memory states for all computations. This way it can process events in real time instead of aggregating it in batches. The software also enables exactly-once state consistency, checkpointing local states to durable storage.
Now that you have the initial lowdown on Flink, stay tuned for more content and news coming up on this topic!
Originally published at https://aiven.io.