In this chapter, we will understand a few test cases in Apache Flink.
Bouygues Telecom is one of the largest telecom organization in France. It has 11+ million mobile subscribers and 2.5+ million fixed customers. Bouygues heard about Apache Flink for the first time in a Hadoop Group Meeting held at Paris. Since then they have been using Flink for multiple use-cases. They have been processing billions of messages in a day in real-time through Apache Flink.
This is what Bouygues has to say about Apache Flink: "We ended up with Flink because the system supports true streaming - both at the API and at the runtime level, giving us the programmability and low latency that we were looking for. In addition, we were able to get our system up and running with Flink in a fraction of the time compared to other solutions, which resulted in more available developer resources for expanding the business logic in the system."
At Bouygues, customer experience is the highest priority. They analyse data in real-time so that they can give below insights to their engineers −
Real-Time Customer Experience over their network
What is happening globally on the network
Network evaluations and operations
They created a system called LUX (Logged User Experience) which processed massive log data from network equipment with internal data reference to give quality of experience indicators which will log their customer experience and build an alarming functionality to detect any failure in consumption of data within 60 seconds.
To achieve this, they needed a framework which can take massive data in real-time, is easy to set up and provides rich set of APIs for processing the streamed data. Apache Flink was a perfect fit for Bouygues Telecom.
Alibaba is the largest ecommerce retail company in the world with 394 billion $ revenue in 2015. Alibaba search is the entry point to all the customers, which shows all the search and recommends accordingly.
Alibaba uses Apache Flink in its search engine to show results in real-time with highest accuracy and relevancy for each user.
Alibaba was looking for a framework, which was −
Very Agile in maintaining one codebase for their entire search infrastructure process.
Provides low latency for the availability changes in the products on the website.
Consistent and cost effective.
Apache Flink qualified for all the above requirements. They need a framework, which has a single processing engine and can process both batch and stream data with same engine and that is what Apache Flink does.
They also use Blink, a forked version for Flink to meet some unique requirements for their search. They are also using Apache Flink's Table API with few improvements for their search.
This is what Alibaba had to say about apache Flink: "Looking back, it was no doubt a huge year for Blink and Flink at Alibaba. No one thought that we would make this much progress in a year, and we are very grateful to all the people who helped us in the community. Flink is proven to work at the very large scale. We are more committed than ever to continue our work with the community to move Flink forward!"