FlinkML is the Machine Learning (ML) library for Flink it has a growing list of algorithms and contributors that aim to provide scalable ML algorithms, an intuitive API, and tools that help minimize glue code in end-to-end ML systems.
Customer Reviews
Marvin P.
Advanced user of FlinkMLI have implemented flinkml for a unified platform to process batch data, the software works brilliantly, is extremely fast and efficient, this software have a wide field of application and is usable for dozens of big data scenarios. Although Flink can run standalone, it usually runs on top of an HDFS installation to read/write distributed files. In addition, Flink can run with YARN support and let YARN deal with the cluster resources, something very useful
The only negative thing I've experienced is that flink are optimized by cost-based optimizer (SQL engines). So Flink applications will be required re-configuration and maintenance whenever the cluster characteristics change and the data evolves over time,but only that, in everything else flink fulfills its function
In my opinion Flink is a great choice, because I do not have to face so much "out-of-memory" problems during the development. Flink has it's own Memory Manager, so in general you do not need to care about it.
Among the benefits are -high performance, flink's data streaming runtime provides very high throughput
-low latency flink can process the data in sub-second range without any delay
-lightning fast speed, it processes data at lightning fast speed (hence also called as 4G of big data)
-fault tolerance, failure of hardware, node, software or a process doesn't affect the cluster.
