This article explains Flink's distributed cache feature, describes its registration and retrieval mechanisms, and provides a complete Java example that demonstrates how to register a file, access it within a RichMapFunction, and print the processed results.
This guide explains how to set up an Apache Flink cluster on CentOS 7 using three deployment methods—Local, Standalone, and Flink on YARN/Kubernetes—including host configuration, SSH setup, package distribution, configuration file editing, cluster start/stop commands, YARN resource manager concepts, session commands, job submission, fault‑tolerance settings, and log inspection.
This article introduces Apache Flink's DataSetAPI and DataStreamAPI, explains their source, transformation, and sink concepts, highlights the key differences in transformation handling, and notes the series' goal of publishing over 500 big‑data tutorials for learners from beginner to expert.
This article introduces the design and implementation of a real‑time user behavior engineering platform at Qunar using Apache Flink, covering Flink's core characteristics, distributed runtime, DataStream programming model, fault‑tolerance, back‑pressure handling, event‑time processing, windowing, watermarks, and practical code examples for filtering, splitting, joining, and state management.
This article outlines a comprehensive series of over 500 planned tutorials covering Java advanced features, distributed theory, Hadoop, Spark, Flink, and various big‑data storage and processing technologies, designed to guide engineers transitioning into big‑data development from fundamentals to expert level.
Alibaba’s Blink, an internal enhancement of Apache Flink, is now open‑sourced, bringing advanced runtime, SQL/TableAPI, Hive compatibility, Zeppelin integration, and a revamped Flink Web UI to dramatically boost performance and scalability for both streaming and batch workloads.
This article analyzes the difficulties of achieving exactly-once delivery in Apache Flink, explains the distinction between state and end‑to‑end semantics, and details how idempotent and transactional sinks—illustrated with the Bucketing File Sink—realize exactly‑once guarantees through checkpoint‑based two‑phase commit.
This article walks through Youzan's real‑time platform architecture, explains why Flink was chosen over Spark Structured Streaming, details practical challenges such as container over‑provisioning and monitoring overhead, shares solutions for Spring integration and async caching, and outlines future directions for SQL‑based streaming and scheduler improvements.
This article outlines Didi's journey from fragmented, self‑built real‑time clusters to a unified, YARN‑managed platform that leverages Spark Streaming, Flink, and StreamSQL, detailing architectural choices, resource isolation, CEP enhancements, and the resulting impact on latency‑critical services.
This article reviews the evolution and characteristics of major big‑data processing engines—from first‑generation Hadoop MapReduce to second‑generation DAG‑based Tez, third‑generation in‑memory Spark, and fourth‑generation real‑time Flink—highlighting their batch and streaming use cases.
This article explains the concept of stateful computation in stream processing, highlights the shortcomings of traditional systems, and details how Apache Flink provides rich state access, various state backends, and robust checkpointing mechanisms to achieve scalable, fault‑tolerant real‑time analytics.
This tutorial demonstrates how to use Apache Flink's Kafka connector to read data from Kafka topics with exactly‑once semantics, covering Maven dependencies, consumer configuration, checkpointing for fault tolerance, and a complete Scala example that writes the streamed data to HDFS.
This article provides a concise overview of numerous Apache open‑source projects—including Ignite, MapReduce, Pig, JAQL, Spark, Storm, Flink, Apex, REEF, Twill, and Beam—that enable distributed in‑memory storage, real‑time and batch processing, and advanced analytics for large‑scale data workloads.
This newsletter introduces Alibaba’s X‑DeepLearning deep‑learning framework, releases a large‑scale cluster dataset for research, announces the upcoming open‑source of the Blink streaming engine, showcases the futuristic FlyZoo Hotel, and shares a welcome letter to new hires, highlighting cutting‑edge technologies across AI, big data, and cloud operations.
This article describes how a large‑scale operations monitoring system at Ctrip evolved from rule‑based alerts to a deep‑learning‑driven anomaly detection pipeline using LSTM models and Flink streaming, achieving a ten‑fold reduction in alert volume while improving fault detection.
From early self‑built Storm and Spark Streaming clusters to a unified YARN‑based Spark platform and finally a low‑latency Flink system with extended CEP and StreamSQL capabilities, Didi’s real‑time computing platform evolved through three stages, delivering multi‑tenant isolation, rich SQL processing, and dramatically reduced development costs.
This talk presents Yanxuan’s real‑time data warehouse built on Flink, covering background challenges, overall architecture and implementation, data quality measures, monitoring, and practical application scenarios, while highlighting design goals of flexibility, high development efficiency, and stringent data quality requirements.
Alibaba’s weekly roundup highlights its re‑election to the Java Global Management Committee, the upcoming open‑source X‑Deep Learning framework, a global mathematics competition, the launch of the DAMO Academy website, massive IPv6 deployment, and two major technical conferences slated for December.
To meet Double‑Eleven’s sub‑second, billion‑item feed demands, Alibaba’s Xianyu selection system evolved from a Solr‑based search pipeline through offline batch and PostgreSQL attempts to a Blink‑powered real‑time stream platform using Niagara’s low‑latency LSM storage, delivering high‑throughput, personalized product feeds.
This article details Vipshop's real‑time platform architecture, the migration from Storm and Spark to Flink, Flink's deployment on Kubernetes, and the latest Unified Data Management system that unifies data access across Kafka, Redis, Tair and HDFS.
This article explains how Meituan tackled growing real‑time data demands by redesigning its streaming platform, adopting a layered real‑time data warehouse architecture, selecting storage and compute technologies such as Cellar, Elasticsearch, Druid and Flink, and sharing practical tips on dimension expansion, joins, and aggregation to achieve higher throughput and lower latency.
This article summarizes Meituan’s real‑time computing platform, detailing its layered architecture built on Kafka, Flink on YARN, state management, resource isolation, fault tolerance, monitoring, and the Petra metric aggregation system, while highlighting the challenges faced and the solutions implemented to achieve high‑throughput, low‑latency stream processing at massive scale.
Meituan replaced its Storm‑based pipeline with a four‑layer real‑time data warehouse powered by Flink, using hybrid storage (Cellar KV, Elasticsearch, Druid, MySQL) to deliver low‑latency, high‑throughput services, dramatically simplifying SQL‑driven development, unifying metrics, cutting compute costs, and paving the way for offline‑grade accuracy and reliability.
This article explains how Alibaba's search offline platform combines massive batch and real‑time processing, leveraging Hadoop, HBase, Flink (Blink) and a unified full‑incremental model to handle tens of billions of daily records and millions of TPS for diverse business lines.
The article summarizes the 2018 Flink Forward Berlin conference, highlighting Apache Flink’s architecture, the new Leager ACID solution, Alibaba’s batch‑stream unification advances, comparisons with Spark, and future directions including AI integration and micro‑service convergence.
As mobile and IoT data surge, real-time stream computing—especially Flink’s low-latency, high-throughput, exactly-once engine—addresses challenges of latency, accuracy, and usability, and Tencent Cloud’s managed Flink service provides elastic, secure, integrated pipelines for applications ranging from online status monitoring to fraud detection and smart transportation.
This article examines Ele.me’s big‑data platform evolution, comparing Storm, Spark Streaming, Structured Streaming, and Flink, detailing their architectures, consistency semantics, performance trade‑offs, and why Flink became the preferred real‑time computation engine for the company.
This article compares Spark Streaming and Flink across runtime models, component roles, programming APIs, task scheduling, time semantics, dynamic Kafka partition detection, fault‑tolerance mechanisms, exactly‑once guarantees, and back‑pressure handling, providing code examples and practical insights for real‑time data processing.
This article explains how the data aggregation platform adopts Apache Flink for high‑throughput, low‑latency stream processing, covering the complete workflow from data source integration, transformation operations, windowing and time concepts, to a concrete order‑count example with custom aggregation logic.
Amid declining Hadoop usage reports, Suning.com’s 2018‑2020 big‑data platform case study reveals why the retailer still relies on Hadoop’s mature ecosystem, how it integrates HDFS, HBase, YARN, Hive, Spark, Flink and emerging tools, and what future resource‑management plans it envisions.
This article examines Ctrip Ticket's real-time data processing ecosystem, comparing batch and streaming frameworks such as Hadoop, Spark, Storm, Flink, and Spark Streaming, detailing Kafka deployment and configuration, and describing how these technologies are applied in production for log analysis, seat‑occupancy detection, and anti‑crawling.
iQIYI’s big‑data collection system has progressed from simple HTTP log uploads to a Flume‑Kafka pipeline and finally to a custom Venus‑Agent architecture with centralized configuration, persistent offsets, dual‑Kafka streams and Flink processing, now handling tens of millions of queries per second and over three hundred billion records daily to power its AI‑driven services.
This article explains how Alibaba built a massive real‑time computing platform using Flink and its Blink extensions, detailing the challenges of ultra‑low latency, exactly‑once guarantees, and high throughput, and showing how these technologies powered the record‑breaking Double 11 shopping festival.
This report presents a systematic performance evaluation of Apache Flink and Apache Storm across multiple real‑time processing scenarios, measuring throughput, latency, message‑delivery semantics, and state‑backend effects, and provides recommendations for selecting the most suitable engine based on the observed results.
This article presents a comprehensive performance evaluation of Apache Flink versus Apache Storm across multiple real‑time processing scenarios, measuring throughput, latency, and the impact of different configurations and delivery semantics to guide framework selection and optimization.
Alibaba’s Double 11 showcase reveals how its upgraded stream computing platform handled a 100% year‑over‑year data surge, achieving 256 K successful payments per second and processing 472 million events per second in real time through a highly optimized Flink‑based architecture.
This article introduces Flink’s background, architecture, and ecosystem, then demonstrates a step‑by‑step tutorial on using Flink SQL to clean and transform streaming data from Kafka, highlighting its advantages over Storm for real‑time ETL.
This article introduces the evolution of big‑data processing from Google’s MapReduce concept to modern open‑source frameworks, defines big data and its 3V characteristics, outlines typical processing pipelines, and compares batch, stream, and hybrid systems such as Hadoop, Storm, Samza, Spark, and Flink.
This article explains Flink’s architecture, detailing the roles of Client, JobManager and TaskManager, walks through a SocketTextStreamWordCount example, and clarifies the four‑layer graph model—StreamGraph, JobGraph, ExecutionGraph, and the physical execution graph—highlighting why each layer exists.
This article explains Flink’s cluster architecture, the roles of Client, JobManager and TaskManager, demonstrates a SocketTextStreamWordCount example, and details the four-layer execution graph (StreamGraph, JobGraph, ExecutionGraph, Physical Execution Graph) to illustrate how Flink schedules and runs streaming jobs.
Alibaba algorithm engineer Shi Xiaogang shares his journey from a Peking University PhD researching real‑time iterative computation on data streams to developing Blink’s state management and recovery features in Flink, highlighting the challenges of transitioning from academia to industry and the impact of large‑scale real‑time systems.
The article recounts the 2017 Flink Forward conference in San Francisco, highlighting key sessions from DataArtisans, Uber, Netflix and Alibaba, and discusses real‑time stream processing use cases, large‑scale deployments, runtime and TableAPI/SQL improvements, and the growing adoption of Flink in the industry.
This article explains how Alibaba’s Blink, built on Apache Flink, transforms batch‑oriented big‑data platforms into a unified, high‑performance real‑time computing engine, detailing its architecture, state management, checkpointing, and successful deployment in e‑commerce, search, recommendation, and online machine‑learning scenarios.
This article explains the concept, implementation, and underlying mechanics of session windows in Apache Flink, covering how Flink assigns and merges windows, the relevant APIs such as SessionWindows.withGap, and detailed source code analysis for both the window assigner and trigger handling.
The article explains how Flink tackles JVM memory challenges by using proactive memory management, a custom serialization framework, cache‑friendly binary operations, and off‑heap memory techniques to reduce GC pressure, avoid OOM, and improve performance in big‑data workloads.
This article explains Flink's core execution‑resource concepts—including operator chaining, task slots, slot‑sharing groups and co‑location groups—detailing their conditions, API controls, internal implementation, and how they together maximize throughput and resource utilization in stream processing.
This article surveys the evolution of Hadoop and its ecosystem, explains core storage and processing concepts, and introduces contemporary big‑data technologies such as Spark, Flink, Kafka, Lambda architecture, NoSQL databases, and cloud‑native solutions, highlighting their roles and trade‑offs.
During a lively “Sit and Discuss” session, experts compared Spark and Hadoop, evaluated Flink against Spark, contrasted HBase with Cassandra, explained why Kafka (and sometimes Flink) is preferred for distributed messaging, and shared insights on Tachyon’s role in modern big‑data ecosystems.
