This article analyzes the current state, development stages, and comprehensive solutions of e‑commerce data‑warehouse governance, covering data quality, cost, security, and efficiency requirements, and presents a roadmap from early‑stage standardization to mature tool‑driven governance with future outlooks.
This article explains the GeoHash algorithm, demonstrates how binary subdivision of latitude and longitude yields compact base‑32 strings, and shows how these hashes can efficiently locate nearby ride‑hailing drivers while highlighting precision limitations and edge cases.
This article discusses the implementation of Apache Kylin as an OLAP engine for logistics data, focusing on optimizing cube building and query performance to handle large-scale, high-dimensional data analytics.
This article summarizes Wang Feng's presentation on the next‑generation Lakehouse architecture, explaining how Apache Paimon provides a unified, real‑time data lake format that bridges batch and streaming workloads, enabling low‑latency analytics and AI integration for modern big‑data applications.
This article presents JD Logistics' one‑stop agile BI platform, detailing the complex data sources, rapid business demands, the UData solution architecture, performance and usability improvements, and future upgrade plans that together enable faster data integration, self‑service reporting, and enhanced decision‑making across the organization.
This guide introduces Apache Spark's Python API, showing how to install PySpark, launch an interactive shell, create a SparkSession, read and write data from various sources, perform transformations, and apply key performance‑tuning practices for efficient big‑data processing.
This article explores the technical maturity curve of indicator systems, covering data collection, modeling, production, management, governance, and application, while analyzing the security, stability, and usability requirements and discussing how large language models can enhance certain clear and complicated scenarios.
This article covers recent AI developments, big data challenges, and industry insights including AI course expansions, regulatory discussions, and tech company updates.
This article details how the WeChat team leverages ClickHouse at massive scale, introduces a suite of performance observation tools, describes lakehouse reading and bitmap optimizations, and explains the integration of AI workloads, demonstrating overall query speedups of up to tenfold across diverse scenarios.
This article presents a detailed case study of how EasyPoint built a scalable, stable advertising platform using ClickHouse Enterprise, covering company background, data architecture with Kafka and Druid, ClickHouse advantages, serverless resource scaling, and extensive performance benchmarks.
This article explains how to construct a comprehensive data indicator system by outlining its background, design, standardization, metadata management, and future applications, while addressing business, technical, and product challenges and showcasing practical examples and visual workflows.
This article provides a detailed technical comparison between StarRocks and Trino, covering their shared MPP architecture, cost‑based optimizer, pipeline execution, ANSI SQL support, differences in vectorized execution, materialized view capabilities, caching systems, data source connectors, benchmark results, high‑availability designs, join algorithms, and real‑world user case studies.
StarRocks combines extreme query speed and a unified architecture to deliver a lakehouse solution that separates storage and compute, supports multi‑warehouse resource isolation, offers Trino compatibility, materialized‑view acceleration, and cost‑effective scaling, making it suitable for real‑time analytics, data‑lake queries, and traditional OLAP workloads.
This article explains Kafka's role as a high‑throughput distributed message queue, detailing its core components, topic‑partition model, consumer groups, storage mechanisms, fault‑tolerance features, delivery guarantees, ZooKeeper coordination, and scalability strategies for building reliable real‑time data pipelines.
Databricks announced the acquisition of Tabular, the company founded by the original creators of Apache Iceberg, aiming to integrate Delta Lake and Iceberg into a unified, open lakehouse architecture that enhances format compatibility, reduces data silos, and supports AI workloads.
The article explains how China Unicom transformed its 5G fully‑connected factory data pipeline from a complex Lambda architecture into a streamlined, real‑time and offline‑integrated solution built on Apache Doris, detailing system requirements, architectural redesign, performance gains, and future plans.
This article presents Ant Group's comprehensive data governance experience, covering data quality management, storage governance, architectural design, operational strategies, case studies, and forward‑looking thoughts on integrated lake‑warehouse governance, data value realization, and AI‑driven automation.
This article details JD Retail’s end‑to‑end data platform, covering data asset certification, 5W2H modeling, unified query DSL, intelligent acceleration, robust governance, visualization components, low‑code orchestration, and large‑model AI applications that together reduce query latency, cut development costs, and empower analysts across the retail business.
This article shares Kuaishou's practical experience with data lake technology (Hudi), detailing the challenges of growing data warehouses, the migration from Hive to Hudi, the promotion strategy, real-world use cases such as CDC sync and batch‑stream integration, and key takeaways for future deployments.
This article provides a comprehensive overview of Apache Kafka, covering its role as a message queue, core components, topic and partition design, consumer groups, storage mechanisms, high‑availability features, delivery guarantees, ZooKeeper coordination, and scalability strategies for building robust real‑time data pipelines.
This article examines why data departments get overwhelmed by massive data‑retrieval requests, identifies root causes such as mindset, requirement handling, and lack of tools, and presents a technical solution centered on dimensional modeling and OLAP multi‑dimensional reporting to streamline data access and empower teams.
This article details how the Dongchedi business team designs, implements, and monitors a comprehensive metric system within its data warehouse, covering metric standards, model construction, metadata management, quality monitoring, application scenarios, and future directions using the DataLeap platform.
This article details JD Retail’s large‑scale HDFS deployment, describing how cross‑region storage challenges were solved with a full‑copy topology, asynchronous block replication, flow‑control mechanisms, and a tiered storage strategy that automatically moves hot, warm, and cold data among SSD, HDD, and high‑density HDD nodes to improve performance and cut costs.
The Athena Data Factory, built by Spark Thinking, is a comprehensive one‑stop data development and governance platform that integrates data integration, development, analysis, and services, offering offline, real‑time, and AI pipelines, modular architecture, extensive monitoring, and cost‑optimisation to empower thousands of users across the company.
This article details how Ctrip, a leading travel company, leverages Alluxio as a distributed cache within its extensive big‑data infrastructure to improve data access speed, implement transparent storage access, support custom authentication and multi‑tenant features, enhance audit logging with CallerContext, and dynamically distribute client configurations via Kyuubi.
The Feitian Technology Salon held on May 16 in Shanghai showcased Arm Neoverse's core advantages and demonstrated how Yitian 710‑based ECS instances deliver significant cost‑effective performance gains for big‑data and video workloads through cloud‑native optimizations and software acceleration techniques.
This article introduces Elasticsearch, an open‑source distributed search and analytics engine, explaining its architecture, core concepts such as clusters, nodes, shards, replicas, indices, inverted indexes, documents and fields, and how these components enable fast, scalable searching and data analysis.
This article explains common data‑quality challenges when rebuilding business systems, compares manual SQL‑based validation with a dedicated data‑comparison product, and walks through practical steps for configuring, executing, and reviewing automated data‑matching tasks in a big‑data environment.
The article introduces Archer, a new big‑data warehouse engine built on Iceberg that adds an inverted‑index mechanism using Tantivy to provide full‑text and JSON search, storage‑compute separation, and significant performance gains over traditional Elasticsearch and Iceberg connectors.
This article presents a comprehensive overview of the Berserker big‑data platform, detailing its overall design, data‑development components, key architectural challenges such as state management, release processes, two‑phase commit, RPC duplication, task routing, message handling, execution isolation, dependency model redesign, and outlines future work including stateless execution nodes, Kubernetes integration, and unified stream‑batch processing.
This article explains the motivations for adopting message queues, outlines core concepts and protocols, compares mainstream MQ products, and provides a detailed walkthrough of Kafka architecture, cluster setup, native Java APIs, and Spring Boot integration with extensive code examples.
This article introduces Tencent's Big Data Processing Suite (TBDS), discusses challenges of data silos, and presents Gravitino's open‑source unified metadata service and permission model, detailing how it integrates Hadoop, MPP, and various catalog plugins to provide consistent access control across heterogeneous data platforms.
This article describes how 58.com designed and implemented a large‑scale user profiling platform using ClickHouse, covering system overview, core modules, major challenges of scale, complexity and performance, and the detailed storage, query, and optimization techniques applied to meet business needs.
This article compares Pandas, Polars, and PySpark across five dataset sizes, showing how Polars' eager and lazy modes dramatically outperform the other tools, and discusses when each framework is the most suitable choice for data processing workloads.
This article presents a complete guide to Hudi-based real-time data lake ingestion, covering overall data integration architecture, batch and streaming ingestion strategies, advanced table design, and practical recommendations for handling challenges such as deduplication, latency, partitioning, and performance optimization.
This article explains how a data security governance platform protects data across its entire lifecycle—from warehouse construction and collection to application—by implementing fine‑grained permission controls, encryption, masking, authentication, and comprehensive auditing, while addressing scalability, high availability, and regulatory compliance challenges.
This article details Xiaomi's sales data warehouse development, covering its history, architecture, dimensional modeling, layer design, streaming‑batch integration, governance, security, and future directions, while also addressing practical Q&A on implementation challenges and best practices.
Didi’s Elasticsearch platform, built on ES 7.6 and deployed on physical machines with containerized gateway and control layers, provides a multi‑tenant, high‑performance search service—featuring a user console, operational controls, ZGC‑based latency reductions, cost‑saving compression, custom security, real‑time cross‑datacenter replication, and a roadmap toward ES 8.13.
This article describes Zhihu's hybrid cloud architecture—including offline, online, and GPU data centers—its self‑built UnionStore cache, the performance and latency challenges faced during large‑scale AI model training, and the subsequent evaluation and migration to Alluxio community and enterprise editions to achieve higher throughput, stability, and lower operational overhead.
The article examines the evolution of data integration architectures—from traditional ETL through ELT to the emerging EtLT model—highlighting their advantages, disadvantages, industry trends, maturity stages, and practical guidance for enterprises and professionals navigating modern big‑data pipelines.
The article presents a suite of interview‑style algorithm and system‑design solutions—including Bloom‑filter URL blacklists, hash‑partitioned word frequencies, missing‑number bit arrays, top‑K min‑heap, low‑memory median, short‑URL encoding, Redis user counting, and extensive Java implementations of sorting, singleton, LRU cache, custom thread pools, producer‑consumer models and various FooBar synchronization techniques.
The article introduces Apache Paimon 0.8, highlighting new Deletion Vectors, a universal file index, memory and I/O optimizations, record‑level TTL, and integration improvements with Flink and Spark, while also discussing broader lake‑house performance trends and future directions.
This article presents the evolution of data platform architectures, the specific challenges of financial‑sector information‑technology innovation, and the design, core components, deployment path, and real‑world case studies of the cloud‑native lakehouse solution DataCyber developed by Shuxin Network.
This article provides a detailed overview of Apache Kafka, covering its core characteristics, distributed architecture, key components such as topics, partitions, brokers, producers, consumers, ZooKeeper, and common application scenarios like log collection, event‑driven architecture, real‑time analytics, and monitoring.
This article explores the complete lifecycle of an indicator system—from defining metrics and addressing common ambiguities, through designing concept consensus, semantic layers, mechanisms, and governance, to productizing platforms, optimizing development, and envisioning future AI‑driven enhancements.
This article outlines the evolution from 1G to 6G communications, explains the third AI wave driven by big data, theory, and compute, introduces federated learning (horizontal, vertical, transfer), and details on‑device AI architectures, decision tree and neural network models, and real‑world use cases such as video preloading and autonomous driving.
This article explains how Lazada’s internal selection platform leveraged Hologres and the RoaringBitmap compression algorithm to dramatically reduce storage costs, accelerate set operations, and break the 200,000‑item pool limit, achieving up to a 99% speed improvement in scheduling.
In the data‑driven era, KNIME offers a free, visual, and highly scalable platform that streamlines massive data ingestion, preprocessing, analysis, automation, and visualization, enabling researchers to handle millions of records efficiently without extensive coding or costly software.
The article presents Ant Group's exploration of applying its data‑driven risk control and credit assessment capabilities to the traditional bulk commodity sector, detailing industry background, data pain points, core technical solutions, and the construction of a secure, explainable data‑model platform for digital transformation.
This article presents OPPO’s next‑generation big‑data and AI integrated architecture on functional cloud, detailing a cloud‑native elastic compute framework, a unified data‑lake solution, real‑time feature platforms, machine‑learning data acceleration, and hybrid‑cloud deployments, highlighting performance gains and cost reductions.
This article explains how Alluxio enables a unified lake‑warehouse architecture by decoupling compute and storage, outlines its core capabilities, evaluates the cost‑saving and performance benefits, discusses the technical challenges, and presents several practical deployment scenarios in finance and AI workloads.
This article presents an in‑depth overview of JD.com’s retail data visualization platform, detailing its product matrix—including EasyBI, a low‑code platform, and JDV large‑screen tool—its architectural layers, key capabilities, business case studies, challenges faced, and future development directions.
This article presents the problems faced by NetEase Cloud Music's data warehouse attribution system and details a comprehensive solution that includes upgrading the event‑tracking framework, redesigning the attribution model, and launching a unified management platform to improve stability, accuracy, and scalability.
The article reviews Apache Paimon's graduation to an Apache Top-Level Project, outlines the essential capabilities of modern lakehouse frameworks—including streaming and batch I/O, multi‑engine integration, and advanced features—and discusses the problems they solve and the promising direction of the lakehouse ecosystem.
This article explores practical SQL optimization techniques for Alibaba's ODPS platform, covering fundamentals, common pitfalls like null handling and select *, advanced strategies such as multi‑insert, partition limiting, UDF placement, data‑skew mitigation, parameter tuning, and real‑world case studies that dramatically reduce query runtimes.
This article shares Ant Group's extensive experience in big data governance, detailing the overall data governance framework, data quality management, data storage governance, and future considerations, illustrated with practical cases and strategies for ensuring compliance, reliability, and cost efficiency.
This article introduces Delta Lake 3.1, detailing its release background, the addition of Deletion Vector to Update and Merge commands, metadata‑driven count/min/max optimizations, the Universal Format for cross‑engine compatibility, and a comparative evaluation with Iceberg and Hudi.
This article provides a comprehensive overview of cloud computing, explaining its definition, major service models (IaaS, PaaS, SaaS), key advantages and challenges, and the essential technologies such as virtualization, distributed systems, automation, security, storage, and big data that enable modern cloud solutions.
To overcome the NameNode write bottleneck caused by a single global read/write lock in Bilibili’s massive HDFS deployment, the team introduced hierarchical fine‑grained locking—splitting the lock into Namespace, BlockPool, and per‑INode levels—which yielded up to three‑fold write throughput gains, a 90 % drop in RPC queue time, and shifted performance limits from lock contention to log synchronization.
The talk explores how the rapid growth of multimodal data and large language models is reshaping data processing, highlighting three key trends—online‑offline integration, vector‑relational database convergence, and the fusion of data processing with AI computation—while presenting practical solutions and future visions for unified data‑AI ecosystems.
This article presents a comprehensive overview of the Apache‑incubated Paimon project, covering its evolution from Flink Table Store, the current features of primary‑key and log tables, management tools such as snapshots, tags and branches, performance optimizations for Flink and Spark, and a detailed roadmap of upcoming functionalities.
This article provides a comprehensive overview of Apache Hudi 1.0, detailing its architectural redesign, five major development directions, and the most important new capabilities such as LSM‑tree timeline, function indexes, file‑group readers/writers, partial updates, and non‑blocking concurrency control, along with performance evaluations and resource links.
This article explains the concepts, formulas, and step‑by‑step implementation of a user‑retention analysis model, covering both Hive‑based offline processing and ClickHouse‑accelerated real‑time queries, complete with SQL examples, architecture diagrams, and practical optimization tips.
This article introduces the TransBigData Python package, explains how to install it, read mobile signaling data with pandas, preprocess and grid the data, identify stay and move events, determine home and work locations, and visualize individual user activity using built‑in functions.
This guide walks through planning a three‑node Hadoop 3.3.5 cluster, explains default and custom configuration files, details core‑site, hdfs‑site, yarn‑site, and mapred‑site settings, shows how to distribute configs, start HDFS and YARN, and perform basic file‑system tests.
This article explains why new data systems are needed, introduces Apache Arrow’s columnar in‑memory format and its zero‑copy advantages, describes how to model data at read time, outlines the execution flow with Acero and SQL planning, and shares practical tips and extensions for building robust, dynamic‑schema data platforms.
Apache Paimon, originally launched as Flink Table Store, has graduated to an Apache Top‑Level Project after a year of incubation, showcasing real‑time lakehouse capabilities, extensive ecosystem integration, and strong adoption by major enterprises, marking a significant milestone for streaming and batch data processing.
This article describes Bilibili’s intelligent optimization project that automatically analyzes historical query workloads to configure multi‑dimensional sorting, various indexes, and pre‑aggregation on Iceberg tables, thereby reducing scan volume by 28% across dozens of tables and improving OLAP query latency.
This article explains how to build and apply a comprehensive metric system by addressing business, technical, and product challenges, outlining design, standardization, metadata management, and future AI‑driven use cases to support data‑driven decision making.
This article explores Uber’s sophisticated real‑time data infrastructure, detailing how the company leverages open‑source technologies such as Apache Kafka, Flink, Pinot, and Presto, and describing the architectural components, scaling challenges, multi‑region resilience, data back‑filling, and operational practices that enable low‑latency analytics for millions of daily rides and deliveries.
This article details Tencent Video’s data business, describing the design and implementation of its metrics middle platform and lake‑warehouse integration, covering architecture, governance, consistency, timeliness, usability, cost optimization, and future plans, with insights into technology choices such as Iceberg, StarRocks, and MQL.
This article outlines Bilibili's data‑driven approach, describing the five‑year development of its big‑data development governance platform, its user segmentation, product positioning, data‑map and governance product designs, operational methods, value evaluation, and future roadmap, highlighting significant efficiency gains and user impact.
This article provides a comprehensive overview of the evolution, current state, and future trends of financial risk control technologies, covering data, feature engineering, modeling, decision-making, product development, challenges, and the impact of large AI models on the industry.
Using Python data visualization and geospatial analysis, this article compares the nationwide distribution of Starbucks and Luckin Coffee stores in China, revealing differences in regional concentration, proximity patterns, and statistical insights such as average Luckin stores within 500 m of each Starbucks location.
This article explains how MaxCompute leverages materialized views as a query accelerator, covering their history, advantages and drawbacks, creation and maintenance details, automatic query rewriting, intelligent recommendation, auto‑materialization, and future enhancements for large‑scale data warehousing.
This article explains MaxCompute’s new integrated offline‑and‑near‑real‑time architecture, Transaction Table 2.0, detailing its unified storage and compute design, automatic data governance, schema evolution, upsert and time‑travel capabilities, and how it simplifies complex big‑data pipelines while delivering minute‑level latency and lower costs.
This article presents a visual case study of a company's data governance and data middle‑platform implementation, outlining the project background, solution architecture, and the resulting business value and effects through a series of illustrative images.
This article explains ElasticSearch’s architecture and core mechanisms—including its reliance on Lucene segments, inverted indexes, stored fields, document values, caching, shard routing, and scaling strategies—while answering common questions about wildcard matching, index compression, and memory usage.
This article explains how a third‑generation metric platform replaces traditional ETL‑heavy data‑warehouse pipelines with a semantic‑driven NoETL approach, reducing cost, improving quality and efficiency, and delivering automated, self‑service analytics for both IT and business users.
This article explains how the Dongchedi team designed, implemented, and monitored a comprehensive indicator system within a petabyte‑scale data warehouse, covering standards, metadata management, model construction, quality monitoring, and diverse application scenarios to improve data reliability and business insight.
When faced with a business requirement to filter up to 100 000 records from a pool of tens of millions and then sort and de‑duplicate them, this article explores four technical solutions—multithreaded ClickHouse pagination, Elasticsearch scroll‑scan, a combined Elasticsearch‑HBase approach, and RediSearch with RedisJSON—detailing their design, implementation, performance testing, and trade‑offs.
This article shares how China Unicom Digital Technology leverages DataOps to build an integrated data governance, research and development, and operations capability, outlining challenges, methodological considerations, a seven-step governance framework, and a multi-center collaborative mechanism to achieve sustainable data-driven value.
Bloom filters are space‑efficient probabilistic structures that answer “definitely not” or “maybe” membership queries, with a controllable false‑positive rate derived from bit array size, element count, and hash functions, and can be implemented via Guava’s Java library, Redisson’s Redis wrapper, native Redis modules, or custom bitmap code, dramatically reducing memory usage and latency in large‑scale systems such as URL deduplication or user‑product checks.
The article presents Turing Data Analysis (TDA), a self‑service BI platform that replaces fragile traditional pipelines with a unified DWD‑based data model, drag‑and‑drop analytics, multi‑engine query optimization and caching, delivering sub‑10‑second queries on billions of rows, fine‑grained permissions, and rapid dashboard creation, while reporting significant usage growth and outlining AI‑driven future enhancements.
Data middle platforms transform raw enterprise data into reusable assets by integrating collection, storage, processing, governance, and service layers, enabling faster deployment, consistent metrics, improved data quality, and business value across digital transformation, while addressing challenges like siloed data, low efficiency, and inconsistent standards.
This article shares Xiaomi's real‑world challenges and solutions when building a new Spark 3.1‑based data platform, covering Multiple Catalog implementation, Hive‑to‑Spark SQL migration, automated batch upgrades, performance and stability optimizations, and future roadmap for vectorized execution.
This article analyzes the challenges of commercial real‑time data processing—such as stability, multi‑stage computation, and frequent schema changes—and explains how the RTS platform provides end‑to‑end managed solutions, auto schema handling, primary‑secondary redundancy, experiment‑first deployment, and metadata generation to unlock high‑velocity data value for advertising operations.
This article presents Li Auto’s end‑to‑end data integration journey, detailing the evolution of its data platform, the challenges of heterogeneous sources, and how a unified Flink‑on‑K8s solution with cloud‑native architecture, operator management, monitoring, and checkpointing addresses batch‑stream convergence and future scalability.
This article explores Apache Kafka’s core storage architecture, explaining how sequential append‑only writes, sparse indexing, segmented log files, and a leader‑based replication mechanism together enable high‑throughput, reliable, and scalable event streaming for massive data workloads.
This article outlines the core capabilities of data-driven development, the background and architecture of DataOps, its research challenges and focus areas, and explores future directions such as data virtualization, platform governance, and data value assessment, providing a comprehensive overview of DataOps practices.
This article introduces a comprehensive ClickHouse training program that covers fundamental concepts, architecture, installation, distributed cluster design, data import, performance tuning, and includes a detailed list of 33 video modules and additional recommended reading resources for large‑scale data analytics.
DataFunCon 2024 Shanghai brings together leading experts from AI, big data, cloud computing, and industry sectors to discuss cutting‑edge technologies, large‑model applications, intelligent operations, and digital transformation across automotive, healthcare, finance, retail, and entertainment.
The DataFunCon 2024 Shanghai conference brings together leading experts from academia and industry to discuss cutting‑edge topics such as large language models, AI‑driven operations, data governance, digital transformation, and emerging applications across automotive, finance, retail, and entertainment sectors.
This article examines ByteDance's DataOps journey, detailing the data‑engineering challenges faced, the concrete solutions and productization through the DataLeap platform, the metrics and best‑practice framework adopted, and the future directions involving AI‑assisted development and open‑source collaboration.
OPPO's data architects combined their self‑developed Shuttle service with Alluxio to double performance, halve system pressure, and double throughput, while building a unified Data & AI lakehouse that integrates structured and unstructured data, metadata management, real‑time ingestion, and cloud cost reductions.
This guide explains how to efficiently load one billion 1‑KB log entries from HDFS or S3 into MySQL by analyzing B‑tree limits, using batch inserts, choosing the right storage engine, sharding tables, optimizing file reading, and coordinating tasks with Redis, Redisson, and Zookeeper.
DataFunCon2024 Shanghai brings together leading experts from AI, big data, cloud computing and various industries such as automotive, biotech, retail, finance and entertainment to share cutting‑edge research, practical case studies and future trends through a series of keynote speeches, panels and technical sessions.
This article analyzes the challenges of building real‑time and batch risk‑control features, compares Lambda and Kappa architectures, evaluates storage‑unified and compute‑unified solutions, and details how StarRocks was selected, validated, and deployed to achieve high‑performance, low‑latency feature serving in a financial context.
This article explains the value of user profiling platforms, outlines their core functions, presents a layered architecture with open‑source options, and details engineering optimizations—from wide‑table design to BitMap caching and task‑mode execution—while also discussing current industry trends.
This guide explains how to use Alibaba Cloud Function Compute to inspect Kafka message keys and automatically perform insert, update, or delete operations on MongoDB, detailing the architecture, advantages, prerequisites, step‑by‑step deployment, and current limitations.
