This article details why 360 chose StarRocks as its OLAP engine, presents performance and operational comparisons with MySQL, Hive, Spark, Druid, Doris and ClickHouse, describes three major production use cases, and outlines ongoing explorations such as cloud‑native integration and Kubernetes support.
This article explains the background, concept, advantages, applicable scenarios, technical implementation, and performance evaluation of Elastic Parallel Query (ePQ) in PolarDB MySQL, showing how multi‑node parallelism leverages idle CPU resources, adaptive scheduling, and cloud‑native architecture to accelerate large‑scale analytical workloads while maintaining HTAP capabilities.
ByteHouse, built on ClickHouse, addresses key limitations such as missing upsert/delete, weak multi‑table joins, scalability issues, and lack of resource isolation by introducing a modular, stage‑based execution engine, advanced join strategies, runtime filters, and a custom optimizer, delivering dramatically faster query performance.
The whitepaper introduces a comprehensive DPU performance benchmark methodology, detailing evaluation criteria across network, storage, compute, and security dimensions, and provides guidance on building test systems, selecting workloads, and designing metrics to enable fair, open, and objective assessment of diverse DPU products.
Songguo Mobility’s data‑center team migrated from a fragmented Impala‑Kudu‑ClickHouse stack to a unified StarRocks‑based real‑time OLAP architecture, iterating through three versions to solve scalability, latency, and maintenance challenges while supporting minute‑level dashboards for orders and vehicle analytics.
Phoronix benchmarked OpenJDK versions 8 through 19 alongside OpenJ9 and GraalVM CE on an Intel Core i5‑12600K Ubuntu 22.04 system, revealing varied performance across SciMark, matrix multiplication, LU decomposition, DaCapo, Renaissance, and other tests, with OpenJDK 18/19 often leading in overall scores.
An in‑depth technical overview describes how a WebAssembly‑plus‑QuickJS sandbox architecture enhances web security, improves developer experience, and delivers benchmarked performance gains for next‑generation open web applications, while maintaining compatibility with W3C standards and integrating with existing e‑commerce platforms.
This article explains how Apache Iceberg’s DataSkipping technique can lose efficiency when many filter columns are used, and presents a data‑organization optimization using space‑filling curves and Z‑Order to improve query I/O, details the OPTIMIZE implementation, and shares performance benchmark results and future plans.
This article presents a technical case study of ZhongAn's Jizhi analytics platform, detailing how switching from ClickHouse to the MPP OLAP engine StarRocks resolved multi‑concurrency and join performance bottlenecks, improved real‑time query speed, and enabled near‑billion‑row data handling for insurance business operations.
The article presents a comprehensive performance benchmark of RedisJSON (with RediSearch) against MongoDB and ElasticSearch, showing RedisJSON’s superior write, read, and mixed‑workload throughput and latency across isolated and mixed scenarios, supported by detailed YCSB test results and analysis.
Benchmark tests reveal that RedisJSON, powered by RediSearch, dramatically outperforms MongoDB and ElasticSearch across isolated reads and writes, mixed workloads, and latency metrics, delivering up to 500‑fold speed gains, lower latency, and higher throughput, making it a compelling choice for modern data‑intensive applications.
This article explains the internal mechanics of Unix Domain Sockets, shows how they differ from traditional 127.0.0.1 TCP loopback, provides sample code for server and client, and presents benchmark results demonstrating significantly lower latency and higher throughput for local IPC.
Recent benchmark tests show that RedisJSON, powered by RediSearch, dramatically outperforms MongoDB and ElasticSearch in isolated writes, isolated reads, and mixed workloads, delivering up to 12.7× faster reads, 5.4× faster writes, and significantly lower latency across various operation types.
This article examines why MySQL benchmark results often show low transactions per second on seemingly powerful servers, exploring hidden factors such as network bottlenecks, SSL overhead, and inappropriate sort_buffer_size settings, and provides practical troubleshooting steps and configuration tweaks to achieve optimal performance.
This article describes AutoHome's real‑time data analysis architecture, the challenges of existing OLAP solutions, the reasons for choosing StarRocks, detailed performance comparisons with Kylin, ClickHouse, Doris, Presto and Spark, and the practical integration of StarRocks with Flink, broker‑load scripts, and monitoring tools.
This article reviews deep learning and AI frameworks, highlights challenges of online model serving, and presents Avior—a lightweight, distributed inference engine designed for high‑performance AI services, detailing its architecture, layer design, benchmark results, and future development plans.
Shopee ClickHouse implements a cold‑hot storage architecture by mounting JuiceFS‑backed S3 as a remote volume, using table‑level storage policies and TTL‑driven background moves to shift data from SSD to cold storage, achieving cost‑effective scalability, improved disk utilization, and validated performance despite operational challenges such as Redis memory pressure and small‑file handling.
Java 17, the first LTS release in three years, brings 14 new JEPs, a free Oracle JDK license, and performance gains demonstrated by OptaPlanner benchmarks that show up to 8.66% faster than Java 11, prompting developers to consider upgrading from older versions.
The article explains Java 17’s release as the first LTS since Java 11, details Oracle’s new free JDK license, and presents a thorough OptaPlanner benchmark that compares JDK 11, 16, and 17 using G1GC and ParallelGC, revealing performance gains and guiding upgrade decisions.
This article explains Java 17’s new long‑term‑support release, Oracle’s free JDK licensing, and presents OptaPlanner benchmark results comparing JDK 11, 16 and 17 with G1GC and ParallelGC, highlighting performance gains and migration considerations for developers in production.
This article presents a detailed performance benchmark comparing MariaDB 10.4.12 and MySQL 8.0.19 on an i7 Windows 10 system, covering single and batch inserts, indexed and non-indexed queries, and analyzes speed and memory usage differences between the two databases.
This article compares Elasticsearch and ClickHouse for log analytics, detailing their architectures, presenting a Docker‑Compose test setup that streams synthetic syslog data to both stacks, executing equivalent queries, measuring response times, and concluding that ClickHouse generally delivers faster performance, especially for aggregations.
This article compares Elasticsearch and ClickHouse as log analytics solutions, detailing their architectures, deployment configurations, query capabilities, and performance benchmarks across various query types, and demonstrates that ClickHouse generally outperforms Elasticsearch in speed and aggregation efficiency.
This article details 58 Group's comprehensive evaluation of DorisDB, TiFlash, and ClickHouse for large‑scale analytical workloads, covering functional and performance benchmarks, real‑world use cases such as security analysis and DBA operations, data ingestion methods, cluster architecture, automation practices, and lessons learned.
This article presents a detailed performance benchmark of MariaDB and MySQL, covering single‑row and batch inserts, as well as various query scenarios with and without indexes, and concludes that MariaDB generally outperforms MySQL while using more memory.
This article explains how OceanBase, a distributed relational database, achieved a new TPC‑H world record, discusses the evolution of HTAP technology, the architectural challenges of combining transactional and analytical workloads, and the future direction of the product.
This article explains how NVIDIA's RAPIDS Accelerator leverages GPUs to speed up Apache Spark 3.0 workloads, detailing the underlying architecture, benchmark results on TPC‑DS and recommendation models, required configuration changes, supported operators, shuffle optimizations, and the enhancements introduced in versions 0.2 and 0.3.
Through a series of benchmark tests comparing simple, logging‑enabled microservices written in Java (Helidon) and Go (Go‑Kit), the authors evaluate whether Java can achieve Go‑like performance, exploring JVM tuning, GraalVM native images, and Kubernetes deployments to uncover speed, memory, and scalability trade‑offs.
Zhao Jianbo, head of Kuaishou's big data architecture team, presents an in‑depth overview of Apache Flink's adoption at Kuaishou, covering reasons for selection, development history, business data flows, technical innovations such as the Slimbase state engine, stability improvements, and future roadmap.
The article analyzes recent Docker adoption trends, compares the benefits and overhead of containers and virtual machines, presents benchmark results showing significant performance gains for container workloads on bare‑metal servers and Kubernetes clusters, and discusses implications for hybrid cloud strategies.
This article benchmarks C and Rust by implementing five classic sorting algorithms—Bubble, Insertion, Selection, Shell, and Heap Sort—across three data sizes, analyzes the execution times on a macOS system, and discusses the nuanced performance differences and broader implications for language choice.
This article details Suning's journey of replacing commercial databases with PostgreSQL and Citus, covering background, feature and performance comparisons, reliability tests, deployment architecture, real‑world use cases, operational tips, and a comprehensive Q&A on migration and maintenance.
This article examines Alibaba's FastJson library, benchmarking its speed against Jackson and Gson, analyzing its limited adoption, exposing unresolved issues, and ultimately recommending a shift to more widely supported JSON parsers for Java projects.
BeetlSQL provides a unified, high‑performance database access framework supporting traditional, big‑data, and time‑series databases, outlines its Maven dependency, and presents detailed M3‑phase performance benchmarks across various query methods and ORM tools.
This article evaluates five popular WeChat mini‑program frameworks—WePY, Mpvue, Taro, Uni‑app and Chameleon—by comparing their features, integration costs, performance metrics, and maintainability, and provides recommendations on which framework best fits different development stacks.
OceanBase, Alibaba’s self‑developed distributed relational database, achieved world‑leading TPC‑C performance in 2019 and shattered its own record in 2020, a journey marked by intense team effort, architectural innovations like Paxos‑based scalability, and a vision to become a global database leader.
This article evaluates Taro’s conversion to JD mini‑programs by comparing package size, long‑list rendering benchmarks, and development experience against native JD and WeChat mini‑programs, revealing where Taro excels, where it lags, and the optimizations it provides.
This article examines Alibaba's FastJson library, benchmarks its serialization speed against Jackson and Gson, discusses its popularity and issue count on Maven, and concludes with recommendations for JSON parsing choices in Java projects.
This article presents a detailed MySQL performance benchmark of Alibaba Cloud RDS, Tencent Cloud RDS, and AWS RDS using sysbench, analyzes configuration differences, shows results across mixed, read‑only, and no‑index update workloads, and offers practical recommendations for cloud database tuning.
This article benchmarks several Linux methods for deleting half a million small files—including rm, find with -exec, find with -delete, rsync, Python, and Perl—showing their execution times and concluding that rsync with the --delete option is the quickest and most convenient solution.
This study evaluates Kubernetes container networking options—Docker host networking, Flannel, IPvlan, VXLAN, host‑gw, and AWS‑VPC—by measuring latency, throughput, and installation cost under high‑load conditions, revealing IPvlan as the lowest‑latency, highest‑throughput solution while host‑gw offers a balanced, low‑dependency alternative.
OceanBase’s distributed relational database achieved the top TPC‑C benchmark score by leveraging cloud‑native resources, optimized storage, advanced transaction engines, and extensive SQL tuning, with detailed insights into preparation, system architecture, performance testing phases, ACID compliance, and storage optimizations that set new industry standards.
This article benchmarks the throughput, average latency, median latency, and maximum latency of five RPC frameworks—Dubbo, Motan, rpcx, gRPC, and Thrift—using a unified protobuf‑based service across varying client concurrency levels, and discusses the resulting performance trends and conclusions.
Using JMH, this article benchmarks the serialization and deserialization performance of four popular Java JSON libraries—Gson, Fastjson, Jackson, and Json-lib—by testing various scenarios with a complex Person model, analyzing results, and providing guidance on selecting the most suitable library for high‑performance applications.
This article presents a detailed performance benchmark comparing Apache Storm and Apache Flink in stream processing, focusing on data transmission methods, reliability mechanisms, operator chaining, and both self‑generated and Kafka‑sourced workloads, and provides practical optimization recommendations based on the results.
The article introduces the Mica Java library, explains its core dependencies and recent updates, and presents benchmark comparisons showing Mica’s superior performance in non‑compile‑time bean copying, UUID generation, and date formatting against tools like Hutool, Spring, and CGLIB.
This article benchmarks popular Java bean‑copy tools—including MapStruct, Selma, mica, CGLIB, Spring BeanUtils, and Hutool—under various scenarios such as simple models, type‑conversion, list mapping, and map‑to‑bean conversion, presenting performance scores, error margins, and functional comparisons to help developers choose the most suitable library.
A detailed benchmark comparing Spring Cloud Gateway, Zuul, and Linkerd shows that, contrary to popular belief, Spring Cloud Gateway delivers 1.6 times the throughput of Zuul and even outperforms Linkerd, after correcting earlier testing errors caused by a Reactor Netty bug.
This article introduces the concept, characteristics, and use cases of time series databases, explains the data model and challenges of traditional solutions, and provides a detailed overview of Tencent's Cloud Time Series Database (CTSDB) along with performance comparisons against InfluxDB.
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.
This article evaluates MongoDB and MySQL for storing and querying large‑scale precious‑metal market data, detailing hardware setup, version configurations, schema designs, and extensive performance benchmarks across million‑ to billion‑record datasets, and concludes with analysis of memory architectures, data models, and suitable application scenarios.
This article explains the core functions of relational‑database middleware, dives into Sharding-JDBC’s architecture, performance characteristics, and implementation details such as sharding rules, SQL parsing, routing, rewrite, execution, result merging, and distributed primary‑key generation, and outlines its future roadmap.
A practical benchmark on three identical servers shows MySQL Group Replication achieving higher TPS than Semi‑Sync despite larger network traffic, and explains how Paxos‑based parallel log transmission gives it an edge over the single‑threaded Semi‑Sync approach.
Pilosa, originally built for sparse high‑cardinality user attributes, is evaluated on a dense, low‑cardinality NYC taxi dataset to see if it can serve as a general‑purpose index, with performance comparisons against Spark, PostgreSQL, Elasticsearch, and kdb+ across multiple query scenarios.
This article explains Kafka's distributed architecture, topic‑partition model, replication and ISR mechanisms, data durability settings, delivery guarantees, deduplication strategies, and presents benchmark results that illustrate how configuration choices affect throughput and latency in real‑world deployments.
This article explains Kafka's distributed architecture, its topic‑partition storage model, replication and ISR mechanisms, leader election, delivery guarantees, configuration for high reliability, and presents extensive benchmark results showing how replication factor, acks settings, and partition count affect throughput and latency.
This article introduces Mongosync, a Qihoo 360‑developed tool for transferring data between MongoDB replica sets, detailing its background, supported features such as full‑data and real‑time oplog synchronization, version compatibility, multithreaded bulk writes, configurable options, performance test results, and future development plans.
The article presents an interview with SPEC chairman Walter Bays at the 2016 SPEC Asia Summit in Beijing, discussing SPEC’s role, its benchmark standards, the organization’s focus on China, and the introduction of cloud‑IaaS and big‑data performance benchmarks for future computing needs.
A detailed benchmark compares Redis 3.0.7 and Hazelcast 3.6.1 on a four‑node HP ProLiant cluster, showing Redis excels with few clients while Hazelcast scales better under high concurrency, highlighting trade‑offs for developers and architects.
The author compares Rust and Go using a series of micro‑benchmarks across common algorithms, presenting CPU time, elapsed time, memory usage and code size, and concludes that while Rust is roughly 30 % faster, Go produces about 30 % less code, making it a better choice for enterprise applications.
This article examines Redis clustering techniques, compares client‑side sharding, proxy sharding, and Redis Cluster, critiques Twemproxy's limitations, and presents Codis's architecture, performance benchmarks, and practical tips for seamless migration and high‑availability in modern operations.
This article explains Apache Kafka's distributed publish‑subscribe design, core components, storage model, ZooKeeper coordination, performance benchmarks against ActiveMQ and RabbitMQ, and provides Java producer and consumer code examples for building high‑throughput messaging applications.
The article introduces SessionDB, a Java‑implemented, LSM‑tree‑based key/value storage engine designed to eliminate sticky sessions by providing high‑throughput, durable, and scalable session data handling, and details its architecture, optimizations, sharding strategy, and benchmark comparisons with BerkeleyDB, LevelDB, and RocksDB.
