The article details the design, architecture, core features, scheduling logic, and performance gains of Vipshop’s self‑built Hera data service, which unifies data‑warehouse access, supports multiple engines, adapts SQL execution, and dramatically improves SLA for both B‑to‑B and B‑to‑C workloads.
This article examines seven typical scenarios where MySQL indexes fail—such as OR conditions, leading wildcards, missing quotes, functions, NULL checks, arithmetic on indexed columns, and composite index order—explaining why they happen and offering practical solutions to maintain query performance.
This article compares PyInstaller and Nuitka for converting Python projects into standalone executables, highlighting Nuitka's dramatically smaller file size, faster build times, and better performance while also providing step‑by‑step installation and usage instructions.
This article provides a systematic, step‑by‑step methodology for diagnosing Redis latency spikes, covering baseline performance testing, slow‑log analysis, high‑complexity commands, big‑key handling, expiration patterns, memory limits, fork overhead, huge‑page settings, AOF configurations, CPU binding, swap usage, memory fragmentation, network saturation, and practical monitoring tips.
This article details the end‑to‑end preparation, execution, reporting, and retrospective of a large‑scale production‑environment load test for the Double 11 shopping festival, covering data preparation, QPS target calculation, multi‑scenario testing, issue analysis, and continuous improvement practices.
This article provides an in‑depth overview of Glide 4.11, covering its basic usage, the four‑step request lifecycle, six‑step image loading flow, caching mechanisms, initialization, configuration options, and common interview‑style questions for Android developers.
This article explains how MySQL InnoDB partition tables can dramatically boost query speed, simplify routine operations, enable parallel writes, and provide efficient data exchange and cleanup compared to non‑partitioned tables, illustrated with concrete SQL examples and performance metrics.
This article explains the difference between memory leaks and overflow, whether young GC triggers stop‑the‑world pauses, distinguishes major GC from full GC, and describes Java's direct memory usage, providing clear insights into Java memory management and performance considerations.
This article examines the origins of Java's infinite loop constructs, compares the frequency of for(;;) and while(true) in the JDK source, and shows through compiled bytecode that both forms are semantically identical and have no inherent performance advantage.
This article presents fifteen practical SQL optimization techniques—including avoiding SELECT *, using UNION ALL, leveraging IN/EXISTS, batching inserts, applying LIMIT, controlling IN list size, incremental queries, efficient pagination, preferring JOINs, limiting join tables, managing indexes, choosing proper column types, improving GROUP BY, and using EXPLAIN—to help developers dramatically improve query speed and resource usage.
This article outlines ten essential Redis best‑practice tips, covering why to avoid the KEYS * command, using SCAN, interpreting INFO stats, leveraging hashes, setting key expirations, choosing eviction policies, handling errors, scaling with clusters, CPU considerations, and ensuring high availability with Sentinel.
This article provides an in‑depth overview of Apache Log4j2, comparing it with Logback and Log4j 1, highlighting its superior asynchronous performance, zero‑GC operation, flexible configuration formats, rich appender ecosystem, and practical code examples for Maven integration, XML setup, and advanced logging techniques.
A Kubernetes‑deployed business app sees its QPS drop from over 100k with a NodePort service to about 50k when exposed via Ingress, but adjusting Nginx keep‑alive parameters in the ingress‑controller can restore and even exceed the original performance while also enabling high availability.
This article presents fifteen practical SQL optimization tips—including avoiding SELECT *, preferring UNION ALL, leveraging small‑table‑driven queries, batching inserts, applying LIMIT, handling large IN lists, incremental fetching, efficient pagination, join strategies, index management, field‑type choices, and using EXPLAIN—to dramatically improve MySQL query speed and resource usage.
This article explains the concept, principles, and practical steps of iOS symbolication, covering crash log processing, address translation, ASLR handling, function‑start and nlist tables, DWARF debugging information, and essential Xcode command‑line tools for accurate source‑level debugging.
This article analyzes why Flutter apps suffer from high memory usage, identifies BuildContext and State as primary leak sources, and proposes a comprehensive, automated detection tool inspired by Android LeakCanary that accurately and efficiently finds memory leaks in production Flutter applications.
Git 2.34 introduces a sparse index for large monorepo repositories, multi‑package reachability bitmaps, the default "ort" merge strategy, OpenSSH‑based tag signing, interactive autocorrect prompts, performance improvements for fetch and push, and various submodule enhancements.
This article introduces Loguru, a third‑party Python logging library, explains why it simplifies logging compared to the standard logging module, outlines its key features such as out‑of‑the‑box setup, async safety, lazy evaluation, and file rotation, and provides practical code examples for quick integration into scripts, libraries, and frameworks.
Rust, a fast and memory‑efficient language adopted by major tech firms, is increasingly replacing JavaScript in web tooling such as bundlers, compilers, and linters, with projects like SWC, Deno, and WebAssembly demonstrating its performance, safety, and growing ecosystem for the future of JavaScript infrastructure.
This article presents practical Python performance tips, covering fundamental optimization principles, avoiding global variables and attribute lookups, reducing unnecessary abstractions, eliminating data copies, leveraging short‑circuit logic, loop improvements, JIT compilation with numba, and choosing efficient built‑in data structures, all illustrated with measurable code examples.
In a large‑scale Redis cluster, expanding nodes caused unexpected CPU spikes and higher latency for MGET operations, prompting a deep investigation that traced the issue to the CLUSTER SLOTS command and its handling of MOVED errors, followed by a code‑level optimization that reduced CPU usage by over 90% and cut command latency dramatically.
This article explains Kafka’s core architecture—including topics, partitions, replication, consumer groups, controller coordination with Zookeeper, and performance tricks like sequential writes and zero‑copy—to show how it achieves high‑throughput, fault‑tolerant messaging for large‑scale systems.
This article explains why iOS memory leaks become critical as apps grow, introduces five representative leak models, details a production‑ready object‑graph scanning solution with custom data structures and a non‑recursive DFS algorithm, and evaluates its performance impact and mitigation strategies.
This article dissects the Android cold‑start sequence of the Alipay app, detailing each stage from the initial input event through process forking, activity initialization, splash screen rendering, and final frame display, while providing timing data and optimization insights.
This article explains React 18’s new concurrent rendering features, focusing on the startTransition API, its problem‑solving benefits, practical demos comparing React 17 and 18, and a deep dive into the underlying source code and scheduling mechanisms that enable prioritized updates.
This article walks through developing a simple Go WebSocket server that echoes the current time, shows a minimal client, compares startup speed with Java, and provides full source code and test logs for practical performance testing.
This article presents a Kotlin compiler plugin that replaces verbose annotation‑based serialization with concise delegated properties, explains its design and implementation details, demonstrates significant compilation‑time reductions of up to 20% and cleaner code, and provides a step‑by‑step guide for developers interested in building similar compiler extensions.
The article details Tencent Cloud's TDSQL‑C cloud‑native database, explaining traditional database bottlenecks, its four standout features—PB‑level storage, second‑level scaling, rapid snapshot backup, and serverless pay‑as‑you‑go—and the deep kernel optimizations that enable these capabilities while outlining future performance and cross‑region goals.
This article presents fifteen practical SQL optimization tips—including avoiding SELECT *, using UNION ALL, driving large tables with small ones, batch operations, LIMIT usage, index control, proper data types, and efficient GROUP BY—to help developers improve query speed, reduce resource consumption, and avoid common performance pitfalls.
Bisheng JDK is Huawei's customized OpenJDK distribution offering high‑performance, production‑ready Java runtime optimized for ARM, compatible with Java SE 8, supporting Linux/AArch64, and featuring JFR and a fast serializer with specific JVM options.
This article examines cloud computing’s core principle of using interchangeable machines, compares parallel‑first and single‑storage‑first database architectures, outlines the advantages such as scalability, cost efficiency, and integration, and discusses the weaknesses including maturity gaps, latency, and vendor lock‑in.
The article investigates why the Feishu app sometimes fails to respond to swipe gestures after a hot start, tracing the issue to the app entering a D (uninterruptible) state during memory compression, and demonstrates how adjusting CPU priority for compression threads can reduce the problem's occurrence.
This guide compares simple Date‑based timing, System.currentTimeMillis(), and Spring's StopWatch utility, showing how to accurately measure and analyze execution durations of individual code blocks in Java applications.
This article examines how cloud computing principles shape database system design, compares parallel‑first and single‑system‑first architectures, outlines the strengths and weaknesses of cloud databases, and offers guidance for selecting the right cloud‑based data platform.
This article provides a thorough overview of Redis, covering its core data structures and common commands, persistence options, memory management and eviction policies, pipelining, transactions, Lua scripting, performance optimization techniques, replication with Sentinel, cluster sharding, and a comparison of popular Java clients.
This article walks through the complete Linux kernel path for sending a network packet, covering the send() system call, TCP processing, IP routing, queueing, driver interaction, DMA mapping, and the role of hard and soft interrupts, while answering common performance questions.
In this article we detail a two‑week sprint that identified and eliminated multiple Java backend performance bottlenecks, optimized CPU usage with jtop, re‑engineered Hystrix circuit‑breaker settings, reduced logging overhead, and fixed Spring data‑binding issues, ultimately doubling QPS and stabilizing service recovery.
PowerImage combines FFI and external‑texture techniques to deliver native‑like Flutter image handling, offering direct ui.Image loading, unified caching, simulator support, custom channels and comprehensive error capture, with texture‑based rendering preferred for typical use and FFI reserved for legacy simulators or direct pixel access, while providing performance gains and upcoming animated‑image support.
This article provides an in‑depth technical overview of Ant Financial’s Cube Card system, covering its cross‑platform dynamic UI architecture, high‑performance rendering engine, thread and data models, development workflow, native optimizations, and upcoming business and IoT expansion plans.
This article provides a detailed analysis of HikariCP’s initialization, core components, startup flow, connection acquisition logic, monitoring metrics, and key configuration parameters, illustrating how Spring Boot 2.x leverages this high‑performance JDBC connection pool and offering guidance for tuning and extending it.
This article explains why Node.js services can suffer from memory leaks, explores a V8 engine bug involving generators, walks through heap snapshot comparison and MAT analysis, lists common leak patterns, and presents practical tools and upgrade strategies to resolve the issue.
This article explains how Pika, a disk‑based Redis‑compatible KV store, overcomes Redis’s memory limits with larger capacity, faster backup, and comparable performance, and describes its deployment, management, monitoring, and migration workflows within 360’s HULK cloud platform.
StarRocks 1.19 introduces a Global Runtime Filter that speeds up joins up to six‑fold, adds BE‑level disk balancing, launches an experimental Primary Key table type with near‑duplicate‑key read performance, enables external‑table writes between clusters, and includes a host of CPU‑cache and query‑optimizations.
This article introduces the fundamentals of Android Runtime (ART) garbage collection, covering its historical background, key milestones, core features such as RegionTLAB, read barriers, pause behavior, generational support, target heap size calculations, multiplier impact, GC trigger thresholds, Systrace analysis, and parameter tuning strategies.
This article explains how MyBatis native batch inserts can exceed MySQL's maximum SQL size, causing errors, and demonstrates a solution using Guava's Lists.partition to split the data into smaller chunks, improving performance and avoiding execution failures.
This article examines frequent Redis pitfalls—including unexpected key expiration, blocking DEL commands, RANDOMKEY performance issues, SETBIT memory spikes, MONITOR OOM risks, persistence challenges, and master‑slave replication quirks—providing detailed explanations and practical mitigation strategies for each scenario.
This article explains how to use Java's FutureTask, Callable, and ExecutorService to run multiple API calls concurrently, aggregate their results, and dramatically reduce total response time compared with sequential execution, complete with code examples and performance analysis.
Zhang Chenyu from Lingque Cloud shares a comprehensive enterprise‑level Harbor case study, covering its role as an image registry, migration from 1.x to 2.x with multi‑architecture support, internal governance policies, integration with ACP, community contributions, performance workgroup findings, and the offered Harbor Enterprise edition.
This article shares a practical three‑step method for turning KPI from a hindrance into a catalyst during agile and DevOps transformation, covering urgency creation, pilot projects, and performance‑assessment redesign, while illustrating how integrated tools and feedback loops boost delivery efficiency and business value.
This article explains why UIWebView was deprecated, how WKWebView improves performance and memory usage, details its multi‑process architecture, outlines page loading flow, delegate methods, container design, caching strategies, cookie handling across processes, and solutions for white‑screen crashes in iOS apps.
This article presents a comprehensive guide to MySQL performance, covering maximum data volume and concurrency limits, recommended query response times, practical table‑design rules, index classification and optimization techniques, as well as detailed SQL tuning patterns such as batch processing, operator, OR, IN, LIKE, JOIN and LIMIT improvements.
This article explores the fasthttp library as a high‑performance alternative to Go's net/http client, demonstrates basic and advanced API usage with object pools, shows how to set up a mock test server, and provides complete Go unit‑test examples with output verification.
The article analyzes the rapid growth of the global data‑intensive HPC market, driven by AI and high‑performance data analysis workloads, outlines shifting storage demands, cloud‑HPC adoption, and the technical challenges and opportunities for future HPC storage architectures.
This article explains why LongAdder was introduced to overcome AtomicLong's CAS contention bottleneck, explores the design of Striped64 and its cells, and walks through key source code snippets of LongAdder, DoubleAdder, and the longAccumulate method, highlighting concurrency mechanisms and performance considerations.
This article details the JD Daojia coupon system's high‑traffic architecture, describing its multi‑layer design, Redis cache challenges such as large‑key and hot‑key issues, and practical optimization techniques including key redesign, expiration strategies, and active‑expire algorithms to improve performance and scalability.
This article investigates the high CPU usage and latency observed after expanding a large Redis cluster, analyzes the root cause in the CLUSTER SLOTS implementation and client MOVED handling, proposes an optimized slot‑traversal algorithm, and demonstrates significant performance improvements through benchmarking and profiling.
This article walks through setting up a lab environment with ESXi, H3C VSR, and CentOS hosts, installing iperf, netperf, and qperf, configuring IP addresses and routes, disabling firewalls, and using these tools to measure bandwidth, latency, jitter, and packet loss, revealing surprising high‑throughput results and practical troubleshooting tips.
In large Redis clusters the original CLUSTER SLOTS implementation traversed every master and all 16,384 slots, causing ~52 % CPU usage and high MGET latency during migrations, but redesigning it to iterate the pre‑built slot array reduced complexity to O(total slots), cutting execution time from 2 ms to 0.17 ms and eliminating the CPU hotspot, a fix now merged into Redis 6.2.2.
This article explores the V8 JavaScript engine’s architecture, from its historical origins and source‑code acquisition to its loading strategies, lexical analysis, parsing, compilation, and execution optimizations, illustrating how V8 achieves high performance in browsers and Node.js.
This article explains theoretical formulas and practical guidelines for calculating the optimal thread pool size in Java, covering CPU‑bound and I/O‑bound workloads, how to measure CPU cores and blocking ratios, and provides example calculations and code snippets.
When a MySQL table is sharded into multiple databases or tables, the usual LIMIT offset, pagesize pagination breaks, and this article examines why simple merge fails, then presents three practical approaches—global limit, secondary queries, and jump‑page restriction—highlighting their trade‑offs and limitations.
This article explains the data collection and reporting component of a complete frontend monitoring platform, detailing performance metrics such as FP, FCP, LCP, CLS, and providing practical JavaScript code examples for measuring, observing, and reporting these metrics, along with error and behavior monitoring techniques.
This article lists ten typical Java programming mistakes—including array‑to‑ArrayList conversion errors, inefficient value checks, improper element removal in lists, misuse of Hashtable/HashMap, raw generic types, access‑level issues, and mutable versus immutable objects—explaining why they occur and providing correct code examples to prevent them.
This article analyzes the main factors that contribute to Young Generation GC (YGC) latency in the JVM, covering safe‑point pauses, live object scanning and marking, object copying, GC log overhead, and OS activity, and offers practical tips for diagnosing and reducing pause times.
Effective logging is crucial for debugging and accountability; this guide presents fifteen practical recommendations—from selecting appropriate log levels and formatting to using SLF4J, asynchronous output, and avoiding common pitfalls—designed to help Java developers write clear, performant, and maintainable logs.
The article outlines the Douyin front‑end team's engineering philosophy, covering the need for modularization of HTML, CSS, and JavaScript, component systems, dependency handling, build tools, loaders, and performance‑focused tooling to improve development efficiency and browser runtime speed.
This article provides a comprehensive comparison between RocketMQ and Kafka, covering data reliability, performance, queue capacity, real‑time delivery, retry mechanisms, ordering guarantees, scheduled messages, transactional support, query capabilities, message tracing, consumer parallelism, filtering, and commercial backing, helping engineers choose the right messaging middleware for their workloads.
This article provides a comprehensive overview of Spring WebFlux, covering its reactive fundamentals, key components, auto‑configuration process, functional endpoints, request handling flow, performance benchmarks against Spring MVC, and recommended reactive data stores for high‑concurrency applications.
This article explains the limitations of CSV for big datasets, introduces the Feather binary format, shows how to install and use it with Python and pandas, and compares its saving/loading speed and storage size against CSV, highlighting Feather's advantages for efficient data handling.
This article surveys the most commonly used Java libraries and tools—from Netty and SpringBoot to JUnit, JMH, OkHttp, HikariCP, Caffeine, Hazelcast, logging frameworks, Jackson, Jolokia, Hibernate Validator, and FreeMarker—highlighting their roles, advantages, and integration tips for modern enterprise development.
This article presents a comprehensive guide to speeding up Python programs by applying optimization principles, avoiding global variables and attribute look‑ups, reducing unnecessary abstractions and data copies, improving loop constructs, using built‑in functions like join, leveraging short‑circuit logic, and employing tools such as numba for JIT compilation.
This article explains MySQL indexes, covering their definition, benefits, various types such as B‑tree and B+‑tree, practical techniques for creating high‑performance indexes like prefix, composite, and clustering indexes, and how to inspect and maintain index information and fragmentation.
The article explains why a simple GROUP BY query on a large user‑view table can become inefficient due to temporary tables, filesort, and memory allocation, demonstrates the execution steps with EXPLAIN, and shows how adding a composite index on user_id, viewed_user_age, and viewed_user_sex eliminates temporary tables and improves performance.
The article examines a MySQL optimizer bug where a query with a WHERE clause on uid, ORDER BY id ASC, and LIMIT 1 incorrectly triggers a full table scan despite a usable composite index, explains the optimizer’s cost calculations, and offers workarounds such as FORCE INDEX or a harmless arithmetic trick to force index usage.
The article analyzes common obstacles that prevent effective use of design patterns and code standards in everyday software development—such as performance overhead, class explosion, team skill gaps, project constraints, agile timelines, and staff turnover—and offers practical strategies to balance quality with productivity.
This article explains the concept of Time To Interactive (TTI) for the Kuaishou Android app, details how to quantify the metric using FrameTime instead of FPS, describes data collection methods, offline and online tracing, a performance lab, and comprehensive governance and task‑scheduling strategies to reduce startup lag.
The article explains how double‑brace syntax creates anonymous inner classes for initializing Java collections, discusses the hidden overhead and memory‑leak risks it introduces, and presents cleaner alternatives such as Arrays.asList, Stream.of, Guava immutable factories, and Java 9's of‑methods.
This article introduces Fair 2.0, a Flutter dynamic framework used in the Anjuke Photo App to enable runtime widget updates, explains how to embed dynamic pages, register local widgets and templates, and presents performance data such as package size, memory usage, startup time, and frame rate.
A Monday morning surge in user traffic exposed a Redis performance crisis, where CPU spiked to 100% due to massive keys* commands, and the investigation using Grafana, Redis info, commandstats, and slowlog revealed the root cause and a temporary mitigation strategy.
This article explains how to improve database insertion performance in Java by replacing per‑iteration inserts with MyBatis‑Plus’s saveBatch method, detailing dependency setup, schema creation, entity, controller, service, and mapper code for efficient bulk operations.
This article presents comprehensive Redis usage standards—including key naming conventions, avoiding big keys, efficient serialization, safe command usage, data persistence strategies, and operational recommendations—to ensure high performance, memory efficiency, and reliable service under heavy traffic.
This article presents a comprehensive collection of Java code optimization techniques—including using final modifiers, reusing objects, preferring local variables, proper resource handling, avoiding unnecessary object creation, and efficient looping—to reduce code size, improve execution speed, and enhance overall application performance.
This article explains DataVisor's risk‑control system architecture, outlines the business challenges such as complex logic, massive data, high QPS and low latency, and demonstrates how OpenResty (Nginx + Lua) is used for high‑performance gateways, micro‑service style calls, hot‑loaded scripts, automatic circuit‑breakers, and progressive architectural evolution toward service‑mesh designs.
Chrome 94, released on September 21, 2021, introduces 13 new features—including the groundbreaking WebGPU API, WebCodecs, prioritized scheduler.postTask, Idle Detection API, JS Self‑Profiling API, Canvas color management, and early hints—each detailed with usage examples, performance insights, and their impact on modern web development.
This article explains why MySQL LIMIT with large offsets becomes slow, analyzes the execution flow, and presents four optimization techniques—including subquery, INNER JOIN, tag‑record, and BETWEEN…AND…—with real‑world performance numbers and sample SQL code.
This article explains how CSS3 transforms and related properties promote elements to composite layers, detailing the browser rendering pipeline, the role of the GPU, practical demos, performance observations, and techniques for inspecting and optimizing animation performance in modern front‑end development.
This guide explains the most common JVM startup flags for controlling garbage collection logging, heap sizing, class loading monitoring, and out‑of‑memory handling, providing sample outputs and detailed interpretations to help developers tune memory usage and diagnose performance issues in large Java applications.
This article explains what memory leaks are, how to identify them in Vue SPA projects using Chrome's memory tools, lists common leak patterns, and provides practical code‑level solutions—including proper event unbinding, timer cleanup, and correct use of keep‑alive—to keep your application performant.
Learn the essential Redis monitoring metrics—including performance, memory, activity, persistence, and error indicators—along with the commands and tools such as redis-cli, INFO, SLOWLOG, and redis-benchmark to collect, interpret, and act on these metrics for effective database operations.
This roundup highlights the latest frontend advancements, including DevTools' cross‑browser support, Gatsby v4's new features, the high‑performance Kraken rendering engine, regex generator grex, the toolb.dev suite, a web‑to‑desktop framework comparison, plus in‑depth articles on Snowpack and a tenfold Table component speed boost.
This article provides a comprehensive overview of microservice architecture, tracing its history, explaining its benefits over monolithic and SOA approaches, and detailing common cognitive, time, cost, development, testing, and operational traps along with practical mitigation strategies for successful adoption.
This article presents nine actionable Java performance‑optimization techniques—from avoiding premature tweaks and profiling bottlenecks to using StringBuilder, primitive types, and caching—complete with code examples and guidance on building a reliable test suite.
This article compares three popular open‑source configuration centers—Spring Cloud Config, Apollo, and Nacos—across product features, user experience, deployment, multi‑language support, migration, performance, and high‑availability to help engineers choose the most suitable solution for microservice environments.
The DeWu ANR Monitoring Platform replaces Bugly with a comprehensive internal system that collects ProcessErrorStateInfo, tombstone files, main‑thread stack samples, Looper message histories, and app state, parses and aggregates this data, visualizes trends and root‑cause details, and guides developers in mitigating ANRs.
This guide outlines ten Vue.js performance optimization techniques—including using stable keys in v‑for, freezing immutable data, defining component data as a function, lazy‑loading components, separating non‑reactive data, employing event delegation, leveraging functional components, and using provide/inject—to boost rendering speed, cut memory usage, and simplify maintenance.
This article explains why using @Async in Spring Boot can cause memory overflow when many requests trigger parallel tasks, examines the default thread‑pool settings that lead to unbounded queues, and provides step‑by‑step configuration of a safe thread pool to prevent out‑of‑memory failures.
This session explains why Kubernetes monitoring is essential for end-to-end observability, describes the five data sources and layers it covers, and walks through discovering and locating architecture, performance, resource, scheduling, and network issues using topology, anomaly detection, and correlation techniques.
This article explores rendering performance optimization from a hardware viewpoint through to front‑end rendering, detailing the three stages of perception, healing, and corrosion, the critical rendering path, CRP analysis, and practical strategies such as NEON acceleration, OOPD/OOPR, and Lighthouse tooling to achieve smoother, lower‑power web experiences.
WebAssembly is a portable, low‑size binary format that complements JavaScript by letting C, C++, Rust, and other languages compile to fast‑loading modules, offering near‑native speed, easy macOS setup with Emscripten, measurable performance gains such as a 1.9× speed‑up on recursive Fibonacci, and seamless interaction with JavaScript for compute‑heavy web applications.
The article explains how MySQL 8.0’s new doublewrite buffer implementation dramatically slows LOAD performance on a 1 GB file, identifies the innodb_doublewrite_pages default setting as the root cause, and shows that increasing it to 64 restores speeds comparable to MySQL 5.7.
