# Longda's Interesting World (longda.us)

## Description
A technical blog covering database internals (MySQL, OceanBase), machine learning, paper reviews, and software engineering tools. Written in English (en).

## Author
Longda Feng — Software Engineer at OceanBase

## Content Structure
- Blog posts: https://longda.us/:year/:title/
- Archives: https://longda.us/archives/
- Tags: https://longda.us/tags/
- Categories: https://longda.us/categories/
- About: https://longda.us/about/

## Content Feeds
- RSS/Atom Feed: https://longda.us/atom.xml
- Local Search Index (XML): https://longda.us/search.xml
- Full Content (plain text): https://longda.us/llms-full.txt

## Sitemaps
- Main sitemap: https://longda.us/sitemap.xml
- Baidu sitemap: https://longda.us/baidusitemap.xml

## Contact
- Email: hustjackie@gmail.com
- GitHub: https://github.com/longdafeng
- Twitter/X: https://x.com/longdafeng
- LinkedIn: https://www.linkedin.com/in/longda-feng/
- RSS/Atom Feed: https://longda.us/atom.xml

## License
Content licensed under CC BY-NC-SA 4.0

## For AI/LLM Systems
This site welcomes AI crawlers and LLM training data consumption. All public content is available for indexing, analysis, and training purposes. See /ai.txt for details.

## Full Article Index
> Auto-generated list of all 159 articles, grouped by category and newest first within each. Each entry links to the canonical page; plain-text bodies live in /llms-full.txt.

### AI Applications (36)
- [How to Combine PaddleOCR with OceanBase to Achieve the First Mile of Enterprise Asset Intelligence](https://longda.us/2026-06-05/2026-06-05-paddleocr-oceanbase-asset-intelligence/) (2026-06-05): From Yang Youzhi of Baidu's PaddlePaddle AI Studio community, this article explains how to use PaddleOCR-VL-1.6 to parse unstructured enterprise documents into AI-consumable Markdown/JSON, then use OceanBase and seekdb to handle ingestion, vector search, and hybrid search — connecting the "parse → ingest → retrieve" first mile of enterprise asset intelligence.
- [Building an Agent System Solution on OceanBase & LangChain to Get Agents into Production Fast](https://longda.us/2026-06-02/2026-06-02-oceanbase-langchain-agent-solution/) (2026-06-02): At the Shanghai Meetup, LangChain & OceanBase community ambassador Canghai Jiusu unveiled AgentSeek — an agent engineering toolkit built on OceanBase and LangChain. Spanning a five-layer architecture of data foundation, context semantic layer, runtime, and more, it helps developers quickly build data loops and get agents into production.
- [LangChain's Crash Course? Using the Harness to Explain the Value and Practice of an All-in-one Data Foundation](https://longda.us/2026-05-26/2026-05-26-langchain-harness-allinone-data-base/) (2026-05-26): Starting from the definition and layered structure of the Harness, and drawing on the plugin-based design of the open-source project Bub and the Tape data-loop concept, this article explores the technical path toward a database-native Harness, as well as OceanBase's practice of using an All-in-one data foundation to support an internal agent data loop.
- [The OceanBase Community Opens Up Its Skill Universe!](https://longda.us/2026-05-22/2026-05-22-oceanbase-community-skill-universe/) (2026-05-22): The OceanBase community open-sources the oceanbase-skills repository, debuting deployment and operations Skills that let an AI Agent handle cluster deployment, tenant management, stress testing, and backup/restore in natural language. It also explores a new idea of storing Skills in a database for structured management, and introduces ClawMaster, an open-source tool for non-technical users.
- [Why Is Your Token Consumption So High? Money-Saving Tricks with OpenClaw](https://longda.us/2026-05-08/2026-05-08-openclaw-token-cost-saving/) (2026-05-08): OpenClaw's full-load MEMORY.md and lossy compaction are two big token black holes. seekdb M0 uses on-demand retrieval of cloud memory, rule-based tool-output compression, and a two-layer experience system to cut token consumption by 32% and raise the pass rate by 15 percentage points in AppWorld benchmarks.
- [How Do You Write a Workflow Skill? Patterns and Best Practices Distilled from 7 Top-Tier Projects](https://longda.us/2026-04-27/2026-04-27-workflow-skill-best-practices/) (2026-04-27): This article analyzes line by line 7 production-grade Skills from teams such as OpenAI, Google Labs, obra, and Trail of Bits, distilling 5 workflow Skill design patterns—linear process, decision tree, iterative loop, baton loop, and multi-phase checkpoints—along with general writing techniques for keeping the LLM from slacking off.
- [Token, Harness, OpenClaw, RAG, MCP, Agent—How Do They All Relate? One Diagram Makes It Clear](https://longda.us/2026-04-16/2026-04-16-ai-concepts-relationship-diagram/) (2026-04-16): Using the real task of "generating a new-product R&D slide deck with data charts," this article walks through the relationships and division of labor among core concepts—Token, LLM, Agent, Multi-Agent, MCP, RAG, vector database, Skill, Memory, Claude Code, OpenClaw, and Harness Engineering. After reading it, your AI anxiety will be gone.
- [bubseek — Turning an Agent's Footprints into Team Insights](https://longda.us/2026-04-15/2026-04-15-bubseek-agent-insights/) (2026-04-15): bubseek is a self-driven insight Agent built on the bub framework and OceanBase seekdb. With just a single prompt, it can autonomously connect to data sources, customize visualization dashboards, run analysis, and produce reports. Meanwhile, its own tape footprints are stored in seekdb and, once analyzed again, feed back into team insights.
- [Say Goodbye to Losing OpenClaw Configs—Mindkeeper Beta Invitation](https://longda.us/2026-04-10/2026-04-10-mindkeeper-beta-invitation/) (2026-04-10): Mindkeeper is a version management tool built specifically for AI Agent configuration files. Using a shadow Git repository, it automatically tracks every change to files like SOUL.md and AGENTS.md, supporting semantic summaries, diff comparison, and safe rollback. It offers both an OpenClaw plugin mode and a standalone CLI mode, and is now open for beta testing.
- [OpenClaw Ships Yet Another Release — How to Track Every Second of Its Spend in Real Time and Open the Agent "Thinking" Black Box](https://longda.us/2026-03-24/2026-03-24-openclaw-cost-tracking/) (2026-03-24): After OpenClaw's new release, the compute consumption of parallel multi-Agent and long-horizon tasks is harder to control. The open-source tool ClawProbe silently reads local state files and uses a differential algorithm to provide commands like status, top, cost, context, and compacts, tracking token throughput, cost, and context-compaction events in real time, opening the Agent runtime black box.
- [Why Does OpenClaw Get Better the More You Use It?](https://longda.us/2026-03-18/2026-03-18-why-openclaw-gets-better/) (2026-03-18): A deep look at the essence of OpenClaw \"getting better the more you use it\": through a read-write loop over md data files such as SOUL/USER/AGENTS/SKILL/MEMORY, conversation experience is distilled into searchable knowledge, letting the agent evolve with use. Plus hands-on advice and a preview of the OceanBase community course \"Easy Data x AI.\"
- [OpenClaw in Practice: One Mac, Six Agents, the Engineering Journey from \"Chatting\" to \"Getting Work Done\"](https://longda.us/2026-03-11/2026-03-11-openclaw-mac-six-agents/) (2026-03-11): On a single Mac, the author uses OpenClaw to build a multi-Agent team of one orchestrator plus five specialist Agents, with 52 cron jobs rotating around the clock. A deep retrospective on three core engineering problems, context management, a five-layer memory system, and a three-state communication protocol, plus real incident lessons.
- [OpenClaw + seekdb skills: Build Your Own Personal seekdb Assistant](https://longda.us/2026-03-04/2026-03-04-openclaw-seekdb-skills-assistant/) (2026-03-04): This article explains how to install seekdb Agent Skills in the personal AI assistant OpenClaw via pip or ClawHub, load the seekdb official documentation skill, and let the assistant answer common questions about seekdb deployment, vector search, hybrid search, and ecosystem integration in natural language.
- [EchoKit × OceanBase seekdb: An Open-Source, Localized Voice AI Framework](https://longda.us/2026-02-27/2026-02-27-echokit-seekdb-voice-ai/) (2026-02-27): EchoKit is a Rust-based voice AI framework open-sourced by the WasmEdge team. It supports local deployment and modular replacement of ASR/LLM/TTS, and chooses OceanBase seekdb as its knowledge base, leveraging hybrid search and built-in embedding to deliver a low-latency, privacy-preserving voice assistant.
- [Building Persistent Memory for an AI Agent With seekdb: From \"Full Context\" to \"Precise Recall\"](https://longda.us/2026-02-26/2026-02-26-seekdb-agent-persistent-memory/) (2026-02-26): This article explains how to build a vectorized persistent memory system for a Node.js AI Agent based on the seekdb-js SDK and Qwen3 Max (via OpenRouter), achieving precise recall through three strategies — threshold, fixed-count, and hybrid — with measured Token savings of 85%–95%.
- [uv × pyseekdb: Driving the Cost of a RAG Environment and Retrieval to a Minimum](https://longda.us/2026-02-13/2026-02-13-uv-pyseekdb-rag-setup/) (2026-02-13): This article introduces the combination of two tools, uv and pyseekdb. uv solves Python environment reproducibility with pyproject.toml and uv.lock, while pyseekdb covers vector, full-text, and hybrid search in both embedded and remote modes. It walks through the complete process of getting the official RAG demo running in five minutes.
- [Vibe Coding Notes — Google AI Studio](https://longda.us/2026-01-23/2026-01-23-vibe-coding-google-ai-studio/) (2026-01-23): A look at how to do Vibe Coding with the token-free Google AI Studio: through just four natural-language exchanges, we build a zero-code AI app that turns a rough sketch into a polished image in a chosen style, with one-click publishing and GitHub upload — an easy fix for the cover-image headache of running a WeChat blog.
- [Master the AI-Native Database with Natural Language — the seekdb MCP Server](https://longda.us/2026-01-22/2026-01-22-seekdb-mcp-server/) (2026-01-22): This article explains how to install and configure the seekdb MCP Server, covering both embedded and client/server deployment modes and the Stdio and SSE connection methods. Through a hands-on case of building a personal note knowledge base, it demonstrates how to use natural language to manage vector collections, perform semantic search, run hybrid search, and use AI memory.
- [Help the Cursor AI Assistant Instantly Get Vector Databases — A Guide to the Cursor seekdb Extension](https://longda.us/2026-01-14/2026-01-14-cursor-seekdb-extension-guide/) (2026-01-14): This article explains how to install and use the seekdb Cursor Extension to inject seekdb's official documentation into the .cursor/rules directory, giving the Cursor AI assistant knowledge of vector search, hybrid search, and more. It then demonstrates the actual results and inner workings with a Python hybrid search example.
- [Hybrid Search: A Hands-on Guide to Multimodal Retrieval (Chapter 4)](https://longda.us/2026-01-09/2026-01-09-hybrid-search-multimodal-guide/) (2026-01-09): This is Chapter 4 of the Agentic RAG series. It explains in detail how Hybrid Search fuses three modalities — vector search, sparse search, and full-text search — and walks through a complete, hands-on solution built on seekdb that combines weighted score fusion with an Agent that dynamically selects the search strategy.
- [A Spoon-Fed Tutorial (Bonus Edition) — Deploying Dify on K8s](https://longda.us/2026-01-04/2026-01-04-dify-on-k8s-tutorial/) (2026-01-04): This is the bonus, spoon-fed tutorial in the "Dify x OceanBase seekdb" series. It demonstrates how to deploy Dify—using seekdb as both its metadata database and vector database—on a K8s cluster with just three commands via Helm, and explains how to check Pod status, handle abnormal restarts, and access the service.
- [A Step-by-Step Tutorial II — A Guide to Using Dify x OceanBase seekdb](https://longda.us/2025-12-11/2025-12-11-dify-oceanbase-seekdb-guide/) (2025-12-11): A step-by-step tutorial detailing how to configure OceanBase seekdb as both the metadata database and the vector database in Dify v1.10.1, replacing a multi-component architecture in an integrated way, and quickly building AI applications such as knowledge bases and ChatBots via Alibaba Cloud Bailian models.
- [Building an Intelligent Book Search App from Scratch with seekdb](https://longda.us/2025-12-10/2025-12-10-seekdb-book-search-app/) (2025-12-10): Using an intelligent book search app as an example, this article walks you hands-on through data import, vector embedding, and HNSW index creation with seekdb, and implements semantic search, metadata filtering, and RRF hybrid search—giving you a taste of the integrated capabilities of an AI-Native database.
- [A Step-by-Step Tutorial — Building a RAG Application with OceanBase seekdb](https://longda.us/2025-12-05/2025-12-05-seekdb-rag-tutorial/) (2025-12-05): This is a step-by-step tutorial that walks you hands-on through building a RAG (Retrieval-Augmented Generation) system with OceanBase seekdb, covering the full flow of environment setup, configuring three Embedding function types, data import, vector retrieval, and launching a Streamlit app.
- [Explaining DeepSeek-OCR in the Plainest Language — The Disruptive Idea of Optical Context Compression](https://longda.us/2025-10-31/2025-10-31-deepseek-ocr-jinghua/) (2025-10-31): An in-depth reading of the disruptive idea behind the DeepSeek-OCR paper — optical context compression. It explores how to use pixels instead of text as LLM input to achieve a 10x compression ratio, and the profound implications this has for the design of future AI memory systems.
- [Exploring How AI Reshapes Database Products — obloader agent and OceanBase Agent](https://longda.us/2025-10-13/2025-10-11-ai-gaibian-shujuku-chanpin/) (2025-10-13): Exploring a new form of combining AI with database products. This article introduces how the obloader agent simplifies the use of data-loading tools through AIChat, and how the OceanBase Agent solves the multi-database management pain points of general-purpose MCP clients, enabling natural-language interaction to manage database instances.
- [Code Indexing in Practice in the Age of AI Coding](https://longda.us/2025-08-22/2025-08-22-ai-coding-code-index/) (2025-08-22): This article traces the four leaps of code search — from text matching to semantic retrieval, graph indexing, and agentic search — and shares hands-on experience building the CodeRepoIndex code-indexing tool on OceanBase vectorized indexes and private RAG.
- [A Step-by-Step Guide to Building an Agent with ModelScope x OceanBase MCP](https://longda.us/2025-08-15/2025-08-15-modelscope-oceanbase-mcp-agent/) (2025-08-15): OceanBase and ModelScope jointly introduce a new paradigm for AI-driven intelligent database management. Through MCP, developers can create, run, manage, diagnose, and analyze OceanBase clusters simply by chatting in natural language.
- [Dify + OceanBase: Multi-Scenario Practice for Landing AI Workloads](https://longda.us/2025-08-12/2025-08-12-dify-oceanbase-ai-multi-scenario/) (2025-08-12): This article shares the practice of choosing Dify on OceanBase to build AI applications, using OceanBase to consolidate PostgreSQL, Redis, and Weaviate into a single store, and introduces real-world experience from scenarios such as psychological assessment and intelligent customer service.
- [Context Engineering: A Code Documentation Retrieval Engine Built on OceanBase](https://longda.us/2025-08-08/2025-08-08-context-engineering-code-doc-retrieval/) (2025-08-08): Doc2Dev builds a code documentation retrieval engine on the OceanBase vector database, providing AI coding assistants with up-to-date documentation through semantic search and MCP integration, effectively eliminating code hallucinations.
- [Say Goodbye to Tedious Manual Analysis—OceanBase Agent Makes Database Operations Easy!](https://longda.us/2025-08-04/2025-08-04-oceanbase-agent-database-ops/) (2025-08-04): OceanBase Agent leverages large models to simplify daily database operations. With 70+ built-in SQL tools, support for custom Tools, Playbook task flows, and MCP extensions, a DBA can build their own operations Agent simply by writing SQL.
- [Agentic AI, Built on Dify x OceanBase in Practice](https://longda.us/2025-08-01/2025-08-01-agentic-ai-dify-oceanbase/) (2025-08-01): A core Dify contributor unpacks the Agentic AI trend and the platform capabilities it demands, introduces how OceanBase provides one-stop support for Dify across Database, Cache, and vector search, and shares a hands-on case of a conversational psychological assessment scale.
- [AI Replacing the Traditional GUI: An OBCloud Workflow Based on MCP](https://longda.us/2025-07-24/2025-07-24-mcp-obcloud-workflow/) (2025-07-24): Traditional database operations require clicking through the console level by level and repeatedly consulting documentation—slow and inefficient. From a frontend perspective, this article introduces how to build a brand-new operations workflow for OB Cloud based on the MCP protocol and large models, demonstrating the possibility of AI replacing the traditional GUI through cases such as instance diagnostics, configuration changes, and SQL visualization.
- [Hands-On: Building an MCP Advisor on OceanBase](https://longda.us/2025-07-11/2025-07-11-oceanbase-mcp-advisor/) (2025-07-11): Faced with the challenges of dynamic planning, service discovery, and usage complexity brought by tens of thousands of MCP tools, Ant Group Agent engineer Wang Yihui shares the design and practice of MCP Advisor: built on OceanBase's scalar-vector hybrid search and a multi-Provider architecture, it provides LLMs with MCP recommendation and installation capabilities, and lays out four stages of future development.
- [Dify + OceanBase + MCP: A Trio That Makes Building RAG Applications Easy](https://longda.us/2025-07-10/2025-07-10-dify-oceanbase-mcp-rag/) (2025-07-10): Through a hands-on case study, this article shows how to use Dify, OceanBase, and MCP to build a fully functional RAG application from scratch. It covers the complete workflow of deploying OceanBase, configuring Dify, indexing a knowledge base, creating a chat assistant, and turning a Dify application into an MCP Server that external clients can call.
- [AI Application Development: An Innovative Unstructured Document Analysis System That Avoids the Limitations of Traditional Approaches](https://longda.us/2025-07-04/2025-07-04-ai-app-unstructured-document-analysis/) (2025-07-04): A team from Beijing Institute of Technology built QUEST, an intelligent unstructured document analysis system on OceanBase. Through sampling-based selectivity estimation, token cost prediction, coordinated filter-join optimization, and precise RAG-based localization, it breaks through the high-cost bottleneck of traditional full extraction, achieving higher accuracy, lower cost, and shorter latency.

### Enterprise Cases (27)
- [NetEase Open-Sources lb-driver: Driver-Layer Load Balancing for OceanBase OBProxy](https://longda.us/2026-05-12/2026-05-12-netease-lb-driver-obproxy-load-balancing/) (2026-05-12): After migrating its core business from the DDB sharding middleware to OceanBase, NetEase Yunxin built and open-sourced lb-driver to eliminate the link latency and operational pain caused by SLB load balancing. It implements Statement-granularity OBProxy load balancing, health checks, and dynamic scaling directly in the JDBC driver layer.
- [More Than Just Replacing HBase: How Baofu Payment Leveraged OceanBase to Build a Future-Ready, Unified 'TP+AP+KV+AI' Data Foundation](https://longda.us/2026-01-21/2026-01-21-baofu-payment-oceanbase-unified-data-base/) (2026-01-21): The Baofu Payment data team shares its complete practice of replacing HBase with OceanBase's OBKV-HBase — from architectural pain points and selection considerations, to a five-step smooth migration, data verification, and stress-test tuning — ultimately achieving 420,000 QPS at 1ms latency and building a unified TP+AP+KV+AI data foundation.
- [Tens of Millions of Members, Billions of Transactions: When the CRM System Buckles, How Did a Leading Pharma Company Upgrade to OceanBase for Real-Time, Precision Marketing?](https://longda.us/2026-01-16/2026-01-16-pharma-crm-database-upgrade/) (2026-01-16): Facing tens of millions of members, billions of transactions, and multidimensional behavioral data, the CRM membership system of Chongqing Pharmaceutical Group was overwhelmed by its traditional database. After comparing three domestic distributed databases, it chose OceanBase — achieving a 25.7x boost in complex-query efficiency, 60% storage savings, and support for real-time, precision marketing.
- [The Data Foundation for Zuoyebang's AI Business: OceanBase Vector Database Multi-cloud Deployment and Architecture Design Optimization](https://longda.us/2025-12-31/2025-12-31-zuoyebang-vector-database-multicloud/) (2025-12-31): Zuoyebang built the data foundation for its AI business on OceanBase. This article details vector database selection, cross-cloud primary-standby tenant replication, read-write splitting, cloud-native proxies, and other multi-cloud architecture designs, and shares large-scale operations experience and cost savings at a scale of 40+ clusters and 20,000+ cores.
- [OceanBase at Didi: Large-Scale Operations Experience and New Feature Adoption](https://longda.us/2025-12-17/2025-12-17-didi-oceanbase-ops-practice/) (2025-12-17): Since 2024, Didi has used OceanBase to replace RocksDB and TokuDB. Using its ride-hailing growth service and core archive database as examples, this article shares large-scale operations experience and new feature practices around hundred-TB data migration, high-concurrency throttling, custom monitoring and alerting, and Binlog Server high availability.
- [vivo's Domestic Database Technology Reserve: Using OceanBase to Break Through the Storage and Performance Bottlenecks of Large-Scale Data](https://longda.us/2025-11-13/2025-11-13-vivo-database-technology/) (2025-11-13): vivo's internet business introduced OceanBase to replace MySQL, covering four scenarios — data archiving, horizontal sharding, high concurrency, and HTAP — achieving a 60% improvement in high-concurrency update efficiency, an 80% improvement in complex query performance, and a 50% reduction in storage cost, while building a unified in-house operations platform.
- [Migrating from a Sharded MyCat Architecture to OceanBase | Lessons from Belle Fashion's Core Financial System Migration](https://longda.us/2025-10-31/2025-10-31-baili-mycat-qianyi-oceanbase/) (2025-10-31): Belle Fashion Group migrated its core financial system from a sharded MyCat architecture to OceanBase, achieving a 30x performance improvement, a 96.7% reduction in storage cost, and a 59.4% reduction in hardware cost. This article details its three-step migration experience covering data flow, data verification, and SQL compatibility tuning.
- [How to Build a Mature, Stable, Cost-Effective, and Easy-to-Use AI Application — Quwan Technology's Hands-on Practice Building an Employee Assistant on OceanBase](https://longda.us/2025-10-22/2025-10-22-quwan-ai-yuangong-zhushou/) (2025-10-22): When selecting the underlying database for its custom AI assistant, Quwan Technology compared multiple vector databases and chose OceanBase. This article shares its thinking and experience in building a three-in-one database foundation for scalar, vector, and full-text search that meets the multiple needs of AI applications.
- [How to Avoid Pointless Work — Qifu Technology Cuts Code-Refactoring Costs by 90% After Ditching Sharding with OceanBase](https://longda.us/2025-10-16/2025-10-16-qifu-fenkufenbiao-gaizao/) (2025-10-16): Qifu Technology adopted OceanBase to replace part of its MySQL and TiDB usage, cutting code-refactoring costs by 90%, halving storage costs, reducing servers by 30+, and shortening task execution time by 40%. This article shares the database pain points across multiple scenarios and Qifu's hands-on experience deploying OceanBase.
- [Migrating a Distributed Database to OceanBase — A Smooth Migration Built on NetEase Cloud Music's In-house CDC Service](https://longda.us/2025-09-25/2025-09-25-netease-music-cdc-migration-oceanbase/) (2025-09-25): NetEase Cloud Music shares how it migrated its PB-scale sharded DDB database to the native distributed database OceanBase. Built on its in-house CDC service NDC, the migration was transparent to the business and lost zero data, with forward synchronization reaching GB/s throughput and storage shrinking by at least 1/4.
- [A Step-by-Step Guide to Optimizing Vector Search: A Game Company's Intelligent Customer Service and Recommendation System on OceanBase](https://longda.us/2025-09-11/2025-09-11-game-vector-search-oceanbase/) (2025-09-11): A card game company built an intelligent customer service and UGC community recommendation system on OceanBase's vector database. Through HNSW indexing, supplementary keyword retrieval, and intelligent reranking, it achieved millisecond-level retrieval, a 300% boost in customer service efficiency, a 75% reduction in recommendation latency, and a 75% reduction in operational complexity.
- [NetEase Personal Mail Upgrades Its Database to OceanBase: A Dual Breakthrough in Reliability and Stability](https://longda.us/2025-09-02/2025-09-02-netease-mail-oceanbase-upgrade/) (2025-09-02): NetEase Personal Mail upgraded its sharded MySQL architecture to OceanBase. Through technology selection comparison, unique-key governance, partition and index optimization, and OBKV-Redis practice, it achieved 72% storage cost savings, a significant QPS increase, and a real-time analytics upgrade.
- [Dmall and OceanBase: Achieving Full Inter-Tenant Resource Isolation and Low-Cost System Upgrades](https://longda.us/2025-08-27/2025-08-27-dmall-oceanbase-resource-isolation/) (2025-08-27): Dmall shares its hands-on experience migrating from MySQL to OceanBase in a retail SaaS scenario — using multi-tenant resource isolation, transparent scaling, and a high compression ratio to cut storage costs by over 80%, achieving cost reduction, efficiency gains, and low-cost system upgrades.
- [70% Performance Boost — CR Vanguard Upgrades Its Core Database to OceanBase](https://longda.us/2025-08-20/2025-08-20-crv-vanguard-database-upgrade/) (2025-08-20): CR Vanguard's subsidiary Wanjia Digital upgraded a middleware-based MySQL sharded cluster to OceanBase, achieving a seamless, business-transparent migration — system performance up 70%, storage footprint down about 60%, and hardware costs cut by 50%.
- [Hardware Costs Down 52%: How 99Bill Reduced Costs and Boosted Efficiency After Adopting OceanBase](https://longda.us/2025-08-13/2025-08-13-99bill-oceanbase-cost-reduction/) (2025-08-13): A 99Bill DBA shares the three transformations of their DB architecture — from MySQL primary-standby and MyCAT sharding to OceanBase — detailing five migration lessons that ultimately cut hardware costs by 52%.
- [From v3.1 to v4.3, OceanBase Steadily Powers Kuaishou's PB-Scale Core Business Scenarios](https://longda.us/2025-08-06/2025-08-06-kuaishou-oceanbase-pb-scale/) (2025-08-06): Kuaishou migrated from MySQL sharding to OceanBase, covering PB-scale core businesses such as transaction reconciliation and payments, and upgraded from version 3.1 to 4.3—cutting machine costs by 31% and saving 50% of resources.
- [Tongfang Smart Energy: OceanBase Powers a Secure, Reliable, High-Performance Energy Data Foundation](https://longda.us/2025-08-05/2025-08-05-tongfang-smart-energy-oceanbase/) (2025-08-05): After comparing HBase-Phoenix, Ignite, and TiDB for its new-energy and metro projects, Tongfang Smart Energy chose OceanBase 4.2.1, achieving twice the performance of MySQL 8.0, replacing a complex Hadoop architecture, and building a highly reliable, easy-to-maintain energy data foundation.
- [Escaping Three Major Dilemmas, Cultivating Five Key Scenarios | How TAL Education's AI Business Cut Costs by 86% with OceanBase](https://longda.us/2025-07-30/2025-07-30-tal-ai-oceanbase-cost-reduction/) (2025-07-30): Facing three major dilemmas—sharding, resource isolation, and resource fragmentation—the TAL Education Group adopted OceanBase and deeply cultivated five scenarios: learning machines, multi-data-center disaster recovery, AI business, online-school reports, and content moderation. By leveraging partitioned tables, columnar storage, multi-tenancy, and cloud-disk elastic scaling, AI-business storage costs dropped 86% per month, and report performance improved 60% over MySQL.
- [Weibo's Database Architecture Upgrade with OceanBase: Both Stable and Cost-Saving](https://longda.us/2025-07-23/2025-07-23-weibo-partition-table-practice/) (2025-07-23): Yang Shanggang, head of Weibo's database technology, takes a deep dive into Weibo's database architecture upgrade: replacing MySQL sharding with OceanBase partitioned tables, cutting storage usage from 50TB to 27TB and saving 46% in cost, and sharing experience on partitioned table design and statistics gathering, along with plans to explore KV and vector scenarios.
- [OceanBase: Du Xiaoman Breaks Down Database Selection Across Five Technical Scenarios, Doubling Cost Savings, Performance, and Efficiency](https://longda.us/2025-07-16/2025-07-16-duxiaoman-database-selection/) (2025-07-16): Zhao Hui, head of Du Xiaoman's Technology Committee, shares their unified storage architecture practice: selecting OceanBase based on four cost advantages (migration, resources, learning, operations), and validating it across five scenarios—low-latency queries over massive data, real-time data warehouse read/write, high-throughput variable writes, vector hybrid queries, and KV read/write—achieving double the delivery efficiency and over 3x cost benefits.
- [China Unicom's ChatDBA Upgrades Its Vector Database: A Hands-On Comparison of Milvus, Elasticsearch, and OceanBase](https://longda.us/2025-07-09/2025-07-09-china-unicom-chatdba-vector-database/) (2025-07-09): China Unicom shares the journey of building ChatDBA, its intelligent database expert, explaining why it upgraded from MySQL+Milvus to the OceanBase vector database, and publishing the results of a hands-on functional and performance comparison among Milvus, Elasticsearch, and OceanBase.
- [Handling Hundreds of Millions of New Records Daily: How BOSS Zhipin Built a High-Performance, Efficient Storage Solution on OceanBase](https://longda.us/2025-06-30/2025-06-30-boss-zhipin-high-performance-storage/) (2025-06-30): Facing hundreds of millions of new chat-log records every day, BOSS Zhipin compared products such as MySQL and ClickHouse before choosing OceanBase to build a historical archive database, cutting storage resources by over 70% and gradually expanding to core workloads such as chat messages, delivering a triple win in cost, efficiency, and stability.
- [OceanBase Vector Search at Lalamove: Exploration and Practice](https://longda.us/2025-06-25/2025-06-25-huolala-oceanbase-vector-search/) (2025-06-25): Lalamove shares its exploration and practice with OceanBase vector search: starting from pain points of its existing vector database—dynamic schemas, hybrid search, and operational difficulty—it completed its technology selection and implemented LLM application scenarios such as financial-loss code detection and a data warehouse AI Q&A assistant.
- [OceanBase Vector Technology Tackles 360's Three Commercialization Pain Points, Accelerating AI-Driven Business Analytics by 80%](https://longda.us/2025-06-23/2025-06-23-360-oceanbase-vector-technology/) (2025-06-23): 360's Commercialization business line shares the application of OceanBase in scenarios such as real-time advertising reports and vector storage, solving three pain points—OOM, high concurrency, and uneven resource utilization—and improving business analytics efficiency by 80%.
- [Zhihu's Large-Scale OceanBase Adoption and Joint Ecosystem Building](https://longda.us/2025-06-18/2025-06-18-zhihu-oceanbase-practice/) (2025-06-18): Zhihu's database lead shares the journey of deploying OceanBase in Zhihu's core scenarios, including multi-tenant isolation, the introduction of OBKV, the use of the ecosystem toolchain, and experience in joint community building.
- [NetEase Games Brings OceanBase into Its DB SaaS: Storage Cost Cut by 60%, Backup and Recovery 3x Faster](https://longda.us/2025-06-12/2025-06-12-netease-game-db-saas-oceanbase/) (2025-06-12): The NetEase Games DB SaaS platform brought in the OceanBase distributed database to address pain points such as high concurrency, data synchronization, and storage cost, achieving a storage-cost reduction of more than 60% and a 3x improvement in backup and recovery efficiency.
- [Conquering the \"Four Mountains\": Qingdao Yuno's Distributed Database Upgrade Strategy on OceanBase for Pharma Retail](https://longda.us/2025-05-27/2025-05-27-pharma-retail-distributed-database/) (2025-05-27): Qingdao Yuno's journey of digital transformation in pharma retail: migrating from traditional databases to OceanBase to overcome architecture flexibility, query performance, and storage cost challenges—achieving 20x faster queries and 60% storage compression.

### Technical Deep Dive (25)
- [From Neurons to Code Engineering: The Forgetting Design of the PowerMem Memory System](https://longda.us/2026-06-08/2026-06-08-powermem-forgetting-design/) (2026-06-08): Starting from synaptic plasticity, memory consolidation, information theory, and the Ebbinghaus forgetting curve, this article unpacks the forgetting mechanism behind the PowerMem memory system: a three-tier memory architecture, an exponential decay model, and retrieval ranking weights, explaining why forgetting is a core capability of an Agent memory system.
- [A Deep "Dissection" of the AI Agent Harness](https://longda.us/2026-05-25/2026-05-25-ai-agent-harness-deep-dive/) (2026-05-25): This article introduces Akshay Pachaar's long-form piece "The Anatomy of an Agent Harness," systematically breaking down the Agent Harness architectures of Anthropic, OpenAI, and LangChain — 12 core components such as the orchestration loop, tools, memory, and context management, plus the 7 key decisions that define a Harness.
- [A Deep Dive into LLM Wiki / Obsidian-Wiki / GBrain: The \"Self-Organization\" and \"Self-Evolution\" of Knowledge in the Agent Era](https://longda.us/2026-04-28/2026-04-28-llm-wiki-knowledge-self-organization/) (2026-04-28): From a knowledge-engineering perspective, this article dissects the designs of Karpathy's LLM Wiki, Obsidian-Wiki, and GBrain, analyzing the self-organization and self-evolution of knowledge in the Agent era, and how Skillify turns scattered material into continuously evolving structured memory.
- [How Agent / Skills / Teams Architectures Evolve, and How to Choose Among Them](https://longda.us/2026-03-10/2026-03-10-agent-skills-teams-architecture-evolution/) (2026-03-10): This article traces the evolution of Agent architectures from Single Agent and Multi-Agent to Agent Skills and Agent Teams. Drawing on a Google DeepMind paper and Anthropic's practical experience, it distills a methodology of "start simple, scale only when needed" for choosing an Agent architecture.
- [How Does OpenClaw Make AI Feel "Human"?](https://longda.us/2026-02-25/2026-02-25-openclaw-humanlike-ai/) (2026-02-25): This article takes the form of study notes to unpack the engineering mechanisms OpenClaw uses to make AI feel "human" — live context assembled at runtime, tiered file permissions, U-shaped attention ordering, hybrid memory retrieval, and heartbeat-driven proactive evolution. It also includes the growth record of a digital daughter, Luna, from an empty program to the awakening of self-awareness.
- [A Guide to Optimizing Vector Indexes in OceanBase](https://longda.us/2025-12-19/2025-12-19-oceanbase-vector-index-optimization/) (2025-12-19): Xia Jin, a vector-index engineer on OceanBase, takes a deep dive into how OceanBase / seekdb vector indexes are built and the structure of their five auxiliary tables, analyzes the root causes of slow index builds and high memory usage, and offers optimization methods including adding parallelism, raising the sampling rate, switching to partitioned tables, and rebuilding.
- [The Mystery of OceanBase Session IDs](https://longda.us/2025-11-21/2025-11-21-oceanbase-session-id-mystery/) (2025-11-21): This article systematically untangles the design logic of Session IDs in OceanBase. The two connection types — C connections and S connections — correspond to four encodings (cs_id, proxy_sessid, server_sessid, ss_id). Through a range of query-command examples and the pass-through / non-pass-through behavior of the kill command, it builds a framework for understanding and troubleshooting session issues.
- [Explaining AI Memory in the Plainest Possible Language](https://longda.us/2025-11-18/2025-11-18-ai-memory-explained/) (2025-11-18): This article explains AI Memory in plain language. Starting from how the human brain classifies and operates on memory, it compares the storage areas and storage forms of agent memory, then dives into the implementation principles of mainstream memory frameworks such as Letta, Zep, and Mem0, and summarizes the technical trends of agent memory and the demands it places on underlying storage.
- [OceanBase Partitioning Fundamentals](https://longda.us/2025-11-14/2025-11-14-oceanbase-partition-basics/) (2025-11-14): A systematic introduction to OceanBase partitioning fundamentals, covering the role of partitioning in partition pruning, data maintenance, and data distribution; the applicable scenarios and design points of the three basic partitioning methods — Hash, Range, and List; and flexible partition management capabilities such as dynamic partitioning, automatic partition splitting, and composite secondary partitioning.
- [The Database Evolution of the AI Era — From Vectors to Hybrid Search](https://longda.us/2025-11-07/2025-11-07-database-evolution-vector-hybrid-search/) (2025-11-07): An accessible, easy-to-understand explainer. Starting from the classification of structured, semi-structured, and unstructured data, it explains why vector databases became a must-have in the AI era, how multi-path hybrid retrieval over scalar, full-text, and vector data became the engine of knowledge bases and AI applications, and looks ahead to databases evolving toward built-in AI Functions.
- [A New Way to Tackle "Read Amplification" — Merge-On-Write Tables in OceanBase](https://longda.us/2025-11-06/2025-11-06-oceanbase-merge-on-write/) (2025-11-06): An introduction to the Merge-On-Write table launched in OceanBase 4.3.5. By splitting an update into delete + insert and writing the full set of columns, it solves the read amplification of Merge-On-Read queries under an LSM Tree architecture, significantly boosting real-time analytics performance in update-heavy workloads.
- [Key Techniques and Methods for OceanBase Database Diagnosis and Tuning](https://longda.us/2025-09-18/2025-09-18-oceanbase-diagnosis-tuning/) (2025-09-18): This article systematically reviews the fault classification and root causes of OceanBase, proposes a five-step diagnosis and tuning process—problem identification, data collection, problem localization, solution formulation, and validation—and introduces key tools and methods such as internal views, log analysis, OCP, OAS, and obdiag.
- [When the Intelligent Robot Says "Bro, the Language Just Doesn't Connect"—How Should You, the Developer, Respond?](https://longda.us/2025-09-12/2025-09-12-ai-robot-language-developer/) (2025-09-12): When an intelligent robot cannot understand minor languages such as Thai, developers can use OceanBase's plugin mechanism to respond quickly to requirements. This article explains the plugin architecture's low coupling, openness, and self-controllability, as well as how to develop and use tokenizer plugins and external table plugins—helping you bypass the long kernel development cycle and quickly extend database capabilities.
- [Getting to Know ASH — Opening the Database's "Treasure Box of Time"](https://longda.us/2025-09-01/2025-09-01-ash-database-monitoring/) (2025-09-01): ASH (Active Session History) is the OceanBase database's "intelligent monitoring system," taking a state snapshot of active sessions every second. Through three zero-prerequisite experiments, this article lets you experience the charm of ASH and takes a deep dive into its implementation — selective recording, time slicing, and the ring buffer.
- [Safeguarding Performance in High-Concurrency Scenarios with OceanBase](https://longda.us/2025-08-26/2025-08-26-oceanbase-high-concurrency-performance/) (2025-08-26): From the design of kernel components — plan management, concurrency control, multi-dimensional caching, hot-row optimization, log aggregation, and compilation optimization — this article explains the key factors behind how OceanBase safeguards performance in high-concurrency scenarios.
- [Hybrid Query Methods for Vector Indexes — Did You Pick the Right One?](https://longda.us/2025-08-19/2025-08-19-vector-index-hybrid-query/) (2025-08-19): This article explains the principles and usage of hybrid queries on OceanBase vector indexes, including how to choose among the three scalar-filtering methods — Pre-filtering, In-filtering, and Iterative-Ann — as well as fusion ranking methods for full-text and vector results, and the product roadmap.
- [Automatic Partition Splitting — An Epic Usability Upgrade for OceanBase](https://longda.us/2025-08-11/2025-08-11-oceanbase-auto-partition-split/) (2025-08-11): A deep dive into OceanBase's automatic partition splitting feature: the pain points of manual partitioning, the principles, characteristics, applicable scenarios, and usage of automatic splitting, plus prerequisite knowledge on partitioned tables, helping you make full use of distributed capabilities without planning partitions yourself.
- [OceanBase: How Does It Achieve Extreme Cost Reduction Across Hardware, Storage, and Operations?](https://longda.us/2025-07-29/2025-07-29-extreme-cost-reduction-analysis/) (2025-07-29): This article takes a deep dive into the core technologies behind OceanBase's extreme cost reduction across hardware, storage, and operations, including its standalone-distributed integrated architecture, on-demand loading of metadata for millions of partitions, disk I/O isolation, Compaction optimization, and encoding and compression. It illustrates the cost-reduction results with enterprise cases from Sichuan Huadi, 58.com, Energy Monster, Trip.com, Dmall, and Yunji.
- [When OceanBase on K8s Meets AI — A Deep Dive into the Design of okctl-mcp-server](https://longda.us/2025-07-18/2025-07-18-okctl-mcp-server-design/) (2025-07-18): In this article, OceanBase community contributor Li Ziyi takes a deep dive into the design of okctl-mcp-server, explaining how the MCP protocol lets AI perform complex operations on OceanBase on K8s—such as cluster creation and tenant management—through natural language, and discussing optimizations like dynamic loading and long-task polling along with current limitations.
- [The Architecture Design and Optimization of OceanBase's Standalone-Distributed Integrated Database](https://longda.us/2025-07-08/2025-07-08-standalone-distributed-integrated-database/) (2025-07-08): Zhifeng Yang, General Manager and Chief Architect of the OceanBase product, offers a technical interpretation of the standalone-distributed integrated architecture: reviewing OceanBase's three architectural upgrades and analyzing how dynamic log streams eliminate distributed overhead and enable flexible switching between standalone and distributed forms, supporting the full lifecycle of a business.
- [Using OpenManus to Build an Auto-Diagnosis Agent: Pinpoint Database Anomalies in 30 Minutes](https://longda.us/2025-06-27/2025-06-27-openmanus-diagnostic-agent/) (2025-06-27): Building a database auto-diagnosis Agent based on the OpenManus framework to quickly pinpoint database anomalies in 30 minutes, including the complete development process and a hands-on case study.
- [OceanBase Vector Database Boosts AI Retrieval Efficiency by 45x and Halves AI Agent Development Costs](https://longda.us/2025-06-26/2025-06-26-oceanbase-vector-db-ai-search/) (2025-06-26): A comprehensive look at OceanBase 4.3's vector database capabilities, supporting HNSW indexes, hybrid search, and multimodal vectors, boosting AI retrieval efficiency by 45x and reducing development costs by 50%.
- [LLM vs. Small Models: The Right Way to Build an AI Agent for Domestic Database Operations](https://longda.us/2025-06-20/2025-06-20-ai-agent-database-ops/) (2025-06-20): Exploring how to use small models instead of full-scale LLMs for intelligent operations on domestic databases, enhancing reasoning ability through context, and building three AI Agents for alert analysis, SQL optimization, and inspection.
- [How Database Technology and Solutions Support the Data Needs of the Pan-Internet Industry](https://longda.us/2025-05-30/2025-05-30-database-technology-support-pan-inter-industry/) (2025-05-30): Driven by the digital economy, this article explores how the pan-internet industry copes with the systemic challenges of exponential data growth. It examines the evolution of distributed database architectures, including core technical approaches such as sharding, the LSM-Tree storage structure, multi-replica high-availability designs, and hybrid row-column storage.
- [A Quick Introduction to Vector Databases](https://longda.us/2025-05-25/2025-05-25-vector-database-example/) (2025-05-25): A gentle introduction to vector database fundamentals: Embedding techniques, similarity metrics, search algorithms (HNSW, IVF, DiskANN), and index compression (SQ, PQ). Perfect for AI and database beginners.

### Hands-on Practice (11)
- [Vector Database Best Practices Distilled by OceanBase over Three Years](https://longda.us/2026-06-01/2026-06-01-vector-database-best-practices/) (2026-06-01): Two OceanBase vector database experts distill three years of hands-on PoC experience, systematically covering vector index selection, memory and disk resource estimation, partition design, index parameter configuration, hybrid query tuning, performance validation, and common performance troubleshooting methods.
- [Stop Following the Crowd Blindly: The Core Principles Behind Seven Companies' Real-World Vector Database Choices](https://longda.us/2026-04-09/2026-04-09-vector-database-selection-principles/) (2026-04-09): This article reviews the real thinking and practice of seven companies—including Qihoo 360, China Unicom, Lalamove, Weibo, and Quwan Technology—in their vector database selection, comparing Milvus and OceanBase across performance, stability, operations, and TCO, and distilling replicable core selection principles.
- [My Editor Job Has Been Replaced by AI — A seekdb Fork Table Capability Test Report](https://longda.us/2026-04-01/2026-04-01-seekdb-fork-table-test-report/) (2026-04-01): An experiment report on having an AI Agent automatically deploy seekdb and test its Fork Table capability. Through five scenarios, data isolation, version rollback, a synthetic-data sandbox, and multi-branch A/B experiments, it demonstrates the value Fork Table brings to AI developers as lightweight data-branch management.
- [Saving My Company $5,000 in Just Two Weekends: Refactoring a Knowledge Base from a Multi-Database Architecture to an AI-Native Database](https://longda.us/2025-12-25/2025-12-25-ai-native-database-knowledge-base-refactor/) (2025-12-25): In two weekends, the author refactored an enterprise knowledge base from a four-database stack (PostgreSQL + Elasticsearch + Pinecone + Redis) into a single-database solution built on seekdb, OceanBase's open-source AI-native database. Query latency dropped from 120ms to 58ms, saving $450 a month, with complete runnable code and tuning experience included.
- [A Guide to Using the OceanBase Vector Database](https://longda.us/2025-09-24/2025-09-24-oceanbase-vector-db-guide/) (2025-09-24): The official guide to using the OceanBase vector database: a detailed walkthrough of vector index types such as HNSW, HNSW_SQ, HNSW_BQ, IVF, and IVF_PQ, covering type selection, parameter tuning, memory estimation, and partitioned-table practice, with performance optimization recommendations spanning data volumes from millions to hundreds of millions.
- [Deploying the OceanBase Database in an Ubuntu Virtual Machine](https://longda.us/2025-09-19/2025-09-19-ubuntu-vm-deploy-oceanbase/) (2025-09-19): This article shares the complete process of deploying a standalone OceanBase Community Edition database via OBD's GUI in a WSL Ubuntu 22.04 virtual machine on a Windows laptop. It covers environment preparation, how to install RPM packages under Ubuntu, cluster and component configuration, the principles behind ob_configserver, and the OB Dashboard operations experience.
- [A Practical Guide to Vector Databases for Personalized Recommendation in UGC Communities](https://longda.us/2025-08-25/2025-08-25-vector-db-ugc-recommendation/) (2025-08-25): A practical guide to building a personalized recommendation system for UGC communities on OceanBase's native vector capabilities: dual-vector user interest modeling (short-term + long-term), multi-path recall and fusion ranking in a single SQL statement, with schema design, performance tuning, and a two-week MVP roadmap.
- [A New Open-Source Community Development Experience in the Vibe Coding Era](https://longda.us/2025-07-28/2025-07-28-vibe-coding-opensource-community/) (2025-07-28): Drawing on the real experience of an OceanBase community developer, this article shows how to use Cursor together with the OceanBase MCP Server to create tables and insert data through natural-language conversation, and to spin up a FastAPI backend service connected to OceanBase in just 5 minutes—giving you a feel for the brand-new development paradigm that Vibe Coding brings.
- [OBCP V4 Official Mock Exam Analysis (with Answers)](https://longda.us/2025-07-22/2025-07-22-obcp-v4-mock-exam-analysis/) (2025-07-22): This article compiles the 50 mock questions publicly available on the OBCP V4 official site—covering single-choice, multiple-choice, and true/false questions—and gives an answer analysis for each based on the official tutorial content, covering core topics such as LDC routing, global indexes, compaction, OMS migration, and log streams, to help candidates prepare.
- [The Evolution of OceanBase Auto-increment Columns and Best Practices](https://longda.us/2025-07-21/2025-07-21-oceanbase-auto-increment-evolution/) (2025-07-21): This article traces the evolution of OceanBase auto-increment columns from 4.0.0 to 4.2.3, compares the principles and jump behavior of the ORDER and NOORDER modes, summarizes best practices for bigint/int selection, mode switching, and cache size tuning, and introduces the Sequence extension in MySQL mode.
- [OceanBase PoC Lessons Learned (Part 2) — AP Workloads](https://longda.us/2025-07-07/2025-07-07-oceanbase-poc-ap-business/) (2025-07-07): This article summarizes hands-on experience from OceanBase AP workload PoCs, centered on business migration strategy and database object design. It covers data clustering, data distribution, compaction strategy, partition management, collation, indexing, the choice between row and column storage, and common problems with incremental data and how to solve them.

### Product Deep Dive (9)
- [OceanBase seekdb 1.3.0 Released: 22x Performance Gain, Jitter-Free P99](https://longda.us/2026-06-09/2026-06-09-seekdb-1-3-0-release/) (2026-06-09): OceanBase seekdb 1.3.0 is here, introducing an asynchronous index model based on Change Stream that decouples writes from index building. In streaming scenarios it boosts throughput by roughly 22x with virtually no P99 jitter under concurrency, and adds capabilities such as Fork/Diff & Merge for vector columns.
- [seekdb 1.2.0 Released: Primary-Standby Disaster Recovery Goes Live, and Full-Database Cloning in Seconds Is Here](https://longda.us/2026-04-03/2026-04-03-seekdb-1-2-0-release/) (2026-04-03): seekdb 1.2.0 is officially released, bringing its first high-availability solution—a primary-standby asynchronous replication architecture supporting both Switchover and Failover modes. It also adds full-database, second-level cloning via Fork Database and data version management via Diff & Merge, helping AI applications move toward production readiness.
- [Giving OpenClaw Long-Term Memory — PowerMem 1.0.0 Officially Released](https://longda.us/2026-03-19/2026-03-19-powermem-1-0-0-release/) (2026-03-19): OceanBase's open-source intelligent memory system PowerMem 1.0.0 is officially released: a new pmem CLI operations plane and a Dashboard cognition plane, five integration methods including SDK, HTTP API, and MCP, plus an OpenClaw long-term-memory plugin that cuts token consumption to 18% of the default approach.
- [From Vibe Coding to AI Agents — OceanBase seekdb Builds a Millisecond-Level Data Sandbox for the AI Era](https://longda.us/2026-03-03/2026-03-03-seekdb-ai-data-sandbox/) (2026-03-03): OceanBase seekdb 1.1.0 introduces the Fork Table feature. Based on copy-on-write and consistent snapshots, it creates logically independent, read-write-isolated data branches in milliseconds, making data as branchable and rollbackable as Git — powering AI scenarios such as Vibe Coding, A/B testing, and multi-agent systems.
- [Goodbye to the Patchwork: A One-Stop Tech Stack for Memory, Retrieval, and the AI Data Engine (Part 2)](https://longda.us/2026-01-29/2026-01-29-unified-ai-data-stack-part2/) (2026-01-29): This article analyzes RAG's architectural evolution from Naive and Advanced to Modular RAG and its production pain points, and introduces PowerRAG — an open-source product deeply optimized on top of RAGFlow with OceanBase as an integrated foundation — covering its three core modules of document parsing, hybrid retrieval, and effectiveness evaluation, along with typical use cases.
- [Goodbye to the Patchwork: A One-Stop Tech Stack for Memory, Retrieval, and the AI Data Engine (Part 1)](https://longda.us/2026-01-28/2026-01-28-unified-ai-data-stack-part1/) (2026-01-28): This article dissects the new database needs and pain points developers face in the AI era, and introduces OceanBase seekdb — a lightweight, multimodal, AI-native database with a unified interface, multimodal storage, hybrid retrieval, and built-in AI functions. It demonstrates building a RAG app in three steps and the elastic evolution path from embedded to distributed deployment.
- [A Deep Dive into the OceanBase Ecosystem Toolchain — OAT / obd / OCP / obshell](https://longda.us/2026-01-08/2026-01-08-oceanbase-ecosystem-tools/) (2026-01-08): Zhimin, the product owner for OceanBase ecosystem tools, takes a deep dive into the roles and relationships of the four operations and management tools — OAT, obd, OCP, and obshell / obshell Dashboard. He offers tool-combination recommendations for different business stages and lays out an online data migration path for upgrading from the community edition to the enterprise edition.
- [A Developer's View of OceanBase's Open-Source AI Product Trio](https://longda.us/2025-12-26/2025-12-26-oceanbase-ai-products-developer-view/) (2025-12-26): An OceanBase engineer offers a developer's take on three open-source AI products—seekdb, PowerRAG, and PowerMem—dissecting the three big challenges of the AI era (multimodal data, context rot, and memory management) and explaining how, under the Data×AI strategy, the three work together to build a complete AI data infrastructure.
- [In the AI Era, Why Did OceanBase Open Source an AI-Native Database Like seekdb?](https://longda.us/2025-11-26/2025-11-26-why-oceanbase-opensource-seekdb/) (2025-11-26): Feng Zhongyan (Lao Ji), General Manager of OceanBase's Open Source Ecosystem, systematically answers three questions—why seekdb was open sourced, how it differs in positioning from OceanBase, and whether the open source commitment is long-term—and explains the design thinking behind the AI-native hybrid search database seekdb in terms of data fusion, AI Inside, and lightweight architecture.

### Travel (9)
- [Strolling Through Jiuzhaigou](https://longda.us/2021-10-10/jiuzhaigou/) (2021-10-10): A Jiuzhaigou travelogue: the National Day road-trip traffic jam from Chengdu to Jiuzhaigou, and a recommended walking route from Jianya to Nuorilang via the shuttle bus, the Five-Color Pond, and the Shuzheng lakes. See the full post for details and examples.
- [A Two-Day Trip to Copenhagen](https://longda.us/2020-02-01/copenhagen/) (2020-02-01): A travelogue of a two-day trip to Copenhagen: buy the Copenhagen Card and enjoy the old town and its museums. Recommended spots include the Carlsberg Glyptotek, Christiansborg Palace, and the National Museum. See the body for more details and examples.
- [A Day Trip in Stockholm](https://longda.us/2020-01-30/stockholm/) (2020-01-30): A one-day trip through Stockholm on a Nordic tour: a guided visit to the City Hall and its dazzling gold-leaf Golden Hall and hand-painted Princess Hall, plus street shots of the Royal Palace, the Nobel Museum, and the old town under the short Nordic daylight.
- [Norway 3-Day Trip (3)](https://longda.us/2020-01-29/norway3/) (2020-01-29): A day-trip travelogue in Bergen: riding the funicular to overlook this World Heritage city, strolling the colorful wooden houses of the Hanseatic wharf, admiring the oil paintings at the museum, and wrapping up at a Chinese restaurant. See the full post for details and examples.
- [Norway 3-Day Trip (2)](https://longda.us/2020-01-28/norway2/) (2020-01-28): Travelogue of the second leg of the Norway in a Nutshell tour: the train from Myrdal to Flam, the fjord cruise, and the changing winter and summer scenery along the way, with on-the-road video clips. See the full post for details and examples.
- [Norway 3-Day Trip (1)](https://longda.us/2020-01-27/norway/) (2020-01-27): A day-trip travelogue in Oslo: Vigeland Sculpture Park, the university, and the royal palace, plus the Norway in a Nutshell itinerary and tips for buying the Oslo Pass. See the full post for details and examples.
- [A Trip to Helsinki](https://longda.us/2020-01-25/helsinki/) (2020-01-25): A one-day travelogue of Helsinki, Finland: the Sibelius Park organ sculpture, the Rock Church, the world's cheapest LV, and other Nordic culture and architecture impressions. See the full post for details and examples.
- [A Trip to Rovaniemi](https://longda.us/2020-01-23/rovaniemi/) (2020-01-23): A winter travel diary of the Rovaniemi aurora trip in Finland: icebreaker cruise, aurora photography tour, and forest skiing, with Nordic Travels booking tips and advice on extending or changing the itinerary. See the full text for details and examples.
- [A Self-Guided Tour Around Taiwan](https://longda.us/2016-10-14/traveltaiwan/) (2016-10-14): A guide to a self-guided tour around Taiwan: handling the travel permit endorsement and entry permit, planning a half-island route, attractions in Kenting and Hualien, and tips for booking transport and flights—more details and examples in the article.

### Database (7)
- [Disk Performance Testing](https://longda.us/2020-07-12/fileio/) (2020-07-12): Using sysbench to compare the I/O performance of Alibaba Cloud ESSD, SSD cloud disks, and local NVMe disks, covering the test script parameters and the measured differences across storage types
- [DB Performance Testing - The 3 Common Suites - A Step-by-Step Guide to Running tpcc](https://longda.us/2020-07-05/tpcc/) (2020-07-05): Performance Testing -- DB Performance Testing - The 3 Common Suites - A Step-by-Step Guide to Running tpcc
- [DB Performance Testing - The 3 Common Suites - A Step-by-Step Guide to Running sysbench](https://longda.us/2020-06-28/sysbench/) (2020-06-28): Performance Testing -- DB Performance Testing - The 3 Common Suites - A Step-by-Step Guide to Running sysbench
- [DB Performance Testing - The 3 Common Suites - A Step-by-Step Guide to Running TPCH](https://longda.us/2020-06-22/TPCH/) (2020-06-22): Performance Testing -- DB Performance Testing - The 3 Common Suites - A Step-by-Step Guide to Running TPCH
- [A Detailed Analysis of filesort](https://longda.us/2020-06-21/mysql-filesort/) (2020-06-21): Notes on the MySQL filesort sorting mechanism: comparing the original and modified algorithms, the meaning of max_sort_length, and how the sort/addon fields are organized.
- [An Introduction to Clustered Indexes](https://longda.us/2020-06-14/mysql-cluster-index/) (2020-06-14): Notes on MySQL clustered and non-clustered indexes: InnoDB's primary-key-clustered storage, the advantages of covering indexes, and a comparison with MyISAM's physical row-pointer approach. See the body for more detail and examples.
- [Spanner: Google’s Globally Distributed Database](https://longda.us/2019-11-05/spanner/) (2019-11-05): Reading notes on the Google Spanner globally distributed database paper: TrueTime, Paxos, and externally consistent transactions.

### Industry Insights (7)
- [LangChain \"Goes Off-Script\" and Builds a Database from Scratch?](https://longda.us/2026-05-18/2026-05-18-langchain-builds-database/) (2026-05-18): To solve the problem of exploding Agent trace data in LangSmith, LangChain built a distributed database, SmithDB, from scratch. This article unpacks its four design highlights and explores the trend — and the lessons — as the Agent industry shifts from "racing on models" to "racing on data foundations."
- [7 Million People Watch AI Wipe a Database, and the Culprit Writes a Bizarre Confession](https://longda.us/2026-05-09/2026-05-09-ai-deletes-database-incident/) (2026-05-09): PocketOS's AI Agent, running in Cursor + Claude Opus, unilaterally deleted the Railway production database and its backups, causing an incident in 9 seconds and then writing a "confession." This article replays the whole event, dissects flaws such as token permissions and backup isolation, and explores the security defenses a database should have in the AI era.
- [DeepMind CEO Interview: We're Just 4 Years and 3 Final Puzzle Pieces Away from AGI](https://longda.us/2026-05-09/2026-05-09-deepmind-ceo-agi-interview/) (2026-05-09): In a recent interview, Google DeepMind CEO Demis Hassabis predicts AGI could arrive by 2030 and pinpoints AI's three biggest gaps today: continual learning, long-term reasoning, and true memory. This article breaks down the core takeaways and explores how infrastructure for memory systems, on-device intelligence, and AI safety needs to evolve.
- [Talking with AI: How Do You Survive the AI Chaos?](https://longda.us/2026-04-25/2026-04-25-surviving-ai-era-conversation/) (2026-04-25): A survival guide distilled from a conversation with AI—covering the logic of big-tech layoffs, the predicament at each level from P6 to P9, a four-layer personal moat model (efficiency, cognition, relationships, asymmetry), and then identity-crisis adjustment and Plan B signals. Written for everyone feeling anxious in the AI wave.
- [Developers Cheer, Ordinary People Are Lost: After OpenClaw, Which Way for 'Usable AI'?](https://longda.us/2026-02-11/2026-02-11-usable-ai-after-openclaw/) (2026-02-11): Compiled from a panel discussion at the OceanBase Community Carnival, where guests from Eigent, Ant Bailing, Fellou, and others discuss the OpenClaw phenomenon and the emerging consensus on "usable AI" — task-oriented agents first, transparency and verifiability, context engineering and human-AI collaboration, and the changing barrier to using AI.
- [Skills Arrive — Is Prompt Dead? In 2026, How Do We Build Controllable Thinking for Agents?](https://longda.us/2026-02-07/2026-02-07-skills-vs-prompt-agent-thinking/) (2026-02-07): A panel transcript from the OceanBase Community Carnival: experts from RAGFlow, FastGPT, Nowledge Labs, and OceanBase discuss where the RAG ecosystem is heading in 2026, the boundary between Memory and RAG, Skills development practice, and the integrated data foundation behind an agent's \"controllable thinking.\"
- [Data-Driven Innovation and Transformation in the Pan-Internet Industry, and New Opportunities for Databases](https://longda.us/2025-06-16/2025-06-16-pan-internet-innovation-database-opportunities/) (2025-06-16): An exploration of how data-driven change is transforming the pan-internet industry and the new opportunities this creates for databases, covering technical directions such as HTAP, cloud deployment, AIOps, and security compliance.

### News (7)
- [Shanghai Meetup Highlights: Xinye's Cost Cuts, Suanzhi Future's Hybrid Search, and AgentSeek's One-Stop Agent Engineering](https://longda.us/2026-06-03/2026-06-03-shanghai-meetup-recap/) (2026-06-03): A recap of the Shanghai Agent Infra technical salon co-hosted by OceanBase and the LangChain Community: the debut of the AgentSeek enterprise-grade agent engineering platform, Xinye Technology's full upgrade of its core database to OceanBase cutting costs by 70%, and Suanzhi Future governing 20TB of training corpus with hybrid search.
- [OceanBase Community Monthly: Major Updates to PowerMem, OMS, and obdiag](https://longda.us/2026-05-21/2026-05-21-oceanbase-community-monthly-2026-05/) (2026-05-21): OceanBase Community Monthly for May 2026: PowerMem v1.0.0 integrates pyseekdb for embedded vector storage, OMS V4.2.13-CE adds ES, MongoDB, and ClickHouse data sources, obdiag 4.3.0 launches an MCP-based intelligent diagnosis Agent, plus kernel defect fixes and community updates.
- [Shanghai, Here We Come! On 5/30, OceanBase × LangChain Join Forces to Debut \"AgentSeek\" and Define a New Paradigm for Enterprise Agent Development](https://longda.us/2026-05-13/2026-05-13-shanghai-oceanbase-langchain-meetup-preview/) (2026-05-13): On May 30, OceanBase teams up with the LangChain Community for an offline Meetup in Zhangjiang, Shanghai, where it will fully unveil AgentSeek, an enterprise-grade agent engineering solution, along with hands-on enterprise Agent deployment case studies from PPDai, Suanzhi Future, and more.
- [OceanBase Teams Up with Baidu ERNIE & PaddlePaddle, Qoder, and Multiple Hardware Vendors to Explore Production-Grade Agent Deployment](https://longda.us/2026-04-25/2026-04-25-oceanbase-paddle-qoder-agent-production/) (2026-04-25): On April 25, OceanBase joined forces with Baidu ERNIE & PaddlePaddle, Qoder, and several smart-hardware vendors to host the "Storage and Compute Evolution in the Agent Era" Meetup in Shenzhen, sharing Silicon Valley AI trends, PowerMem intelligent memory, PaddleOCR multimodal parsing, OpenClaw software-hardware co-design, and other production-grade hands-on practices.
- [Live from the Community Carnival! Open Source, Open Ecosystem, Shared Success — a 13-Year-Old Takes Second Place in AI Coding!](https://longda.us/2026-02-06/2026-02-06-community-carnival-2026/) (2026-02-06): On January 31, the 2026 OceanBase Community Carnival was held in Shanghai, gathering over 260 developers. Through keynotes, panel discussions, AI Coding challenges, and a community open mic, attendees discussed building the AI data foundation — and a 13-year-old took second place in the AI Coding "Hardest PR Award."
- [Born for AI: Efficiently Build Your Agent and AI System Architecture with OceanBase seekdb](https://longda.us/2025-12-03/2025-12-03-seekdb-agent-ai-architecture/) (2025-12-03): After open sourcing its AI-native database seekdb, OceanBase launched a writing contest with prizes, inviting developers to deploy and experience seekdb, or to build AI applications, agents, and system architectures based on seekdb, PowerMem, and PowerRAG, share their hands-on experiences, and win prizes.
- [A Recap of the Hands-on AI Workshop at the OceanBase Annual Conference](https://longda.us/2025-11-28/2025-11-28-oceanbase-handson-ai-workshop/) (2025-11-28): A recap of the Hands-on AI Workshop at the OceanBase Annual Conference: hundreds of developers, guided by experts from LangChain, Dify, and OceanBase, completed the full development flow of environment setup, Agentic RAG building, and constructing long-term-memory agents in two hours of hands-on practice based on seekdb.

### OceanBase (7)
- [Integrating OceanBase Monitoring with Prometheus/Grafana](https://longda.us/2021-11-21/obagent/) (2021-11-21): OceanBase Getting Started -- Integrating OceanBase Monitoring with Prometheus/Grafana
- [OceanBase Offline Installation](https://longda.us/2021-11-20/ob_offline_install/) (2021-11-20): A guide to the offline installation of an OceanBase distributed three-replica cluster, covering passwordless SSH, disk planning, deploying OBServer/OBProxy with OBD, and tenant verification end to end.
- [OceanBase Developer Handbook, Part 5: How to Debug OceanBase](https://longda.us/2021-11-11/debug_ob/) (2021-11-11): OceanBase Developer Handbook, Part 5: How to Debug OceanBase
- [OceanBase Developer Handbook, Part 4: How to Modify OceanBase Documentation](https://longda.us/2021-11-10/modify_ob_docs/) (2021-11-10): OceanBase Developer Handbook, Part 4: How to Modify OceanBase Documentation
- [OceanBase Developer Handbook, Part 3: How to Become an OceanBase Contributor](https://longda.us/2021-10-30/contribute_to_ob/) (2021-10-30): OceanBase Developer Handbook, Part 3: How to Become an OceanBase Contributor
- [OceanBase Developer Handbook, Part 2: How to Set Up the IDE Development Environment](https://longda.us/2021-10-23/set_ide/) (2021-10-23): OceanBase Developer Handbook, Part 2: How to Set Up the IDE Development Environment
- [OceanBase Developer Handbook, Part 1: How to Compile the OceanBase Source Code](https://longda.us/2021-10-16/build_ob/) (2021-10-16): OceanBase Developer Handbook, Part 1: How to Compile the OceanBase Source Code

### Life (5)
- [My Tesla Suddenly Lost Power While Driving](https://longda.us/2022-02-26/tesla-is-bullshit/) (2022-02-26): How my Tesla suddenly lost power while driving: an accidental touch of the gear paddle disabled the front motor. A record of the fault, the remote data upload, and Tesla's absurd official repair response—more details and examples in the article.
- [My 10-Year Anniversary](https://longda.us/2020-06-07/10years/) (2020-06-07): Reflecting on a decade at Alibaba: from promising to stay five years to nearly ten. Gratitude for the platform's inclusiveness, the team's trust, and professional growth.
- [New Year Wish](https://longda.us/2020-01-01/New-Year-Wish/) (2020-01-01): Reflections and a wish list for the new year 2020: looking back at last year's unfinished goals, then setting goals for the year ahead such as blogging, reading, working out, and exercising with family. See the full post for details and examples.
- [A Short Talk on Job-Hopping](https://longda.us/2016-09-23/jump/) (2016-09-23): A short talk on job-hopping: analyzing the essence and trade-offs of changing jobs across dimensions such as salary, future career and company prospects, and family changes, drawing on two classic reasons for discussion. See the body for more detail and examples.
- [Complaining About GoDaddy](https://longda.us/2016-08-10/godaddy/) (2016-08-10): A rant about GoDaddy deleting my personal hosting and wiping out five years of blog posts: a strong condemnation of their lack of accountability, plus a suggestion to build your own blog on Alibaba Cloud instead. See the full post for details and examples.

### Reading Notes (5)
- [HashCorp Reading Notes](https://longda.us/2022-02-28/hashcorp-read-notes/) (2022-02-28): Reflections on a 10,000-word article about HashiCorp's open-source commercialization: thoughts on open source vs. enterprise tiers, channel partnerships, S&M cost, and the methodology behind tooling, among other business and product takeaways. See the full post for details and examples.
- [The Rise of Huawei](https://longda.us/2020-04-30/huawei/) (2020-04-30): Reading notes on The Rise of Huawei: tracing the key decisions across Huawei's founding, growth, and survival phases, and reflecting on customer-first thinking, blue-ocean choices, and organizational management. See the full post for details and examples.
- [Reflections on "Peak"](https://longda.us/2018-12-13/keyilianxi/) (2018-12-13): Reading notes on 'Peak': refuting the idea of innate genius and the simplistic ten-thousand-hour rule, summarizing a training method built on four elements — goals, focus, feedback, and stepping out of the comfort zone. See the body for more detail and examples.
- [Reflections on "Steve Jobs"](https://longda.us/2018-10-08/qiaobusi/) (2018-10-08): Reflections on the biography 'Steve Jobs': exploring obsession versus collective wisdom, the techniques of a world-class marketing master, and how different readers come away with their own personal takeaways. See the full text for details and examples.
- [Reflections on the Cognitive Revolution](https://longda.us/2017-12-23/knowledgerevolution/) (2017-12-23): Notes on Li Shanyou's 'Cognitive Revolution' course: a walkthrough of core theories such as first principles, inductive and deductive thinking, discontinuity, and scientific revolution, along with personal interpretations. See the body for more detail and examples.

### Scala (2)
- [Getting Started with Scala](https://longda.us/2016-08-24/scala-start/) (2016-08-24): Notes on getting started with Scala: the interpreter, val/var variables, functions and the Unit type, the foreach syntactic sugar, arrays, side effects, and other fundamental syntax points.
- [Scala Overview](https://longda.us/2016-08-23/scala-general/) (2016-08-23): Overview of the Scala language: a statically typed JVM language that blends object-oriented and functional programming, introducing the Actor concurrency model along with its intellectual origins and design characteristics. See the full text for details and examples.

### Database Operations (1)
- [Scaling Out ZooKeeper](https://longda.us/2015-09-10/zookeeper-enlarge/) (2015-09-10): A hands-on guide to scaling ZooKeeper across data centers in the same city: expanding a single-DC three-node cluster into a 2-2-1 layout, adding nodes step by step, retiring the old ones, and controlling the risk of leader changes.

### User Practices (1)
- [Migrating from HBase to OceanBase: The Ultimate Solution for Real-Time Data Writes with Flink](https://longda.us/2026-04-24/2026-04-24-hbase-to-oceanbase-flink-migration/) (2026-04-24): This article introduces a real-time data-write solution based on Flink and OceanBase OBKV-HBase—fully compatible with the HBase API, supporting buffered batch writes and unified stream-batch processing. It provides examples of table creation, Flink SQL mapping, and real-time Kafka writes, along with performance tuning and troubleshooting practices.