HiddenMerit Daily · Issue 15

📊 HiddenMerit Daily · Issue 15

Focus on Database Frontiers, Practical Insights for DBAs
May 10, 2026 ｜ 5 Selected Global Breaking News

01｜Community Controversy Continues: MySQL 9.7 LTS "True Vector" Capabilities Shrunk, Some Are Voting with Their Feet

Although MySQL 9.7 LTS moved several previously enterprise‑only features (vector functions, Hypergraph Optimizer, JSON Duality Views) to the Community Edition at the end of April, the open‑source "sweet spot" has fallen far short of community expectations. The reality: in Chapter 14.21 of the 9.7 Reference Manual, the entire vector section is missing core support such as vector indexes and nearest‑neighbour search. Vector columns cannot be used as primary keys, foreign keys, or unique keys, nor do they support any aggregate functions. The only function usable for vector retrieval, DISTANCE(), is marked as an enterprise‑only paid feature in the manual. This means Community Edition users have the VECTOR_TYPE concept but cannot efficiently perform vector search. Some developers even accuse Oracle of lacking strategic sincerity in opening up the community with 9.7, and community activity and user satisfaction have fallen to a 24‑month low. The strained relationship between the MySQL community and Oracle may accelerate developer migration to forks like PostgreSQL and MariaDB.

· DBA Perspective: The "incomplete" vector implementation in MySQL 9.7 LTS is a clear trap for DBAs building RAG applications on MySQL. Before deploying hybrid queries on 9.7 LTS, be sure to measure actual recall performance. Evaluate the engineering risk of alternatives using PostgreSQL's pgvector extension to avoid extremely slow distance searches due to missing vector index support, which could force a costly second‑round data layer refactoring.
· CTO Perspective: Oracle making DISTANCE() a paid feature is a very dangerous signal – it means building a truly usable AI database service still requires purchasing licenses and taking on additional technical debt. Technical decision‑makers should elevate vector capability completeness to a mandatory qualification item in database selection. If 9.7 LTS's actual vector performance falls short, quickly add a smooth migration path to open‑source alternatives in the technology roadmap.
· Investor Perspective: Oracle's move to push vector capabilities to Community Edition while keeping the key function as a paid enterprise feature may damage long‑term community trust and push AI‑needy SMEs toward the PostgreSQL ecosystem. Investors should monitor funding and user growth metrics of startups around the PostgreSQL ecosystem (e.g., managed pgvector service providers) as leading signals of open‑source database market share shifts.

02｜OceanBase Accelerates AI Ecosystem Integration, GitHub Ultimate C++ Open‑Source Score Ranks Among Global Top

OceanBase recently accelerated its AI ecosystem efforts, announcing its first batch of over 60 AI ecosystem partners, including major industry frameworks such as LlamaIndex, LangChain, Dify, FastGPT, Camel AI, and OpenManus. It has also deeply integrated with MCP (Multi‑Agent Control Protocol), providing AI developers with full‑stack ecosystem support from models to databases. Leveraging its distributed database's HTAP capabilities, vector search, and low‑latency transaction engine, OceanBase is positioning itself as one of the most common "data foundations" for AI application development. Additionally, OceanBase's Ultimate C++ open‑source score on GitHub rose to the global top tier for the first time in May 2026, enhancing domestic enterprises' global influence in open‑source core competitiveness. Community contributor activity and code quality have been highly recognised, and its enterprise‑grade open‑source governance is attracting more overseas financial‑grade customers.

· DBA Perspective: OceanBase's deep integration with over 60 mainstream AI frameworks means DBAs will get involved earlier in AI project development – from configuring vector indexes and building batch‑stream pipelines to performance monitoring. DBAs should pay attention to OceanBase's AI ecosystem partner tooling, especially changes in data access patterns when interfacing with agent‑class applications via the MCP protocol.
· CTO Perspective: OceanBase's "data × AI" ecosystem partner model orchestrates Model‑as‑a‑Service and DB‑as‑a‑Service together, technically shortening the data‑layer timeline from AI application development to deployment. Technical managers with AI projects can prioritise OceanBase's integrated AI solution for development efficiency gains.
· Investor Perspective: OceanBase's GitHub open‑source score reaching the global top tier for the first time, combined with its rapidly expanding AI agent ecosystem, will enhance its technical evaluation and procurement premium ability among overseas financial and tech innovation companies, ultimately reflected in enterprise‑class order growth. Open‑source standing and the degree of integration with international AI frameworks can serve as important reference indicators for market share changes.

03｜NineYou Database Completes Series A+ Funding, Targeting AI Hyperconvergence and Multi‑Modal Technology Opportunities

Recently, domestic hyperconverged database startup NineYou Database announced the successful completion of its Series A+ funding round, attracting investments from Songhe Capital, Qinghao Capital, and CMBC Dinghong Investment. This marks an important exploration to support the rapid evolution of modern AI applications and the challenge of heterogeneous data governance. Founded in 2021, NineYou Database's core product is a self‑developed multi‑modal hyperconverged database covering key areas such as vector search, time‑series analysis, and federated queries over heterogeneous data sources. Its goal is to provide a one‑stop data technology infrastructure solution for AI applications.

· DBA Perspective: The renewed funding for domestic hyperconverged database startups sends a clear market signal: AI applications are creating more urgent demand for integrating traditional SQL processing capabilities with next‑generation vector, time‑series, and heterogeneous data federation. The DBA role will evolve into a bridge between AI application frameworks and multi‑modal data construction, requiring early learning of design thinking for combining multiple data models.
· CTO Perspective: Market recognition through the Series A+ round reflects that AI‑native demands are forcing structural upgrades in next‑generation databases – single‑modal databases will struggle to fully meet scenarios requiring multi‑modal data and complex query fusion. Technical managers should focus on internal and external integration testing of multi‑modal databases, proactively creating POC environments for different business lines to validate.
· Investor Perspective: NineYou Database's continued funding indicates that the domestic hyperconverged database track is gaining strong capital interest. AI requires the underlying data infrastructure to seamlessly accommodate multiple data models – the "hyperconverged landscape" is capturing more market expectations. Investors should pay attention to other database startups with self‑developed breakthroughs in vectorisation and multi‑modal technologies.

04｜AWS ElastiCache Fully Upgraded: Vector + Full‑Text Hybrid Search Enhances AI Agent Memory Foundation

On May 6, AWS announced that Amazon ElastiCache now supports real‑time hybrid search, combining vector similarity search with full‑text search in a single query, eliminating the need for additional separate search components. ElastiCache for Valkey delivers the lowest retrieval latency and highest throughput at a 95% recall rate, with excellent price‑performance.

· DBA Perspective: Previously, when building RAG applications on AWS, DBAs often had to juggle complex component stacks among ElastiCache (caching), OpenSearch (full‑text search), and pgvector (vector storage). Now ElastiCache handles hybrid search (scalar + full‑text + vector) in a single cluster, dramatically simplifying the caching and real‑time retrieval architecture for AI applications. AWS‑experienced DBAs should validate the actual latency improvement of ElastiCache for long‑context agent memory calls in their next architectural design.
· CTO Perspective: AWS's unified in‑memory database layer hybrid search capability significantly reduces the number of operational components and complex dependencies for AI applications. It is especially suitable for agentic applications and e‑commerce recommendation systems with extreme requirements for real‑time performance, concurrency, and global distribution. Technical decision‑makers need to examine ElastiCache's cache hit rate and stability under multi‑modal retrieval at millions of QPS.
· Investor Perspective: AWS once again demonstrates its platform engineering ability to "avoid moving data" in the agent era. Third‑party specialised vector database vendors will face competitive pressure from hyperscale cloud providers who are continuously expanding their database feature sets. Investors should assess whether those vendors' differentiation remains solid.

05｜Tencent Cloud's "AI Integration Strategy" to Be Unveiled at May 29 Product Launch

Tencent Cloud Database previously announced a major "Database + AI" product launch on May 29, 2026, with the core theme of moving beyond simple tool‑combination thinking to achieve deep internalisation and native integration of large models with the database kernel (AI‑In‑Database). Tencent Cloud will present, for the first time, its full strategic landscape under a dual‑track route, covering DB for AI (optimising the database for AI workloads) and AI in DB (embedding AI capabilities into the database kernel).

· DBA Perspective: AI‑In‑Database is being systematically integrated for the first time by a leading domestic cloud vendor. Tencent Cloud's specific engineering implementation is worth studying in depth. Since AI agent access patterns to databases are completely different from traditional OLTP, DBAs should early sort out performance baselines, resource budgets, and monitoring systems for hybrid scenarios (multi‑dimensional queries, vector similarity search, scalar filtering) to leave a migration window for post‑launch adoption.
· CTO Perspective: With Tencent Cloud joining the "AI‑In‑Database" array, all leading domestic cloud database vendors will now have kernel‑level AI‑native integration capabilities. If the engineering efficiency and kernel performance demonstrated at Tencent Cloud's launch are truly delivered, enterprises will make database AI maturity one of the most important factors in their selection process.
· Investor Perspective: Tencent Cloud Database's shift from "cost optimisation" to AI‑focused kernel innovation indirectly signals that after achieving profitability, Tencent Cloud will focus on enhancing its technology brand premium around AI. The technical substance and roadmap presented at the May 29 launch will affect Tencent's competitive valuation in the IaaS+PaaS market and represent a new milestone in catching up with the industry's top tier.

📅 Recent Database Hot Topics Recap

Date Event Core Highlights
May 6 AWS ElastiCache supports vector + full‑text hybrid search One cluster handles scalar, full‑text, and vector retrieval
May 7 MongoDB 8.3 GA + Atlas Automated Voyage Embeddings Built for AI speed; automatic vectorisation on write
May 8 Oracle 26ai LTS officially released 300+ new features; AI + developer experience upgrades
May 9 ScyllaDB Cloud GA announced Real‑time NoSQL + vector search integration enters production
May 9 NineYou Database completes Series A+ funding Domestic hyperconverged database track gains capital boost
Mid‑May OceanBase GitHub open‑source score ranks among global top Milestone for domestic database open‑source ecosystem
May 12 MySQL 9.7 LTS official webinar Deep dive into features of the latest LTS release
May 20 Snowflake Q1 FY2027 earnings Key validation point for AI product revenue growth
May 29 Tencent Cloud "Database + AI" product launch Domestic AI‑In‑Database roadmap officially unveiled

📌 Issue Summary

News Core Keywords DBA Actions CTO/Decision‑Maker Focus Investor Perspective
MySQL 9.7 LTS community controversy Vector shrunk, DISTANCE() paywalled, trust erosion Measure actual vector retrieval recall for RAG; assess PostgreSQL pgvector alternative risk Make vector completeness a mandatory database selection criterion Community split accelerates SME migration to PostgreSQL; monitor managed DB service provider growth
OceanBase accelerates AI ecosystem 60+ frameworks, MCP protocol, GitHub score global top Learn OceanBase AI partner tooling; study Agent access pattern changes under MCP Orchestrate Model‑as‑a‑Service + DB‑as‑a‑Service to shorten time to innovation Open‑source standing and AI framework integration boost overseas order acquisition
NineYou Database Series A+ Hyperconverged AI database, multi‑modal, heterogeneous federation Master multi‑modal design thinking (vector, time‑series, federated queries) Create POC environments for non‑standard multi‑modal business lines Multi‑modal data management track heats up; watch startups with self‑developed vectorisation tech
AWS ElastiCache hybrid search Vector + full‑text, one cluster three retrieval types Validate ElastiCache latency improvement for long‑context agent memory calls Examine cache stability under million‑QPS multi‑modal retrieval Hyperscale cloud vendors squeeze specialised vector database vendors' value
Tencent Cloud DB+AI launch AI‑In‑Database, May 29, dual‑track roadmap Prepare monitoring for hybrid scenarios; leave migration window after launch Elevate database AI maturity to a top‑tier selection factor Launch roadmap affects Tencent Cloud's valuation in IaaS+PaaS market

HiddenMerit Team Production
Slogan: 绩优隐于内，金石启新程 | Hidden deep. Merit bold. Forge ahead.