# 📊 HiddenMerit Daily · Issue 51
> Focus on Database Frontiers, Practical Insights for DBAs
> June 26, 2026 ｜ 5 Selected Global Breaking News
## 01｜Alibaba Cloud Tops China Relational Database Market for Seventh Consecutive Year with 26.1% Market Share
On June 24, IDC released the “China Relational Database Software Market Tracker, H2 2025.” The report shows that Alibaba Cloud ranked first with 26.1% full‑year overall market share, achieving a seven‑consecutive‑year championship since 2019. China’s relational database software market reached $4.83 billion in 2025, up 14.5% year‑on‑year.
Public Cloud Market: Alibaba Cloud’s full‑year 2025 market share reached 37.6%, further increasing to 38.1% in the second half, with its leading advantage continuing to expand.
On‑Premises Market: Alibaba Cloud’s revenue grew 25.4% year‑on‑year, far exceeding the industry average of 11.1%, with continued implementation cases at the four major state‑owned banks and central SOE headquarters in government and finance sectors.
Market Forecast: IDC projects China’s relational database software market will reach $11.24 billion by 2030, with a 2025‑2030 CAGR of 18.4% and a domestic production rate exceeding 70%. AI integration has become the most important product and technology trend in relational databases – “AI for DB” improves autonomous operations, while “DB for AI” supports multi‑modal data fusion processing.
- DBA Perspective: Alibaba Cloud’s seven consecutive championships confirm the absolute dominance of cloud databases in the overall market (the public cloud market is approximately twice the size of the on‑premises market). Alibaba Cloud’s 25.4% growth in the on‑premises market far exceeding the industry average means hybrid cloud and multi‑cloud deployments are becoming the preferred path for government and enterprise customers. When planning career development, DBAs should further increase the weight of cloud‑native database skills (PolarDB, AnalyticDB, etc.). IDC’s forecast of an $11.24 billion market by 2030 with over 70% domestic production rate provides long‑term confidence for DBAs to deepen their domestic database technology stack.
- CTO Perspective: Alibaba Cloud’s seven consecutive championships and continued market share expansion reflect that the combined advantages of cloud databases in scale, elasticity, and AI integration capabilities are creating a “winner‑takes‑all” effect. IDC’s forecast of continued concentration among leading vendors means the “cost of correcting a wrong technology choice” will keep rising. When planning data architectures, CTOs should prioritise database products deeply integrated with leading cloud vendor ecosystems to reduce long‑term technical debt.
- Investor Perspective: Alibaba Cloud’s 37.6% share in the public cloud market (several times that of the second‑placed vendor) forms one of the foundational pillars of Alibaba Cloud’s revenue. IDC’s forecast of an $11.24 billion market by 2030 with an 18.4% CAGR provides quantifiable growth space for the database track. AI integration has become a confirmed industry trend – database vendors with both “cloud‑native” and “AI‑native” capabilities will win in this round of competition.
## 02｜Dameng Nanjing Channel Ecosystem Conference: Four Full‑Stack Self‑Developed New Products Deepen Regional Market Penetration
On June 25, the 2026 Dameng Channel Ecosystem Conference, themed “Dreaming a New Chapter, Gathering in Jinling,” was successfully held in Nanjing. Dameng showcased four full‑stack self‑developed new products and engaged in in‑depth discussions with core partners in the Jiangsu region, focusing on product innovation, industry applications, and coordinated regional market development.
Core Highlights of the Four New Products:
| Product | Core Capabilities | Key Data |
|---------|-------------------|----------|
| DM9 Database | Centralised and distributed integrated architecture, supporting multi‑tenancy, autonomous disaster recovery, and AI agents | Over 450 functional upgrades |
| New Generation Database Appliance | Ultra‑fast storage system with four major technological innovations | IOPS starting at 12 million; I/O latency reduced from 400μs to 80μs |
| Qiyun Database V4.0 | AI operations system covering intelligent tuning, fault self‑healing, cross‑cloud and cross‑region deployment | Focuses on three pain points: “low O&M efficiency, complex disaster recovery, fragmented compute power” |
| Graph Database V4.0 | Graph‑native and vector‑integrated architecture, HyperRAG full‑domain retrieval | Supports over 100 billion‑level graph data; AI construction cycle shortened by 60%; total cost reduced by 40% |
Channel Ecosystem Building: At the conference, Dameng explained its 2026 channel policies, providing full‑spectrum support to partners in technology certification, project collaboration, and market promotion. Eight core partners received the “Dameng Regional Channel Contribution Award,” including representatives from China National Building Materials Information, Digital China, and Hygon Information, who shared their cooperation experiences with Dameng in Xinchuang ecosystem development. Dameng Vice President Shuai Lili stated in his address that Jiangsu and the East China region, as frontier hubs for China’s digital economy development, will create broader practice space for domestic foundational software.
- DBA Perspective: Dameng’s Nanjing channel ecosystem conference signals a clear regional deepening strategy – Jiangsu and East China, as digital economy hubs, will see further release of job demand for domestic databases. The new generation database appliance’s I/O latency reduction from 400μs to 80μs provides DBAs with a quantitative reference for hardware selection in high‑performance scenarios. The graph database V4.0’s support for over 100 billion‑level graph data and 60% AI construction cycle reduction means DBAs can complete knowledge graph + vector retrieval AI scenario setup within a single Dameng database, eliminating data movement between multiple systems.
- CTO Perspective: Dameng’s four new products cover the full spectrum from general‑purpose relational databases to cloud‑native, software‑hardware integrated, and graph computing, forming a complete enterprise‑grade technology matrix. The deep participation of channel ecosystem partners (China National Building Materials Information, Digital China, etc.) indicates that the delivery system for domestic databases is evolving from “direct vendor sales” to a mature “channel + service” model. When making Xinchuang selections, CTOs can use this to assess vendors’ regional service capabilities and ecosystem support.
- Investor Perspective: Dameng’s regional deepening strategy at the channel ecosystem conference, combined with its financial performance (2025 revenue RMB 1.306 billion, core product gross margin 99.73%), forms a “product + channel” dual‑drive. The graph database V4.0’s “AI construction cycle shortened by 60%” data is an important reference for assessing Dameng’s product competitiveness in the AI direction. The progress of channel expansion in Jiangsu and East China can serve as an observation indicator for Dameng’s regional market penetration capability.
## 03｜AkasicDB: Korean Team Develops Tri‑Modal Converged Database, 78% RAG Accuracy Improvement, 20x Response Speed Increase
Recently, scientists from KAIST and startup GraphAI jointly developed next‑generation database technology AkasicDB, creatively integrating vector database, graph database, and relational database capabilities into a single system, capable of simultaneously understanding deep relationships between documents, data, and entities. The related paper was published at the top database conference in the “Proceedings of the International Conference on Data Engineering” (ICDE).
Technical Breakthroughs:
- Omni RAG Solution: Unifies vector similarity search, graph traversal, and relational filtering into a single query and execution plan, simultaneously mining semantic information from documents, relationship information from knowledge graphs, and structured conditions from tables.
- Unified Storage Architecture: AkasicDB integrates the storage of three data models into a single database management system, optimising and executing queries across three data models as a single execution plan through a unified query planner and “traverse‑join‑similarity” operators, minimising unnecessary intermediate results and data movement.
Performance Data:
- Response accuracy improved by up to 78% compared to traditional RAG
- Complex search query response time reduced from 21.3 seconds to under 1 second, a performance leap of over 20x
AkasicDB is positioned as next‑generation database technology for the AI agent era, with potential for widespread application as core data infrastructure in fields with extremely high reliability requirements such as defence, manufacturing, finance, law, and technology.
- DBA Perspective: AkasicDB represents the “ultimate form” of database multi‑modal convergence – vector + graph + relational tri‑modal integration, rather than simple plug‑ins. Its Omni RAG solution unifies three retrieval capabilities into a single execution plan, eliminating intermediate results and data movement overhead for cross‑data‑model queries. For DBAs, this means multi‑modal databases are no longer “multiple databases patched together” complex architectures, but a true “single engine serving all queries.” AkasicDB’s 20x performance improvement and 78% accuracy improvement provide DBAs with a quantitative reference for evaluating the AI scenario applicability of multi‑modal converged databases.
- CTO Perspective: The joint achievement of KAIST and GraphAI shows that the academic frontier is accelerating vector + graph + relational tri‑modal convergence. The Omni RAG solution breaks through traditional RAG’s inability to simultaneously handle document semantics, entity relationships, and structured conditions, directly addressing the core pain point of “hallucination” in enterprise AI implementation. When planning AI data architectures, CTOs should make “multi‑modal convergence capability” a core consideration in database selection, not just focusing on single‑vector retrieval performance.
- Investor Perspective: AkasicDB’s achievement marks that academic validation of the evolution from “multiple databases patched together” to “one database for multiple models” has gained recognition from top conferences. The vector + graph + relational tri‑modal convergence technology path may become the standard paradigm for next‑generation AI databases. Database startups and academic teams with “tri‑modal convergence” technology储备 will attract capital market attention in this round of AI database competition.
## 04｜AWS Defines Five Key Capabilities for Databases in the Agentic AI Era: Open Architecture, Zero Ops, Extreme Elasticity
During the 2026 AWS China Summit, Ganapathy “G2” Krishnamoorthy, Vice President of Database Services at Amazon Web Services, engaged in an in‑depth dialogue with the media, clearly defining the five key capabilities for databases in the Agentic AI era.
The Five Key Capabilities:
| Capability | Core Meaning |
|------------|--------------|
| Open Data Access | Supports the MCP protocol, enabling databases such as PostgreSQL, MySQL, and DynamoDB to be smoothly accessed by Agentic frameworks |
| Vector Embedding and Hybrid Search | Empowers applications to store and retrieve Agentic memory; multiple database services have already integrated vector retrieval |
| Model Invocation Capability | Whether the database environment itself can directly invoke AI models for inference |
| True Zero Ops | Out‑of‑the‑box, no manual instance specification evaluation, automatic scaling, self‑optimisation; traditional databases with built‑in AI agents take over O&M |
| Extreme Elastic Scaling | Scaling to tens of millions of concurrent requests during bursts, scaling to zero during idle periods |
Selection Path Recommendations: G2 clearly stated that database construction in the Agentic AI era is not “one‑size‑fits‑all” – new projects should decisively adopt modern open‑source foundations (MySQL/PostgreSQL + Iceberg‑format data lakes); existing applications should prioritise releasing data value rather than overthrowing and rebuilding.
- DBA Perspective: AWS’s “Zero Ops” capability has profound implications for the DBA role – developers of AI‑generated applications often lack traditional DBA skills; they need an automated database foundation. This means the DBA role will evolve from “manual operations” to “O&M policy definer.” MCP protocol support becoming the “standard interface” for agents to access databases requires DBAs to learn MCP protocol security configuration and permission models. Scale‑to‑zero extreme elasticity requires DBAs to redesign monitoring and alerting systems – the traditional capacity planning logic based on “fixed resource pools” will be overturned.
- CTO Perspective: AWS’s pragmatic path recommendation – “open‑source foundation for new projects, prioritise releasing data value for existing applications” – provides CTOs with a clear decision‑making framework for database selection in the AI era. Open‑source foundations (MySQL/PostgreSQL + Iceberg) offer enterprises extremely high future flexibility and overwhelming cost advantages.
- Investor Perspective: AWS’s support for the MCP protocol marks that MCP is becoming the “de facto standard” for agents to access databases. Data access layers, security controls, and observability tools around the MCP protocol will see market demand. AWS’s “Zero Ops” direction will further increase the penetration rate of managed database services.
## 05｜Weekly Security Vulnerabilities Focus: OpenLink Virtuoso DoS, PostgreSQL MD5 Side‑Channel Patches Continue
OpenLink Virtuoso Denial‑of‑Service Vulnerability (CVE-2025-61024) : Affects openlink virtuoso‑opensource version 7.2.11. The vulnerability originates from the sqlo_try_in_loop component, allowing attackers to cause denial of service through crafted SQL statements.
PostgreSQL MD5 Timing Side‑Channel Patch Continues: Red Hat Enterprise Linux 8 has released security updates for postgresql:12 and postgresql:13 modules, fixing CVE-2026-6478 (MD5 password hash comparison timing side‑channel) and CVE-2026-6473 (integer overflow).
- DBA Perspective: The OpenLink Virtuoso DoS vulnerability reminds DBAs to pay attention to the security of “niche but critical” database components. Virtuoso still has widespread deployments in knowledge graph and enterprise data integration scenarios. It is recommended to check whether affected versions (7.2.11) are in use and evaluate upgrade priorities. The continued patching of the PostgreSQL MD5 timing side‑channel vulnerability (covering PG 12 and PG 13 modules) reminds DBAs to migrate authentication from MD5 to SCRAM-SHA-256 as soon as possible. MD5 password hash comparison has a timing side‑channel risk – attackers can recover credentials by measuring authentication time.
- CTO Perspective: Virtuoso, as a representative open‑source multi‑modal database, its vulnerability patch cadence reflects the open‑source community’s maintenance commitment for older versions. When selecting technologies, CTOs should incorporate vendors’/communities’ security response speed into their evaluation systems.
- Investor Perspective: The continued exposure of vulnerabilities in multi‑modal databases and open‑source database components creates sustained market demand for database security scanning and compliance auditing service providers.
## 📚 SQL Little Knowledge Point
This Issue’s Knowledge Point: What is “Zero Ops”?
“Zero Ops” is a new database capability standard proposed by AWS for the Agentic AI era – referring to database services that are out‑of‑the‑box, requiring no manual instance specification evaluation or daily operations, with automatic scaling and self‑optimisation. In the AI era, developers using AI‑generated applications often lack traditional DBA skills; what they truly need is an automated database foundation.
Traditional Ops vs. Zero Ops:
| Dimension | Traditional Ops | Zero Ops |
|-----------|-----------------|----------|
| Specification Selection | DBAs select instance specifications based on business estimates | System automatically adapts |
| Capacity Planning | Resources reserved for peaks, leading to waste | Automatic scaling, Scale to zero |
| Performance Tuning | DBAs manually analyse slow SQL, adjust parameters | AI Agent automatically optimises |
| Fault Handling | DBAs manually troubleshoot and recover | System self‑heals |
| Skill Requirements | Requires professional DBA | Usable by developers |
Implementation Path: Traditional databases (such as PostgreSQL) can have all O&M work taken over by built‑in AI agents, including automatic diagnosis, automatic tuning, and automatic scaling.
Significance for DBAs: Zero Ops does not mean DBAs will lose their jobs, but rather that the focus of work will shift from “repetitive O&M” to “policy definition” – defining boundary conditions for automatic scaling, auditing AI agent operation traces, and handling complex anomalies that AI cannot cover.
> HiddenMerit Team Production
> Slogan: 绩优隐于内，金石启新程 | Hidden deep. Merit bold. Forge ahead.
No comments yet