Here is the corrected fully English version of “HiddenMerit Daily · Issue 6” with the remaining Chinese phrases properly translated and a full review ensuring no other Chinese characters remain.

📊 HiddenMerit Daily · Issue 6

Focus on Database Frontiers, Practical Insights for DBAs
May 2, 2026 ｜ 5 Selected Global Hot Topics

01｜IBM Db2 12.1.5 Officially Announced: Vector Index + Native AI Model Integration, Aiming for Enterprise-Grade AI Data Platform

On May 1, IBM officially announced that Db2 12.1.5 will be generally available on June 9, 2026. This release builds on the AI foundation of Db2 12.1, introducing several major capabilities with the goal of making Db2 a unified platform for enterprise-grade AI data.

Core Technical Highlights:

Five Enhancement Areas in Detail:

1. Vector Search and AI Integration: Vector index powered by DiskANN algorithm enables similarity search directly within the database. Developers can use SQL functions like TO_EMBEDDING and TEXT_GENERATION to directly invoke external AI models for embedding generation and text generation, without moving data out of the database.

2. High Availability and Disaster Recovery: Supports configurations with more than 3 standby databases across geographic regions, standbys dedicated to reporting workloads, and cascading HADR for global deployment. Pacemaker cluster management adds arbitration disk support as a replacement for Qdevice host solutions, simplifying pure Scale deployments.

3. Security and Compliance: GSKit 9 raises compliance standards to FIPS 140-3, supporting both classical and quantum-resistant security algorithms. New ADMIN_GET_TLS_CERT interface allows querying server TLS certificate information via SQL, simplifying certificate monitoring.

4. Full-Text Search Modernization: Integrates OpenSearch and Elasticsearch as self-hosted external search engines, replacing the legacy ECMTS engine to provide better performance and scalability.

5. Operational Efficiency: Column-organized tables support schema changes without downtime. pureScale installation and update speed improved by up to 45%. New license usage telemetry capabilities added.

DBA Perspective
This is one of the most noteworthy technology releases this week. The IBM Db2 12.1.5 update sends a clear signal: traditional enterprise databases are fully embracing AI, and they are choosing a different technical path than Oracle.

Three In-Depth Observations:

1. “SQL-calling-AI-models” is a differentiator: Oracle has taken the “Agent factory + natural language query” route, while IBM has chosen the lower-level approach of “SQL functions directly invoking models.” Which is better? It depends on the scenario. For developers already using SQL, IBM’s approach has a lower learning curve; for business users, Oracle’s Agent approach is more user-friendly. DBAs need to make judgments based on their enterprise’s technology stack characteristics.

2. DiskANN vector index is worth attention: Microsoft previously open-sourced DiskANN, a disk-based approximate nearest neighbor (ANN) algorithm that has significant advantages over HNSW in memory footprint. IBM’s choice of DiskANN over HNSW indicates different considerations for scaling vector search. For enterprises with hundreds of millions of data points, this could be a deciding factor.

3. The high availability “arms race” continues: More than 3 standby databases, cascading HADR, five-nines availability — these metrics indicate that the bar for enterprise database high availability is being continuously raised. DBAs should consider: does your business need such high availability? If so, can your existing architecture support it?

One sentence summary: The release of Db2 12.1.5 marks that the three traditional enterprise database giants (Oracle, IBM, Microsoft) have all completed the deployment of “AI-native” capabilities. DBAs can no longer wait to build AI skills.

02｜Top Database Conference ICDE 2026: Inspur KaiwuDB and Xidian University Paper Accepted, Bala-Join Technology Breaks Cross-Domain Query Bottleneck

ICDE 2026, one of the three top conferences in the database field, will be held in May in Montreal, Canada. The joint work between Inspur KaiwuDB and Xidian University, titled “Bala-Join: Adaptive Hash Join Technology for Cross-Domain Distributed SQL Databases,” has been accepted into the Industry Track, which has an acceptance rate of only about 25%.

Technical Core:

Bala-Join technology addresses three major challenges faced by cross-domain distributed databases: data skew, load imbalance, and high network overhead. It introduces three core innovations:

Performance Results:

• In TPC-H benchmark tests and real customer business workloads, system throughput improved by 25% to 61%
• Performance advantages are particularly significant in high data skew and cross-domain WAN scenarios
• Effectively solves problems of traditional solutions: load imbalance, high network overhead, and long-tail latency

DBA Perspective
This is technical news with practical reference value for DBAs, not purely an academic achievement.

Three Key Points:

1. Cross-domain queries are a common pain point in distributed databases: When business data is distributed across multiple data centers or cloud regions, JOIN query performance often degrades dramatically. If Bala-Join’s solution can be productized, it will be a major benefit for DBAs working with distributed databases.

2. The distance from research results to engineering implementation: The ICDE Industry Track features solutions that “address industry pain points and have engineering implementation value.” This indicates that Bala-Join is not theoretical — it has been validated with real business workloads. DBAs should pay attention to whether KaiwuDB will integrate this technology into its products.

3. Domestic database vendors are deepening their “technical moats”: From Dameng DM9 to Inspur KaiwuDB’s ICDE paper, domestic database vendors are increasing their investment in core technologies. For DBAs, this also means the domestic technology stack will become richer and more professional in the future.

One sentence summary: Bala-Join’s breakthrough shows that domestic databases can now converse at the level of top international conferences on the world-class challenge of distributed query optimization.

03｜CETC Kingware Speaks at Digital China Summit: AI Empowerment Shows Results, Large-Scale Replacement Exceeds 20,000 Sets

At the 9th Digital China Construction Summit, CETC Kingware, as a leading domestic database vendor, disclosed the latest results of large-scale replacement. The summit was co-hosted by the National Development and Reform Commission, National Data Bureau, Cyberspace Administration of China, Ministry of Industry and Information Technology, and the Fujian Provincial People’s Government, with the theme “Accelerating the Innovation and Development of Digital Intelligence Technology, Deeply Advancing the Construction of Digital China.”

Core Data:

Key Customers: State Grid Corporation of China, China’s “three oil giants” (CNPC, Sinopec, CNOOC), China Mobile, and other mega-enterprises.

AI Empowerment Achievements:

• Self-developed intelligent O&M system enables fault prediction, SQL optimization, and automatic self-healing
• High-performance vector retrieval supports RAG scenarios such as financial knowledge Q&A and government semantic search
• Promotes the transformation of O&M from “passive firefighting” to “proactive self-healing”

Market Interpretation: User selection of domestic databases has shifted from “policy-driven” to proactive selection driven by “experience, cost, and security.”

DBA Perspective
The first and second items in this issue covered IBM and Inspur, while this news focuses on the large-scale implementation results of domestic databases.

Three Key Insights:

1. The milestone significance of “20,000 sets”: If Dameng DM9 represents “technical height,” then CETC Kingware’s 20,000 sets of deployment represent “market breadth.” Together, they show that domestic databases have formed a complete closed loop from “usable” to “good-to-use” to “widely used.”

2. AI empowerment is not empty talk: CETC Kingware explicitly positions “intelligent O&M” and “vector retrieval” as core selling points, with practical implementation scenarios. This aligns with the trend reported in Issue 1 about Oracle AI Database — database + AI has become a global consensus.

3. The decision logic for choosing domestic databases is changing: The report mentions that “users have shifted from policy-driven to proactive selection driven by experience, cost, and security” — this is a very important signal. For DBAs still hesitating about building domestic database knowledge, this shows that domestic replacement is no longer a “future trend” but an “ongoing reality.”

One sentence summary: Behind CETC Kingware’s 20,000 sets of deployment is a fundamental shift of domestic databases from “policy-driven” to “value-driven.”

04｜University of Miami Launches Modernization: Saying Goodbye to Oracle, Embracing Snowflake

On May 1, the University of Miami’s IT Services department officially announced the launch of the “Next Generation Data Architecture” project. The core goal is: removing dependency on Oracle products and adopting Snowflake as the new cloud data warehouse platform.

Project Background:

• Current data warehouse is based on Oracle Analytic Services (OAS), which has been running for over ten years
• Oracle licenses will expire in June 2028, requiring migration before that date
• The university is transitioning from the Banner system to Workday, which has different data structures requiring a new architecture

Technology Selection:

• New data warehouse will use the Snowflake cloud platform
• Follows a “cloud-appropriate strategy” — if a service can leverage cloud technology, migration should be considered
• Data visualization tools include Tableau, R, Excel, etc.

Project Goals:

• Implement a modernized data platform in phases
• Provide decision-makers with convenient access to needed data
• Preserve existing data structures and critical business logic

DBA Perspective
This news appears to be a university IT transformation project, but it reflects major trends in the enterprise market.

Four Points to Consider:

1. Oracle is losing its “default option” status: The University of Miami is not unique. More and more enterprises and institutions are choosing alternatives when licenses expire rather than renewing. For DBAs, this means that while Oracle skills remain valuable, they can no longer be the only skill.

2. The rise of Snowflake: As a representative of cloud-native data warehouses, Snowflake is eroding the market share of traditional database vendors. DBAs need to start learning the architectural concepts of cloud data warehouses — separation of storage and compute, elastic scaling, pay-as-you-go.

3. “Cloud-appropriate” strategy becomes mainstream: The University of Miami explicitly adopts a “cloud-appropriate strategy,” a pragmatic hybrid cloud approach. For DBAs, this means needing to have operational capabilities for both on-premises and cloud environments simultaneously.

4. Urgency of the migration window: There are approximately 26 months until June 2028, but they have already started. This indicates that large-scale system migration has a long preparation cycle. If your enterprise also faces a similar license expiration timeline, you should start evaluating now.

One sentence summary: The University of Miami’s “de-Oracleization” is just one example. The “great migration” in the database market is accelerating, and DBAs need to prepare in advance.

05｜KaiwuDB 4.9 Million Yuan Procurement Contract: Inspur Doubles Down on Domestic Time-Series Database

On April 29, Inspur Software Group issued a direct procurement announcement, planning to spend 4.9 million yuan to procure the “KaiwuDB Predictive Analysis Engine,” with Shanghai Yunxi Technology Co., Ltd. as the supplier. The direct procurement method was chosen because “procuring the group’s own products or products of invested/acquired companies.”

Project Information:

• Project Name: Inspur Software Group Database Predictive Analysis Engine Procurement Project
• Procuring Entity: Inspur Software Group Co., Ltd. Project 817
• Subject: KaiwuDB Predictive Analysis Engine
• Supplier: Shanghai Yunxi Technology Co., Ltd.
• Budget Amount: 4.9 million yuan

KaiwuDB Background: KaiwuDB is a domestic distributed multi-model database independently developed by Inspur Group for IoT and AIoT scenarios. It features industrial time-series data processing, multi-model integration, and AI-native capabilities. The ICDE 2026 paper mentioned in the second item of this issue was a joint achievement between Inspur KaiwuDB and Xidian University.

DBA Perspective
This procurement announcement provides an indirect perspective on the commercialization progress and large-scale deployment trends of domestic databases.

Three Observations:

1. Inspur’s database strategy: From self-developed KaiwuDB to procuring the KaiwuDB Predictive Analysis Engine, Inspur’s investment in the database field is sustained and strategic. For DBAs following domestic database development, Inspur is a player worth continuous tracking.

2. Reference value of the procurement amount: The 4.9 million yuan procurement for a predictive analysis engine can serve as a reference benchmark for DBAs or enterprises evaluating domestic database procurement costs. Of course, prices vary by product and deployment scale.

3. The industry logic of “priority procurement of invested/acquired products”: The announcement explicitly mentions “procuring the group’s own products or products of invested/acquired companies,” reflecting the strategy of large technology groups to cultivate core products through internal ecosystem collaboration. For DBAs, this means that choosing a particular technology ecosystem will come with richer supporting products and services.

One sentence summary: The 4.9 million yuan procurement of KaiwuDB is a microcosm of domestic time-series databases entering large-scale commercial use. Demand for time-series databases in IoT and AIoT scenarios is exploding.

📅 This Week’s Database Hot Topics Review

Today is May 2, 2026, the second day of the May Day holiday. Let’s review this week’s important events:

Summary of This Issue

📌 Editor’s Note
The second day of the May Day holiday, DBAs enjoy your rest! This week’s news reveals several clear signals:

1. The “AI arms race” for enterprise databases is fully underway: The release of IBM Db2 12.1.5 means that the three traditional database giants — Oracle, IBM, and Microsoft — have all completed their deployment of “AI-native” capabilities. DBAs can no longer wait to build AI skills.

2. Domestic databases are moving from “replacement” to “leadership”: CETC Kingware’s 20,000 sets of deployment and Inspur KaiwuDB’s acceptance into ICDE — these are not isolated news items but an overall signal of the maturity of the domestic database industry.

3. “De-Oracleization” is becoming a global trend: The University of Miami case is just the tip of the iceberg. The three factors of license costs, cloud transformation, and domestic replacement are converging. DBAs need to prepare for a multi-technology-stack future.

Three Suggestions for the Remainder of the Holiday:

1. Use the last two days of the holiday to learn the basics of vector search (the difference between DiskANN and HNSW)
2. If your enterprise is still using MySQL 8.0, make upgrading assessment the first thing you do after the holiday
3. Pay attention to the architectural concepts of cloud data warehouses (Snowflake, Redshift, BigQuery) — this is the future direction

Welcome to leave comments: Is your enterprise also planning “de-Oracleization”? What alternative solutions are you choosing? See you after the holiday!