HiddenMerit Daily · Issue 2

📊 HiddenMerit Daily · Issue 2Focus on Database Frontiers, Practical Insights for DBAsApril 28, 2026 ｜ 5 Selected Global Hot Topics

01｜Dameng Releases Four New Products: Domestic Databases Make Waves AgainOn April 22, the 2026 China Database Technology and Industry Conference, themed "Dameng Builds Foundation, AI Opens a New Era," was held in Beijing. Dameng Data centrally released four strategic new products: DM9 Intelligent Database, DAMENG PAI V2.0 All-in-One Machine, Dameng Qiyun Database V4.0, and GDMBASE V4.0 Graph Database, covering centralized, distributed, cloud-native, all-in-one, and graph database scenarios.Core Highlights:

ProductCore CapabilitiesKey MetricsDM9Centralized & distributed integration, supports vector/multi-model data, database agents450+ new featuresDAMENG PAI V2.0Ultra-fast storage + compute offload + full-stack RDMAI/O latency reduced from 400μs to 80μs, 12M+ IOPSQiyun V4.0AI-powered intelligent O&M, multi-cluster hybrid disaster recoverySolves "low O&M efficiency, complex disaster recovery, fragmented compute power"GDMBASE V4.0Graph-native + vector integration, AI long-term memorySub-500ms retrieval for billion-level nodes/edges

Market Data: According to CCID Consulting, China's database market reached 43.02 billion yuan in 2025, with domestic vendors already leading foreign vendors in market share. Dameng firmly ranks first among domestic vendors.

DBA PerspectiveThis is a milestone event for domestic databases moving from "usable" to "good-to-use". The most noteworthy signal: Dameng is betting on four tracks simultaneously — centralized, distributed, cloud-native, and graph databases — indicating that domestic vendors are no longer satisfied with single-point breakthroughs but are pursuing full-stack deployment.

Three takeaways for DBAs:

1. DM9's "centralized & distributed integration" is worth watching: the same product can meet both traditional OLTP and distributed scaling needs, reducing selection complexity.
2. GDMBASE's "graph-native + vector integration" directly addresses AI Agent pain points — when business needs to store knowledge graphs, there's no need for an additional graph database.
3. Domestic-tech replacement enters deep waters: China Unicom completed full-stack ERP transformation across 31 provinces, and Fuzhou First General Hospital's 8 campuses saw core system CPU usage as low as 3% — these cases prove that domestic databases can "stay stable" in core scenarios. DBAs should start building knowledge of Dameng, OceanBase, and other domestic databases, starting with non-core business pilots.

02｜Oracle Partners with Google Cloud: Query Databases in Natural Language — Is SQL Going Extinct?

At the Google Cloud Next 2026 conference, Oracle announced that Oracle AI Database Agent for Gemini Enterprise is now available on Google Cloud Marketplace.

Core Capabilities:

• Users ask questions in natural language (e.g., "What were sales trends in East China last quarter?"), and the Agent automatically understands intent, queries the Oracle database, generates analytical results and visualizations
• No SQL writing required throughout the process, no need to build data pipelines
• Inherits users' original database permissions via OAuth, query results limited to authorized tables/columns/policies

Business Model Progress:

• Already available in 15 regions globally, will expand to Europe and North America over the next 12 months
• Oracle's Q3 financial report shows: multi-cloud database revenue grew 531% year-over-year, cloud infrastructure revenue grew 84% to $4.9 billion

DBA PerspectiveThis news may seem like "DBAs are about to be replaced," but upon calm analysis: natural language to SQL is not a new concept — Microsoft Copilot for Azure SQL, Amazon SageMaker Data Agent are all working on similar things. Oracle's differentiation is embedding the Agent into Gemini, an office collaboration platform, directly serving business users rather than developers.

Real impact assessment:

• Current stage: Agents can handle simple queries ("this month's sales"), but complex multi-table joins and performance tuning still require humans
• Impact on DBA work: The DBA role shifts from "SQL writer" to "permission policy designer" — because the Agent's query results depend entirely on the permission boundaries (row/column/table level) defined by the DBA
• Security challenge: If permission policies are poorly designed, business users might combine natural language queries to "inadvertently" piece together sensitive data

Recommendation: DBAs should proactively understand the permission model of Oracle AI Database Agent and evaluate whether there are any "over-privileged" scenarios in existing databases — this is the risk most amplified in the Agent era.

03｜MySQL 8.0 Officially EOL: Farewell to a Classic Version, Upgrade is Urgent

Major News: On April 21, 2026, MySQL 8.0 officially reached End of Life (EOL) and entered Oracle Sustaining Support phase.

What This Means:

• No more new features, bug fixes, or security patches
• Extended support is still available (paid), but only for critical security issues and key bugs
• Oracle officially recommends upgrading to MySQL 8.4 LTS or 9.7 LTS

Version Evolution Timeline:

text

MySQL 8.0 (released 2018) → April 2026 EOL
MySQL 8.4 LTS (released 2024) → current stable release
MySQL 9.7 LTS → future LTS release

DBA PerspectiveThis is the end of an era in MySQL history. 8.0 has served countless production environments since its release in 2018. DBAs now face a real problem: 8.0 users must weigh between "delaying upgrade" and "upgrading urgently."

Upgrade Path Recommendations:

Current VersionSuggested ActionTime Window8.0.46 and belowPlan upgrade to 8.4 LTSWithin 6 months8.4 LTSContinue stable operationNo pressure (LTS supported until 2030+)5.7 and earlierMust upgradeAlready overdue, prioritize

Upgrade Checklist:

1. Run mysqlcheck to check all tables
2. Confirm application compatibility (especially caching_sha2_password vs mysql_native_password)
3. Prepare rollback plan (8.0 → 5.7 does not support downgrade; make sure to have full backups)

One sentence summary: If you're still running MySQL 8.0, now is the time to make an upgrade plan — don't wait until security vulnerabilities hit your production systems.

04｜PostgreSQL 19 Preview: Graph Queries, REPACK Kernelization, Upgraded AI Capabilities

PostgreSQL 18 was released in 2025, and the industry is already looking ahead to the upcoming PostgreSQL 19. According to expert interpretations at the HOW 2026 conference, PG 19 will introduce several major features.

PG 18 Review (already released):

• Asynchronous I/O: Significantly improved performance for large I/O operations in cloud storage scenarios (index creation, VACUUM, large table scans)
• Statistics retention: Execution plan statistics retained after pg_upgrade upgrades, ending the era of "upgrade equals incident"
• disabled_nodes: More precise execution plan control, replacing the blunt disable_cost approach
• pg_overexplain: Structured output of optimizer internal decisions, helping diagnose execution plan problems

PG 19 Preview (in development):

FeatureDescriptionDBA ValueREPACK CONCURRENTLYTable reorganization function kernelized, replacing third-party pg_repack extensionSolves table bloat, reduces cloud storage costsGraph queriesNative graph structure query support, no need for separate graph databaseAI Agent knowledge graph, memory storage scenariosWAIT FORRead replicas can wait for primary's LSN replay before executing queriesStrong consistency guarantee in read-write separation scenariosParallel VACUUMParallel garbage collection across multiple indexesReduces VACUUM time, reduces business impactLogical replication supports SEQUENCESequence values can be synced via logical replicationImproves logical replication functionality

DBA PerspectivePG's evolution path is very clear: kernelization of proven third-party extensions. REPACK kernelization is a typical example — previously DBAs relied on pg_repack to solve table bloat, but there were compatibility risks. Now with direct kernel support, stability is greatly improved.

Upgrade Strategy Recommendations:

ScenarioRecommended VersionRationaleCloud deployment + large I/O operationsPG 18Asynchronous I/O brings significant performance gainsTroubled by table bloatWait for PG 19REPACK kernelization, more stableAI Agent + knowledge graphWait for PG 19Native graph queries, unified data planeDomestic-tech + cross-CPU architecturePG 17/18Built-in character sorting compatibility

Production environment suggestion: Maintain "pain point driven" principle — don't upgrade just for new features; only upgrade when the new version solves an existing pain point.

05｜Oracle April Critical Patch Update: Fixes 481 Vulnerabilities, 27 in Database

Oracle released the second quarter 2026 Critical Patch Update (CPU) , fixing a total of 481 security vulnerabilities — one of the largest patch updates in recent years.

Key Data:

• Total vulnerabilities: 481
• Of which 376 (approx. 78%) come from third-party open-source components included in Oracle products
• Database products: 27 security updates, highest CVSS score 7.5
• Affected products: Oracle Database Server, GoldenGate (10 updates), MySQL, WebLogic Server, Java SE, etc.

Vulnerability Distribution by Product Line:

Product LineNumber of VulnerabilitiesPercentageOracle Communications13928%Financial Services7515.6%Fusion Middleware5912.3%GoldenGate10-Database Server8-

DBA PerspectiveThis is a routine quarterly CPU, but the scale is alarming — 78% of vulnerabilities come from third-party open-source components, indicating that software supply chain security has become a major risk for modern databases.

DBA Action List:

1. Assess impact scope: Log into Oracle Support, download the CPUAPR2026 patch, check for vulnerabilities affecting your environment
2. Prioritize CVSS 7.5+ vulnerabilities: 8 new security updates for database products, highest score 7.5
3. Pay attention to third-party components: Even if the Oracle database kernel has no vulnerabilities, introduced open-source libraries can become a breach point
4. MySQL users note: The April CPU also includes security updates for MySQL Server — need to assess as well

One sentence summary: DBAs can't just focus on the database kernel; they also need to pay attention to "which open-source libraries are inside the database" — this may become the new normal for security operations.

Summary of This Issue

TopicKeywordsDBA Action SuggestionsDameng 4 new productsDomestic database, full-stack deploymentBuild domestic database knowledge, start with non-core pilotsOracle natural language AgentAI + database, natural language queryStudy permission model, audit existing over-privileged accessMySQL 8.0 EOLVersion upgrade, security complianceMake upgrade plan, migrate to 8.4 LTS within 6 monthsPG 19 previewGraph query, REPACK kernelizationPain point driven upgrade, focus on PG 18 for cloud storageOracle CPU patchSecurity vulnerabilities, third-party componentsAssess vulnerability impact, focus on software supply chain security

📌 Editor's NoteThis issue's news reveals a clear signal: databases are evolving in layers — the bottom layer is security compliance (MySQL 8.0 EOL, Oracle CPU), the middle layer is AI capability integration (Dameng vector+graph, PG 19 graph queries), and the top layer is human-computer interaction change (natural language query Agents). As DBAs, we need to advance on three fronts simultaneously: hold the security bottom line, build AI skills, plan upgrade paths.

Welcome to leave comments: After MySQL 8.0 EOL, are you planning to upgrade to 8.4 LTS or go directly to 9.x? See you tomorrow!