Executive Summary

In November 2026, security researcher Zheng Yu of Depthfirst published proof-of-concept exploits demonstrating remote code execution (RCE) against PostgreSQL, MySQL, ClickHouse, and Oracle Database, exposing a critical architectural gap: low-privilege database accounts with no access to application data can still trigger memory-corruption bugs in the underlying C/C++ engine code through ordinary SQL operations. Four CVEs were assigned and patched across PostgreSQL (CVE-2026-116331), MySQL (CVE-2026-101786), Oracle Text (CVE-2026-114819, CVSS 9.9), and Oracle REST Data Services (CVE-2026-91609). The Oracle chain requires no login at all, exploiting exposed web endpoints pre-authentication. ClickHouse remains without a confirmed fix.

This research reframes database privilege models as insufficient security boundaries: SQL-level access controls do not constrain memory-unsafe engine execution paths. With four major database platforms affected simultaneously and one unauthenticated Oracle chain active, organizations relying solely on database-level permissions face systemic RCE exposure across hybrid and multi-cloud data tiers.

Why This Matters Now

Database servers sit at the core of every regulated workload—PCI, HIPAA, SOX. These exploits prove that least-privilege SQL grants alone cannot prevent RCE when the engine itself is reachable via ordinary queries or unauthenticated web endpoints. With patches still incomplete for ClickHouse and the Oracle unauthenticated path only recently closed, organizations must urgently audit temporary-table privileges and enforce network-layer controls around database tiers to contain lateral movement and exfiltration risk.

Attack Path Analysis

Related CVEs

MITRE ATT&CK® Techniques

Potential Compliance Exposure

Sector Implications

Sources

Frequently Asked Questions

Database SQL permissions govern which tables and schemas an account can read or write, but they do not constrain the memory-unsafe C/C++ engine code that executes every query. A minimal grant such as TEMPORARY table creation is sufficient to craft input that triggers memory-corruption bugs deep in the engine, escalating from SQL-level access to operating-system-level code execution.

Cloud Native Security Fabric Mitigations and ControlsCNSF

Based on the attack progression modeled above, these are the defensive controls that would constrain each stage.

Aviatrix Zero Trust CNSF is directly relevant here because attackers exploited exposed database endpoints to gain OS-level footholds and move freely across internal tiers — conditions that identity-aware segmentation and controlled egress enforcement are specifically designed to constrain. By limiting reachability to database workloads and restricting outbound paths, CNSF would likely reduce the attacker's ability to pivot, beacon, and exfiltrate at scale.

Initial Compromise

Control: Cloud Native Security Fabric (CNSF)

Mitigation: Restricting which network sources could reach database service ports would likely reduce the exposed attack surface, making it harder for external actors to deliver exploit payloads directly to vulnerable database endpoints.

Privilege Escalation

Control: Zero Trust Segmentation

Mitigation: Enforcing least-privilege identity-based segmentation policies around the database OS user context would likely constrain what resources and services the compromised process could reach, reducing the blast radius even after OS-level execution is achieved.

Lateral Movement

Control: East-West Traffic Security

Mitigation: Workload-to-workload segmentation enforced across internal tiers would likely constrain the attacker's ability to pivot freely from the compromised database host to adjacent services, reducing the number of reachable targets within the environment.

Command & Control

Control: Multicloud Visibility & Control

Mitigation: Multicloud visibility into outbound connection patterns from database workloads would likely surface anomalous egress behavior early, giving teams the context needed to constrain unauthorized C2 channels before they become persistent.

Exfiltration

Control: Egress Security & Policy Enforcement

Mitigation: FQDN-based egress allowlists and outbound traffic policy enforcement would likely constrain the attacker's ability to transfer bulk data to unauthorized external destinations, reducing the volume and reach of data that could be exfiltrated from the compromised host.

Impact (Mitigations)

Because lateral movement and egress were likely constrained by earlier CNSF controls, the attacker's impact would likely be scoped closer to the initially compromised database workload rather than spreading across dependent application tiers or additional cloud environments.

Impact at a Glance

Affected Business Functions

  • Database-Backed Application Services
  • Data Integrity and Confidentiality of Application Records
  • Database Server Operating System Integrity
  • API and Web Endpoint Security (Oracle REST Data Services)
Operational Disruption

Estimated downtime: 7 days

Financial Impact

Estimated loss: N/A

Data Exposure

Any data stored within affected PostgreSQL, MySQL, or Oracle Database instances is at risk of full exfiltration or modification. Because the exploits achieve OS-level code execution, attackers could access all application tables, credentials, cryptographic material, and configuration files residing on the database host. The Oracle chain requires no prior authentication, meaning externally exposed Oracle REST Data Services endpoints place entire database contents at risk without any attacker foothold. ClickHouse deployments running version 26.7.5.10 with user-configurable binary format options are similarly exposed to full data compromise and shell access. No confirmed exfiltration has been reported; risk is currently limited to proof-of-concept exposure.

Recommended Actions

  • • Apply all vendor patches immediately for CVE-2026-116331 (PostgreSQL), CVE-2026-101786 (MySQL), CVE-2026-114819 and CVE-2026-91609 (Oracle), and restrict ClickHouse deployments to vendor-confirmed fixed versions; patch the Oracle GraphQL/REST endpoint exposure as the unauthenticated entry point represents the highest-priority risk.
  • • Enforce Zero Trust Segmentation by auditing and revoking TEMPORARY table privileges and broad temporary-table rights from all database accounts that do not explicitly require them, applying least-privilege identity-based policies aligned with ZTMM.Network and NIST.800-53.AC-6 to constrain the SQL-to-OS attack surface.
  • • Deploy East-West Traffic Security controls to enforce workload-to-workload microsegmentation between database hosts and adjacent application or infrastructure tiers, ensuring that OS-level compromise of a database server cannot be used for uninspected lateral movement to other internal services.
  • • Implement Egress Security & Policy Enforcement with FQDN-based allowlists and Cloud Firewall (ACF) rules to block unauthorized outbound connections from database hosts, preventing reverse-shell C2 establishment and direct data exfiltration from compromised database OS contexts.
  • • Enable Multicloud Visibility & Control and Inline IPS (Suricata) with signatures covering known database exploit traffic patterns and anomalous bulk outbound data transfer, providing detection and inline blocking capability for exploit attempts and post-compromise exfiltration across all database tiers.

Secure the Paths Between Cloud Workloads

A cloud-native security fabric that enforces Zero Trust across workload communication—reducing attack paths, compliance risk, and operational complexity.

Cta pattren Image