Ref Sp DB Security

Database Security

Purpose

Provide portable defaults for protecting databases as socio-technical systems where confidentiality, integrity, availability, and privacy depend on more than SQL permissions alone.

When to use this skill

• Designing or reviewing database access-control models, roles, and privileges.
• Protecting sensitive datasets, regulated data, or administrative interfaces.
• Deciding how to use views, auditing, encryption, integrity controls, or secure recovery workflows.
• Reviewing how backups, logs, replicas, exports, and operational tooling expose database data.
• Building a database threat model that includes accidental damage as well as malicious misuse.

Scope boundaries

This skill owns protecting a database: threats, privileges, exposure, auditing, encryption, and secure recovery.

• ref-sp-agents-security — despite the similar name, an unrelated subject: which files an agent may read in a repo. Nothing to do with database access control.
• ref-sp-db-schema-design — constraints that keep data correct. This skill uses constraints and views to keep data protected; the two overlap in mechanism, not in purpose.
• ref-sp-db-operations — how backup, restore, and migration work. This skill owns keeping those artifacts and paths from becoming an attack surface.
• ref-sp-db-distributed — where replicas and cross-site links live. This skill owns securing them.

Defaults

• Treat database security as a combination of confidentiality, integrity, availability, and privacy, not as authentication alone.
• Assume the DBMS is only as secure as the surrounding operating system, network, procedures, and human controls.
• Grant the narrowest privileges that still let each user, service, or operator do its job.
• Prefer role separation and controlled access paths over broad direct table access.
• Treat backups, logs, replicas, exports, and recovery artifacts as sensitive data, not as harmless copies.
• Pair preventive controls with detection, auditability, and tested recovery paths.

Core Rules

Start from threats and risk, not from vendor features

• Identify both intentional and accidental threats, including theft, fraud, misuse, privacy loss, corruption, hardware failure, and loss of availability.
• Decide which datasets require confidentiality, integrity, availability, or privacy most strongly.
• Include insiders, operators, applications, infrastructure failures, and physical media exposure in the threat model.
• Match controls to realistic business damage rather than applying every feature uniformly.

Control access explicitly

• Grant and revoke privileges deliberately instead of relying on default broad access.
• Separate application roles, read-only users, administrators, and auditors so each path has only the permissions it needs.
• Use discretionary access control when object owners can safely manage sharing, and mandatory access control when centrally enforced classification rules are required.
• Restrict privileged utilities and direct administrative access as tightly as ordinary data access.

Reduce exposure with views, boundaries, and integrity controls

• Use views, stored interfaces, masking, or row and column restrictions to minimize direct exposure of underlying tables.
• Enforce integrity rules in the database so unauthorized or malformed changes cannot silently corrupt trusted data.
• Encrypt sensitive data at rest and in transit, and keep key management separated from routine application access.
• Treat redundancy technologies such as RAID as availability controls, not as substitutes for confidentiality or authorization.

Audit and recover securely

• Log meaningful access, schema changes, privileged actions, and failed authorization attempts when the workload needs accountability.
• Protect audit logs from tampering and uncontrolled retention.
• Keep backup copies, log files, and restore media in secure locations with controlled access.
• Recovery plans should restore the latest consistent state without bypassing the normal protection model.

Treat procedures and people as part of the design

• Define onboarding, offboarding, credential rotation, incident response, and emergency access procedures.
• Limit direct production access and make break-glass paths explicit, monitored, and reviewable.
• Patch and harden the DBMS, operating system, drivers, and adjacent tooling together instead of assuming the database can be secured in isolation.
• Test restore and incident procedures so security controls still hold during failures.

Gotchas

• Database permissions alone do not secure exported files, backups, replicas, or logs.
• Encryption without good key management or role design is incomplete protection.
• Backups improve recovery but also expand the attack surface if they are copied broadly or stored carelessly.
• Views reduce exposure but do not replace least privilege or auditing.
• Security designs that ignore operations, people, or the surrounding host environment are brittle.

Validation

• The design names the major confidentiality, integrity, availability, and privacy risks for the database.
• Roles, privileges, and administrative paths are explicit enough to verify least privilege.
• Views, auditing, encryption, and integrity controls match the sensitivity of the stored data.
• Backups, logs, replicas, and exports receive the same security attention as primary tables.
• Incident response and restore procedures preserve both availability and access control expectations.

References

• W3Schools Database Security: https://www.w3schools.in/dbms/database-security
• W3Schools Data Recovery in DBMS: https://www.w3schools.in/dbms/data-recovery-in-dbms
• W3Schools Database Architecture: https://www.w3schools.in/dbms/database-architecture
• Wikipedia: https://en.wikipedia.org/wiki/Database