Database Security Gateway

Stop unauthorized database
access before data leaves.

Identity-aware access control, policy enforcement, dynamic masking and complete audit — an inline security proxy that sits between your applications and your databases. No changes to your clients. No changes to your database.

One policy layer for every access to your data — by people, applications, and AI agents.

Enterprise Security
Compliance Ready
High Performance
Zero Downtime

Your database is the last place you can’t see

Applications talk to the database directly, over a shared login, on a protocol your security tools don’t read. When something goes wrong, you find out from the logs — after the data has already left.

One login, many users

Apps connect with a single service account, so you can’t tell one user’s access from another — or stop just one of them.

Logging is after the fact

Audit logs tell you what happened once it already has. They record the breach; they don’t prevent it.

Sensitive data walks out

Query results return raw PAN, PII and secrets in full. Nothing sits in the path to mask or block them.

Inspect, don’t implement

DSGuard sits inline between your applications and your databases. It reads the wire protocol to understand each request, applies your policy, then forwards the traffic unchanged — no protocol emulation, no client or database changes.

Applications

Your existing clients & services, unchanged

DSGuard Gateway

Inspect · authenticate · enforce · mask · audit

Databases

PostgreSQL, Oracle, SQL Server, MySQL, MongoDB, ClickHouse

1

A query arrives

The client connects to DSGuard instead of the database. DSGuard authenticates the user (MFA) before any query runs.

2

It’s inspected

The SQL is extracted from the wire protocol and checked against injection patterns, dangerous-function rules, object permissions and time/app/IP policy.

3

Allowed, masked or blocked

Permitted traffic passes through; sensitive columns are masked in the response; disallowed requests are denied by default — every outcome is decided inline.

4

Everything is logged

Query, identity, decision and reason land in the audit store and dashboards — a complete, attributable record of who did what.

Inside the gateway

  • Authentication — MFA/TOTP, Ident/Peer
  • Policy engine — object permissions, time, application and IP rules, connection limits
  • Threat prevention — SQL-injection patterns, dangerous functions, rate limiting
  • Masking — column-level, full/partial/hash, format-preserving and Lua-scripted
  • Audit — query, identity, decision and block reason, with GeoIP context

Fits your existing stack

  • Identity — TOTP MFA
  • Observability — Prometheus metrics endpoint, provisioned Grafana dashboards
  • Log pipeline — centralized query and application logs for your SIEM
  • Transport — TLS-encrypted client connections on every engine, TLS to supported backends, Oracle Native Network Encryption
  • Deployment — on-prem, private or public cloud, hybrid; no client or schema changes

Complete Database Protection

Everything you need to secure, monitor, and manage access to your database infrastructure.

Data Protection

Keep sensitive fields out of query results — without modifying your applications.

  • Dynamic Data Masking (Full, Partial, Hash)
  • Column-Level Access Control
  • Response Leak Detection (OWASP CRS)
  • Result Limiting (Rows/Bytes/Time)
  • Query Result Truncation

Authentication

Put a verified human identity in front of every shared database account.

  • Multi-Factor Authentication (TOTP)
  • IP-Based Access Control
  • Ident/Peer Authentication

Access Control

Least privilege enforced at query time — not just at login.

  • Role-Based Permissions (RBAC)
  • Object-Level Permissions
  • User/Group Management
  • Session Management & Ban
  • Connection Limits per User/IP

Threat Prevention

Attacks are blocked inline, before they reach the database.

  • SQL Injection Protection
  • Dangerous Function Blocking
  • Rate Limiting (Token Bucket)
  • SQL Pattern Rules
  • Application Fingerprinting

Audit & Logging

An attributable record of who did what — and why it was allowed or blocked.

  • Complete Query Logging
  • Centralized Application Logs
  • Session & Transaction Tracking
  • Block Reason Audit Trail
  • GeoIP Location Tracking

Operations

Keep serving traffic through failures, maintenance and load spikes.

  • Time-Based Access Rules
  • Connection Pooling
  • High Availability & Failover
  • Read/Write Splitting
  • Health Check Monitoring

What DSGuard protects against

Each of these is stopped by a control that runs inline on the request path — not by a report you read afterwards.

Credential theft

Stolen database credentials alone aren’t enough: MFA authenticates the human, and IP rules bound where they can connect from.

SQL injection

Injection patterns are matched against the extracted SQL before it reaches the database, and the attempt is blocked and logged.

Insider threat

Object-level permissions, time-of-day rules and application rules constrain what a legitimate account may do — not just whether it may connect.

Data exfiltration

Result limiting, query-result truncation and response-leak detection cap how much data any single session can pull out.

Privilege abuse

Dangerous-function blocking stops destructive and privilege-escalating statements even when the account technically holds the grant.

Unauthorized DBA access

Administrative sessions run under the same policy and audit as everyone else, and can be terminated live from the console.

Shared accounts

Per-user proxy logins put an identity in front of a shared service account, so activity is attributable to a person again.

Compliance violations

Dynamic masking keeps regulated fields out of results, and every decision is recorded with its reason for the audit trail.

Deploy Anywhere

Flexible deployment options to match your infrastructure and compliance requirements.

On-Premises

Your data center, your control

Private Cloud

Dedicated cloud infrastructure

Public Cloud

AWS, Azure, GCP compatible

Hybrid

Unified multi-environment

Built for

Securing critical database infrastructure for regulated companies.

Banking & Finance
Insurance
FinTech Platforms
Data-Intensive Systems
High-Volume Databases

Comprehensive Coverage

One platform for your entire database ecosystem.

Relational (SQL)

PostgreSQL
Oracle Database
Microsoft SQL Server
MySQL
MariaDB*

NoSQL

MongoDB

Analytical

ClickHouse

* MariaDB is supported through the MySQL-compatible wire protocol.

Give AI agents a shell you can trust

The same inspect-and-enforce approach, applied to command execution. Operators and AI agents run commands through a sandboxed shell where every command is policy-checked, sensitive output is masked, and the whole session is recorded.

Command rules

Allow, deny or require approval per command — with a secure default-deny policy so nothing runs unless you permit it.

PII & column masking

Detect and mask secrets, PII and sensitive columns in command output before it ever reaches the caller — including format-preserving masking.

Network & egress control

Per-domain egress rules control exactly which hosts a session can reach — fail-closed when a rule can’t be applied.

Filesystem allowlisting

Landlock-based path allowlists and per-file access rules scope every session to exactly the files it is allowed to touch.

Live session control

Watch active sessions and revoke or ban one instantly — the enforcement is server-side, so a client cannot talk its way around it.

Complete audit

Every command, decision and masked result is logged with its identity and reason — the same audit discipline as the database gateway.

Controls that map to your obligations

DSGuard’s access control, masking and complete audit trail give you concrete, demonstrable evidence for the frameworks your auditors ask about.

FrameworkControl areaDSGuard control
PCI DSS Req. 3.3 — Mask PAN when displayed Dynamic masking keeps full PAN and other regulated fields out of displayed query results.
Req. 7 — Restrict access by need‑to‑know Role-based and object-level permissions with a default-deny policy.
Req. 8 — Identify and authenticate access Per-user proxy logins with MFA/TOTP — in front of shared service accounts.
Req. 10 — Log and monitor all access Every query logged with identity, decision, block reason and GeoIP context.
ISO 27001 Access control Centralized least-privilege policy across every supported database engine.
Operations security & logging Continuous audit trail plus Prometheus metrics and Grafana dashboards.
Communications security TLS to clients and backends; Oracle Native Network Encryption support.
SOX Integrity of financial reporting data Attributable record of who read or changed financial data, and under which rule.
HIPAA PHI access & safeguards Column-level restriction and masking of PHI, evidenced by continuous logging.
GDPR / KVKK Data minimisation & accountability Masking limits what is exposed; purpose-bound access rules and processing records.

DSGuard provides technical controls and audit evidence that support these frameworks; it is not, by itself, a certification. Control-area names are given by theme, as framework numbering differs between revisions.

Start free, license when it matters

Download and run the full gateway with no licence and no registration — all features, no time limit. Move to a licence when you need more datasources, more ShellAI users, or support.

Community Edition

No licence · no registration

  • One-command install — the full stack ships as a single all-in-one container, available on Docker Hub
  • Up to 3 datasources
  • Up to 3 ShellAI users
  • All features included — SQL-injection blocking, dangerous functions, object permissions, masking and more, with no time limit
  • MFA/TOTP

No expiry and no feature-gating — the full gateway stays protected. Add a licence when you need more datasources, more ShellAI users, or support.

Enterprise

Annual subscription · custom terms

  • Everything in Professional — all features, capacity as licensed
  • 24×7 support with a 4-hour response target — around the clock, every day
  • Custom licence terms — capacity, contract length and invoicing shaped to your organisation

Licensing is applied per service category — database proxies are counted by datasource, the AI Sandbox Shell by user. For pricing and evaluation terms, email us at dsg@ds-guard.com.