DataBridge Cloud

DataBridge Cloud is the managed version of the data quality monitoring platform. It provides a visual UI for profiling datasets, defining quality checks, scheduling automated validation runs and alerting on failures.

Beta Status

DataBridge Cloud is currently in closed beta. Sign up for early access.

---

Why Use DataBridge Cloud?

Visual Interface

• No YAML Required: Define checks using a visual builder
• Dataset Explorer: Browse tables and columns interactively
• Live Profiling: See column statistics, distributions and quality metrics
• Historical Trends: Track data quality over time with check history

Scheduled Monitoring

• Automated Runs: Schedule profiling and checks on recurring intervals
• No Infrastructure: No cron jobs or Airflow DAGs to maintain
• Reliable Execution: Built-in retry logic with exponential backoff

Built-In Alerting

• Multi-Channel: Slack, email, webhooks, Telegram
• Per-Dataset Configuration: Attach different alert destinations to different datasets

---

Supported Data Stores

DataBridge Cloud supports quality monitoring for:

• ClickHouse -- High-performance analytics databases
• PostgreSQL -- General-purpose relational databases
• MySQL -- Popular open-source databases

You can connect data stores in two ways:

• Cloud Mode: DataBridge connects directly to your warehouse using credentials stored in the platform
• Agent Mode: A DataBridge Agent runs in your infrastructure, connecting to your warehouse locally and reporting results back to the cloud

See the Data Stores section for connection configuration details.

---

Dataset Profiling

Once a data store is connected and datasets (tables) are imported, DataBridge profiles them on a recurring schedule. Profiling collects:

• Row count and table size
• Column statistics: null/blank counts, cardinality, unique values
• Distribution metrics: min, max, average, standard deviation
• Most frequent values: top values by occurrence
• Row samples: preview of actual data for context

Profiling results appear in the dataset detail view and are stored historically so you can track changes over time.

---

Quality Checks

Check Types

DataBridge supports two categories of quality checks:

Standard checks use an expression format: function(column) operator threshold

Raw query checks use custom SQL that returns a single value, validated against a threshold.

Standard Check Functions

Table-Scope Functions

These operate on the entire table, without targeting a specific column:

Function	Description	Example
row_count	Total number of rows	row_count > 1000

Column-Scope Functions

These target a specific column:

Function	Description	Example
not_null(col)	Checks that no null values exist in the column	not_null(user_id)
uniqueness(col)	Checks that all values in the column are unique	uniqueness(email)
freshness(col)	Checks recency of the most recent timestamp value	freshness(updated_at) < 24h
min(col)	Validates the minimum value in a column	min(price) > 0
max(col)	Validates the maximum value in a column	max(discount) <= 100
sum(col)	Validates the sum of values in a column	sum(quantity) between 1000 and 5000
stddev(col)	Validates the standard deviation of values	stddev(latency_ms) < 200

Expression Format

Standard check expressions follow this pattern:

function(column) operator threshold

Supported operators: >, >=, <, <=, =, !=, between ... and ...

Threshold values:

• Numbers: 100, 3.14
• Time durations (for freshness): 1d, 24h, 30m

Examples:

row_count > 1000
not_null(user_id)
uniqueness(email)
freshness(created_at) < 7d
min(price) >= 0
max(quantity) <= 9999
sum(revenue) between 10000 and 500000
stddev(response_time) < 150

Raw Query Checks

For validation logic that standard functions cannot express, use a raw SQL query that returns a single value:

SELECT COUNT(*) FROM orders WHERE status = 'pending' AND created_at < NOW() - INTERVAL '24 hours'

The result is compared against a threshold using the same operators as standard checks.

Failure Actions

Each check has an on_fail setting that controls what happens when the check fails:

Action	Behavior
warn	Log a warning; the check is marked as failed but does not trigger error-level alerts
error	Mark as failed and trigger configured alert destinations

---

Creating Checks in the UI

Visual Check Builder

1. Select a dataset -- Navigate to a data store, then select the table you want to monitor
2. Profile the dataset -- Click Profile to see current column statistics and identify potential quality issues
3. Add a check -- Click Add Check and choose:
   • Standard check: Pick a function from the dropdown, select a column (if column-scope), set operator and threshold
   • Raw query: Write custom SQL
4. Add a description -- Describe what the check validates and why
5. Save -- The check runs on the next scheduled validation cycle

Example: Creating a Not Null Check

1. Navigate to your PostgreSQL data store
2. Select the public.users dataset
3. Click Add Check
4. Select function: not_null
5. Select column: email
6. Add description: "Email is required for all users"
7. Click Save

---

Notifications & Alerting

DataBridge supports four alert destination types. You create alert destinations at the organization level and then attach them to individual datasets.

Supported Alert Destinations

Email

Send alert notifications to email addresses when checks fail.

Slack

Post alert messages to a Slack channel via an incoming webhook URL.

Webhook

Send a JSON payload to any HTTP endpoint when checks fail. Useful for integrating with internal tools and custom workflows.

Telegram

Connect a Telegram bot to receive alerts in a Telegram channel or group.

Configuring Alerts

1. Navigate to Alert Destinations and create a new destination (email, Slack webhook, custom webhook, or Telegram)
2. Go to a dataset's detail page
3. Attach the alert destination to the dataset
4. When any check on that dataset fails with on_fail: error, the attached alert destinations receive a notification

---

Data Privacy

What DataBridge accesses:

• Table and column metadata
• Row counts and aggregate statistics
• Sample values for profiling

What DataBridge never stores:

• Full table data
• Credentials in plain text (encrypted at rest)

When using Agent Mode, your data warehouse credentials never leave your infrastructure -- only profiling results and check outcomes are sent to the cloud.

---

Getting Started

1. Request Beta Access -- Join the closed beta
2. Connect a data store -- Add your ClickHouse, PostgreSQL, or MySQL connection
3. Import datasets -- Select which tables to monitor
4. Profile -- Run an initial profile to understand your data
5. Add checks -- Define quality expectations using the check builder
6. Set up alerts -- Attach alert destinations so you are notified on failures

Questions? Email us at hi@databridge.tech

---

Cloud vs CLI Comparison

Feature	Open Source CLI (dbqctl)	Cloud
Pricing	Free forever	Free tier + usage-based
UI	CLI only	Full web dashboard
Scheduling	Cron / Airflow	Built-in scheduler
Alerting	Exit codes + scripts	Built-in (Slack, email, webhook, Telegram)
Databases	PostgreSQL, MySQL, ClickHouse	PostgreSQL, MySQL, ClickHouse
Profiling	One-off profiles	Continuous, historical
Best For	CI/CD, local dev	Production monitoring

Both options use the same core validation engine (dbqcore) and check syntax.