Skip to main content

General

Data Quality & Metrics

Understanding the health, freshness, and quality of your industrial data streams in Proxus.

Industrial data quality is crucial for reliable operations. Proxus provides comprehensive data governance metrics to ensure your dashboards, alerts, and analytics are based on trustworthy data.

The 5 Dimensions of Data Quality

Proxus evaluates data quality using five orthogonal dimensions, each scored from 0-100%, providing a holistic view of data trustworthiness:

1. Data Accuracy

Percentage of successful complete cycles where operations succeed AND all expected data is returned. This measures the reliability of the data acquisition process.

2. Data Completeness

Percentage of expected data items that are actually received. This tracks whether all expected values are present in each data collection cycle.

3. Data Integrity

Percentage of received data that passes validation checks. This ensures the data content is valid and not corrupted during transmission.

4. Data Consistency

Rolling window consistency based on recent success/failure patterns. This measures the stability of data delivery over time.

5. Data Freshness

Time-based freshness score calculated from how recently data was received. This ensures you're working with current information.

Device Diagnostic Metrics

Every connected device (via Fieldbus, MQTT, or OPC-UA protocols) maintains detailed diagnostic metrics through the ProxusActorMetrics structure:

MetricDescriptionPurpose
Total RequestsTotal number of communication attemptsOverall activity measurement
Successful RequestsNumber of successful communicationsSuccess rate indicator
Failed RequestsNumber of failed communicationsError tracking
Last Communication TimeTimestamp of last communication attemptActivity awareness
Last Data Received TimeTimestamp of last successful data receptionFreshness calculation basis
Total Data ReceivedVolume of data (bytes) received from deviceThroughput measurement
Average Response TimeAverage response time in millisecondsPerformance indicator
Connection Error CountNumber of connection-related errorsNetwork stability indicator
Read Error CountNumber of read operation errorsData acquisition health

Freshness & Stale Data Detection

Proxus automatically tracks data freshness using time-based interpolation against predefined thresholds:

  • ≤ 30 seconds: 100% freshness
  • ≤ 1 minute: 95% freshness
  • ≤ 2 minutes: 90% freshness
  • ≤ 5 minutes: 80% freshness
  • ≤ 15 minutes: 60% freshness
  • ≤ 1 hour: 40% freshness
  • ≤ 12 hours: 10% freshness
  • > 24 hours: 0% freshness (completely stale)

The system calculates freshness by comparing DateTime.UtcNow against the LastDataReceivedTime and LastCommunicationTime, with data older than 24 hours considered completely stale.

Quality Propagation & Offline Handling

When a device disconnects, Proxus automatically updates the metrics to reflect the disconnection. The DataFreshness score decreases over time based on the predefined thresholds, ensuring that disconnected devices are properly identified as providing stale data.

The system does not emit automatic "Bad" quality packets for tags when offline, but the freshness score provides clear visibility into data staleness.

lightbulb
Quality in Rule Engine

Your Rule Engine logic can be configured to consider data quality metrics when triggering alerts. Rules can be designed to ignore low-quality data or adjust sensitivity based on quality scores to prevent false alarms from stale or inconsistent data.

System Health & Internal Metrics

The platform maintains comprehensive health metrics through the actor system:

  • Actor Statistics: Each device actor maintains its own metrics instance
  • Thread Safety: All metrics updates are synchronized using device-specific locks
  • Accumulators: Internal tracking of rolling statistics
  • Memory Management: Efficient pooling and management for performance

These metrics are available through the UI and API, providing real-time visibility into system health and data quality across your entire infrastructure. You can view per-device metrics in Device Details or troubleshoot missing data in Data Flow Problems.