Building an Industrial Data Analytics Platform with AWS IoT SiteWise - Equipment Data Collection and Asset Modeling
Collect industrial equipment data through edge gateways, structure it with asset models, and visualize it on dashboards. This article introduces an OPC-UA compatible equipment data analytics platform.
Overview of IoT SiteWise
IoT SiteWise is a service for collecting, organizing, and analyzing industrial equipment data. It gathers sensor data from factory production lines, power plant turbines, building HVAC systems, and other industrial equipment via the OPC-UA protocol, then logically organizes it using asset models. Edge gateways ensure offline resilience, while dashboards provide real-time visualization of equipment operating status. Traditional industrial data collection typically stores each device's data separately in SCADA systems or proprietary databases, but IoT SiteWise consolidates these into a unified cloud platform, enabling cross-equipment analytics.
Asset Models and Dashboards
Asset models define the hierarchical structure of your equipment. You define properties (temperature, pressure, RPM) at each level of the hierarchy: Factory → Line → Equipment → Sensor. Metrics automatically calculate aggregations (average, maximum, count) within time windows, computing equipment utilization rates and OEE (Overall Equipment Effectiveness) in real time. Transforms define formulas between properties (e.g., Celsius to Fahrenheit conversion, difference between two sensor values) and automatically generate derived data. SiteWise Monitor is a no-code dashboard builder that lets on-site operators drag and drop charts and gauges to monitor equipment status. The alarm feature enables notifications when metrics exceed thresholds, sending alerts to personnel via SNS.
Edge Gateways and Data Collection
IoT SiteWise gateways are installed within the factory network to collect sensor data from OPC-UA servers and transmit it to the cloud. Gateways run on AWS IoT Greengrass and buffer data locally during network outages, sending it once connectivity is restored. Data source configuration defines the OPC-UA server endpoint, authentication credentials, and the nodes (tags) to collect. Sampling rates can be set per tag - 10-second intervals for slowly changing data like temperature, and 1-second intervals for high-frequency data like vibration. The service also supports Modbus TCP and EtherNet/IP protocols, enabling data collection from legacy equipment. A single gateway can connect to multiple OPC-UA servers, providing centralized data ingestion from PLCs and DCS systems distributed throughout the factory. For a comprehensive understanding of IoT SiteWise, related books (Amazon) can be a helpful resource.
Design Best Practices and Pitfalls
The most important consideration in asset model design is structuring hierarchies from an analytics perspective rather than simply mirroring physical equipment layout. For example, when comparing identical equipment types across multiple factories, adding an equipment-type hierarchy simplifies aggregation. Metric computation windows that are too short increase computing costs, while windows that are too long delay anomaly detection - use 1-minute, 5-minute, or 1-hour windows based on the use case. A common pitfall is configuring indiscriminate collection of all nodes from an OPC-UA server. When tag counts reach tens of thousands, data point charges escalate rapidly, so selecting only tags needed for analysis is critical. Additionally, gateway buffer capacity has limits, so environments expecting prolonged network outages require pre-calculation of buffer sizes and alert configuration.
Comparison with Alternative Services
The main alternatives to IoT SiteWise are the IoT Core + Timestream combination and Amazon Managed Service for Prometheus. IoT Core + Timestream offers greater flexibility but requires building asset models and OPC-UA integration from scratch, adding development effort for industrial protocol support. IoT SiteWise has native OPC-UA support and built-in asset models, making it deployable in a short timeframe even in environments with limited factory IT staff. Managed Service for Prometheus specializes in metrics monitoring and suits time-series equipment data storage, but lacks logical asset hierarchy structures and no-code dashboards. Compared to on-premises SCADA/MES systems, IoT SiteWise eliminates server management and automates scaling, but retrieving data from existing SCADA requires building OPC-UA bridges. Depending on data ownership and latency requirements, a hybrid architecture (edge for real-time control, cloud for long-term analytics) is often the practical approach.
IoT SiteWise Pricing and Optimization
IoT SiteWise pricing consists of messaging (data point ingestion), data storage, and computing (metric calculations). Data point ingestion costs approximately $2 per million data points. Optimize costs by setting appropriate sampling rates and avoiding redundant transmission of unchanged data. Deadband settings that filter out data where value changes fall below a threshold are also effective. Automatic tiering between hot tier storage (recent data) and cold tier storage (historical data) reduces storage costs. Computing charges are proportional to the number of metrics and calculation frequency, so avoid setting 1-minute windows for use cases where 5-minute windows suffice. Adjust dashboard refresh intervals to match business requirements and reduce unnecessary queries. As a reference, a configuration collecting 50 tags per device from 100 devices at 10-second intervals results in approximately 13 million data points per month (roughly $26).
Summary
IoT SiteWise is an industrial IoT service that collects sensor data from industrial equipment via OPC-UA, logically organizes it with asset models, and provides visualization. Edge gateways ensure offline resilience, while metrics and dashboards enable real-time monitoring of equipment operating status. Analytics-oriented hierarchy design in asset models is important, and tag selection with deadband settings efficiently manages data ingestion costs. Choose between IoT Core + Timestream or SCADA based on OPC-UA requirements and operational team size.