Easy as PI: Integrators
The PI System is a powerhouse for collecting real-time data, contextualizing it, and storing it. But what if you want to take things a step further?
Enter PI Integrators.
John Baier, OSIsoft's director of solution strategy, and Landry Khounlavong, product marketing manager, are both intimately acquainted with the in's and out's of PI Integrators, making them the perfect people to interview about what they are, why we made them, and where they're headed.
What are PI Integrators?
“In short, it's a data preparation tool that allows you to easily send operations data to advanced analytics platforms,” says Khounlavong, who has been at OSIsoft since 2010 and has held numerous positions in technical support, cloud support strategy, and product marketing. Integrators were “built to speak the language of higher analytics,” according to Baier, who has directed OSIsoft's engineering and product management organizations, “they translate OT to IT.”
Why did we create them?
Many products in the analytics space don't have a way to query time-series data, such as what's in the PI System, and data preparation takes too much time for operations people to be involved in the process. Converting time-series data into something that analytics tools can work with is a time-consuming, labor-intensive process, so we created PI Integrators to convert that information quickly and easily, letting the customer get back to what matters most, analyzing the data, outlines Baier.
Another major reason - it allows customers to send contextual data, including asset and event context, so there's an enriched data set going to these external platforms. “If you send raw data from an industrial operation without context to a data lake, it's going to look completely foreign to a data scientist, because they won't get the naming and nuances of the operation,” says Khounlavong, highlighting a Gartner article about the common ways data lakes fail.
How does the underlying technology work?
“Most of the product is in the base layer,” says Baier, “it takes time-series data and makes it conform to a row/column format while enriching it with context” while following data cleansing rules that the customer can set up. There are also adaptable pieces that work on top of this base layer to natively connect and send the enriched dataset to different data lakes, data warehouses, and messaging hubs.
How is it changing the way our customers do business?
“The most salient value proposition is saving time,” says Khounlavong. The people who work in operations understand the data they use day to day, so they're often the ones asked to prepare it. If they do it on their own, they're likely using spreadsheets or may even write code - with the PI Integrator, it's a drag-and-drop, web-based interface with native integration to various advanced analytics platforms. With PI Integrators, data can be uploaded in bulk, in scheduled loads, or streamed live.
What platforms do PI Integrators work with?
AWS, ESRI ArcGIS, Hadoop, Microsoft Azure, and SAP HANA. In addition, PI Integrator can send to destinations such as Microsoft SQL Server and Oracle Database. A full list can be found here, and if there's a destination we don't support, we can build support for it if there's enough need.
What updates are on the horizon for PI Integrators?
Very recently, the PI Integrator has been enhanced with more destinations that it can integrate to. The recent version adds the ability for customers to send data from their PI Systems to Google Cloud Storage, BigQuery, and Pub/Sub as well as SAP HANA (via ODBC). Beyond that, “we continue to update how we support existing destinations since these are native integrations to live-and-evolving advanced analytics platforms, so as other platforms update their services, we're updating as well,” says Khounlavong. "And, we are always listening to customer feedback on new destinations and feature requests via https://feedback.osisoft.com."