When we are requested to retrieve data from an OSI PI System for long term storage or
further processing/analytics.
Batch files are the
easiest way to get data out of the OSI PI System. Batch files can be created via Excel or VB
using data link connector or via the PI SDK.Batch data exports from OSI can also be generated via the OSI provided
executable. These various components provide
an interface into the PI data archive storage mechanism and exposes tables,
tags and values.
While using batch
files for loading PI Archive data into other systems is reliable it is
cumbersome to maintain, because changes in data model mean changes to scripts,
file formats and client interfaces.
OSI provides a JDBC
interface that a programmatic mechanism for interfacing with the OSI data
archive.However there is very little
information available for performance of the JDBC connection for large
datasets. Since the OSI System is usually considered critical infrastructure
and understanding the performance implication of an interface technique is
necessary for production deployments.
The last option I will call out is OSI's BI Integrator which can forward data from the PI database into Hive periodically. I have no experience with this particular method.
