Jim Cahill recently had a post about migrating 10 years of historical process data from one historian to another.   Yet another application that can make use of OPC HDA.  OPC Historical Data Access – migrating historical data… who’da thunk it?

I’ve done a lot of database migrations, and having the right tools to get the data out of one proprietary format into another was always a challenge.  Developing a custom tool always had risks and even using flat files had it’s pitfalls  (text wrapping, file size limitations, speed of transfer… the fun never ended).  A well built OPC HDA server cures a lot of ills, and you can get them for pretty much any major historian on the market.

The basics of migrating data between historians boils down to reading blocks of history from one and writing blocks of history into the other.  OPC HDA is used to move history in many architectures, such as hub and spoke, guaranteed data delivery and event capture.  Historical process data migration is a natural fit.  Features of OPC HDA such as ReadRaw with bounding values ensure you don’t miss data, and the Insert, Replace and InsertReplace functionality gives the tools control over updating the data in the target historian.

Moving 10 years of data is not a trivial task, regardless of what tools you’re using.  Jim outlined a lot of the key project steps;  design, configuration, data migration, testing, training etc.   An important step he didn’t highlight was data quality auditing.  Identifying things such as accuracy, integrity, consistency, completeness, and validity.  In other words  Did you get everything that was in Historian A into Historian B, and is it EXACTLY the same?    OPC HDA helps out here too with the standard aggregate functions.  You can quickly check data integrity by running the same standard aggregates like TimeAverage, Minimum, Maximum, etc on both data sets.

Migrating historian databases is not just the result of upgrading legacy equipment.   Changing or standardizing on a vendor due to company mergers often results in historian changes.   Of course if your data visualization and reporting tools had been OPC based in the first place, having multiple historians is less of an issue.