If I understand the documentation correctly, the PI Interface for Relational Databases (RDBMS) does not have a high availability configuration. If I'm incorrect, I'd love for someone to send through some information to advise how this could be set up.
If my understanding is correct though, has anyone attempted to achieve this using a bespoke/customized installation? I was thinking about configuring two RDBMS interfaces to different PI Data Points and then using PI AF/PI Analytics to filter any stale or bad data values to provide a good result (kind of like a 1oo2 voting architecture), but this seems a tad complex.
Has anyone played with something like this? Any plans for this feature in future versions of this interface?
To attempt to answer some of the "why would you want this?" kind of questions:
- this is a feature being asked for by a client
- the client uses a RDBMS interface to bring their LIMS data into PI
- if the LIMS data is updated, but this data is not uploaded into the PI Data Archive due to come problem with the RDBMS PI Interface, there is significant potential for economic loss. The normal buffering and recovering historical data is insufficient for this application.