Connector-level Failover - OPC UA 2.x
Support connector-level failover in PI Connector for OPC UA 2.x (relay-enabled version). Similar to UniInt failover (interfaces), failover should be data source specific and always favor the data source in the better state.
Failover is needed so that maintenance can be performed (Such a windows patching) while maintaining data flow and in case of a server failure or connector application failure
This feature is necessary for the stability and should be standard
This is critical to ensuring the stability of our system.
We would like to have PI Connector, PI DCM and Relay as well to support failover
Uniint failover is important to keep up data availability in case of pi-node downtimes due to failure, maintenance, migration or other reasons. It is required to have a similar mechanism for connectors. In general: PI UA Connector should combine the features of OPC HDA interface and OPC DA interface (especially data handling options; post processing; interactive HR recovery...). Then it would be really useful.
Connector level failover should be a standard feature for every connector, the lack of this feature is really putting us in a tedious situation.
We need this feature!
Kenneth Barber commented
PI Adapters, which are the successors to PI Connectors, don't support failover either. This includes the PI Adapter For OPC UA.
Carl Van Laer commented
Urgent feedback needed from the product owner of this connector - redundancy at connector level is a must have!
How on earth was losing connector level failover ever allowed to go forward? My guess: the developers found it too difficult so persuaded those who should know better that it wasn't needed.
We need this feature! Please implement this important feature. This is essential for a meaningful use in the production area.
Please give feedback on that. Fail-over is also very important for us !!!
High Availability is critical for our business. Particular in regulated industries like pharma it is a must have. I would like to know if fail-over functionality is on the OPC-UA 2.x road map?
We invested in HA solution at server side,
as data capture process needs to be secured end to end OPC UA failover is required.
Redundancy should be available for our mission critical data. End users are expecting high availability.
This is a dealbreaker for us. We won't be using the connector because it's lacking client redundancy.
Right now it seems to be the better solution for us to have an OPC UA to OPC DA bridge and the legacy OPC DA interface, adding another hop but redundancy makes up for that.
Balu (Balasubramaniam) Karthikeyan commented
The legacy SCADA systems are getting replaced by UA servers out-of-the-box. So OPC UA Connector is the preferred route.
Echoing what others have shared already, I think we can all agree with these two points:
1) Failover is a key pillar for the "critical infrastructure" very frequently present in all OSIsoft pitches; and
2) OPC UA is probably the most common standard in industrial communication.
Seeing that the OPC UA connector dropped the connector level failover feature in the new version (plus not having a clear roadmap for this) conveys a not so reassuring message to your customers who decided/take into consideration to engage a long strategic journey with you.
I hope to see this implemented very soon, thanks!
Hi OSIsoft, as others have pointed out, I'm baffled by the apparent backwards step in v2.x (i.e. no HA / failover support at the interface-level).
However I notice that OPC UA has native support for something called "Client Failover" (see the 'OPC UA Online Reference' here: https://reference.opcfoundation.org/v104/Core/docs/Part4/6.6.1/ )
I'm wondering if your intent is to support native OPC "Client Failover" under the new v2.x connector fork?
• You have two diverse PI OPC UA connectors (OPC UA 'clients') – both Active/Active – but one is operating in ‘backup’ mode
• The OPC UA ‘server’ maintains information on it’s active client sessions
• The ‘backup’ UA client monitors the client-session information on the UA server – and detects when the ‘active’ UA client has “failed”
• The ‘backup’ UA client then instructs the UA server to redirect all future traffic to itself
Is that on the cards for support in a future version of v2.x?
From Case:00812539, I need a Failover at the Connector Level just the same as we have for the OPC-DA Interface.
I learned that none of the "new" Connectors are able to have a redundancy on the Connector Level, so every Connector of a current Version is not useable for a Productive Environment in the chemical Production Environment.