[
https://issues.apache.org/jira/browse/MINIFI-3?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15217176#comment-15217176
]
Joseph Witt commented on MINIFI-3:
----------------------------------
It is important that whatever method is chosen (in band or out of band) that it
allows for the full directed graph of provenance to be determined. If you
encode the provenance of a flowfile into that flowfile this will likely support
going backward to its origin but makes resolving the full graph more difficult
when flows are non linear. I am not saying I am opposed to this in-band (on
the flowfile) approach but I am skeptical of its benefit. It would be
important to describe how you can achieve the full provenance graph for events
following such a model. We know from NiFi how it can work with an out of band
model.
> Provide new approach for provenance generated in agents
> -------------------------------------------------------
>
> Key: MINIFI-3
> URL: https://issues.apache.org/jira/browse/MINIFI-3
> Project: Apache NiFi MiNiFi
> Issue Type: Improvement
> Components: Data Format, Data Transmission
> Reporter: Aldrin Piri
> Fix For: 0.0.1
>
>
> Provenance is handled by the Provenance Reporter and stored in the provenance
> repository. This approach is not ideal for the agent case where the events
> also need to be transmitted to another system or store. Some approaches
> could include the bundling of this with the FlowFIle itself and/or an out of
> band mechanism that sends this information asynchronously and separately from
> the data being transmitted to its destination system.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
