Details
-
Architectural Decision
-
Resolution: Done
-
None
-
None
Description
Why now? As we are starting to actually get ingest workflows in the MvP for PSS+PST, it seems like the right time to re-visit the decision of what protocol we are going to use here (at least initially).
Scope? SDP, PSS, (PST?)
Qualities? Maintainability, Testability, Portability, minor performance
In contrast to the stream-like visibility receive, for both PSS and PST ingest involves sending file-like blocks of data. The data rates are not extremely high (compared with visibility ingest) and we would like to guarantee delivery.
Furthermore receive involves little more than writing those files to disk for either post-processing or data product delivery. Possibly some signal displays, but it doesn't seem like we will be forced to hook directly into the receive process for this.
Therefore it seems like the decision should be primarily about selecting a simple solution that we can maintain and test easily.