env: INO414, java 14*
SUN SOLARIS 64 bit, but it shouldn’t
matter.
problem: as the original documents are very large, sometimes about 300MB, a node_level_update to only one attribute acting as a flag will take very long and therefore will affect the overall performance. to overcome this situation i am thinking about to create/maintain a status_info_doc, also used sometimes for fast look aside access.
solution i am planning:
1. trigger onInsert will create the add_on document. but i have some questions:
in the documentation is mentioned the trigger is called before the validation phase: what does this mean? what happens, if the validation phase fails? is the process synchron or asynchron? can i access node content of the document as i need some information from the original document?
2. trigger onDelete will delete the add_on document to keep the information in sync. is the trigger part of the transaction concept in this case?
3. update the add_on document when doing an extract to an application like read the original document and update an flag, that this document was read. is there any Xtension/trigger available or must this be done by the application?
4. what about joins in tamino in this case: is there any optimisation possible?
5. how secure is the idea to use the ino:id of the original document inside the add_on document as the unique index information. this information is provided by the trigger functionality?
i know a lot of stuff but looking forward to some tips/hints/tricks
regards
MichaelK