With all refresh issues I'd recommend using what's known as the "Cartesian flush", which basically means adding a cartesian ("Give me everything") join to the relationship(s) on which the portal(s) depend. Daniel of Digital Fusion in New Zealand explains the technique very clearly , and explains why unnecessary flushing of the cache is such a no-no, especially in a WAN setup.
I use this all the time now, and it's solved many issues.
Thx for the aricle, considering the chearfull comments, that should do the trick.
Testing scheduled for this evening ;-)
To be continued,
with kind regards,
It took me a couple of years to find that bottleneck which slowed down performance. After some changes, performance increased by about factor 6!
Avoid cartasians (table1::field1 x table2::field2) and comparative-operators (< > ≤ ≥) at any costs!!!
Have a field "rel_1" in every table as a formula:
rel_1 = 1 [number, indexed].
Though you can build a "Cartasian" like:
table1::rel_1 = Table2::rel_1
At this verry moment, those are my most valuable tipps regarding FileMaker performance
efficientbizz - I don't understand what you're saying - surely not that catesians and other comparitives should be avoided always?
If possible, try to setup some tests with both relational-models, then compare performance.
I've rebuilt all cartasians with '=' relations based on fields 'rel_1'.
More significiant the performance-increase when avoiding comparative-operators.
This discussion has been initiated by HOnza: