I started to dive into the linked tables in AF and found some very powerful uses. We have a lot of OLEDB compliant data in RDBMS and linking these tables provides the ability to use this data for notification criteria and a more powerful approach for analysis. I also viewed the data in processbook because my linked table field attributes were using a PI tag for filtering criteria. This also has numerous benefits for providing a much more detailed means of visualization.
I do have questions though concerning the cache updating and how this will effect visualization when needing to go back months in time and needing several thousands of records.
It appears that the way linked tables work is that the data is cached in main memory and when it is refreshed, it will refresh all records. This limits the amount of data in the cached set to the desired amount of memory usage but more importantly, the efficiency of retrieving the data is significantly impacted.
In many of my cases, I would like to have a significant amount of data cached or possibly even stored locally but I do need to refresh a small amount of the current data and retrieve new records. What I am saying is that it would be nice to have just a small amount of refreshing to be added to a cache instead of having to refresh the entire cache. Is this possible or is this something that is under consideration?