[
https://issues.apache.org/jira/browse/LENS-197?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14274690#comment-14274690
]
Amareshwari Sriramadasu commented on LENS-197:
----------------------------------------------
if Cost(F1) < Cost(F2) and F1 is an aggregate fact built on top of F2, then F2
will continue to have the column C through out d1 to d3. If column is in F1 and
F2 in mutual exclusive ranges as given in the example above, then we would pick
raw fact if data is still available, otherwise fail the query as we are doing.
bq. Are we saying if the query time range is [d1, d3] We have to pick up both
F1 and F2?
Not required as part of this jira. But can be done as union query in future.
> Allow start and end times for columns in fact tables
> ----------------------------------------------------
>
> Key: LENS-197
> URL: https://issues.apache.org/jira/browse/LENS-197
> Project: Apache Lens
> Issue Type: Improvement
> Components: cube
> Reporter: Amareshwari Sriramadasu
>
> We can have columns being promoted/demoted from one aggregate fact to other,
> with usage of columns in reports.
> We need this for the following use cases:
> # If field is moved to a lower cost fact: Since the fact is having lower cost
> it will be picked up for processing, but if the life of column queried does
> not fit in the range queried, a higher level fact should be picked up. If
> lower level fact is picked, it would lead to wrong results.
> # If field is moved to a higher cost fact, i.e. its life ending in a lower
> cost fact. Again, depending on the range queried, either lower or higher
> level fact should be picked up. It will be more like optimization for this
> case.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
