Bidirectional filtering or crossfilter?

[HxH]

Hey guys,

I am working on a model that has a lot of bidirectional relationships and I remembered reading from SQLBI and other sources that bidirectional filters are bad and that it's better to use crossfilter in measures instead. What I wanted to ask is if there's any reason I should keep bidirectional relationships enabled and not use crossfilter every time. What I mean is: what are the situations in which you absolutely can't do without bidirectional relationships (maybe because crossfilter doesn't work in that particular instance) ? Do these situations exist? Or I can just eliminate them and use crossfilter? This is probably a stupid question but I'm a beginner so this is all really new to me. 

Thanks 

[parry2k]

@HxH keeping bi-directions will have performance impact and sometime you will find unexpected results because of bi-directional filter, if you use crossfilter in measure, bi-directional will only applicable where you will use that measure otherwise it will not have perfomance impact. Hope it helps.

[HxH]

Yes I got that but I'm trying to understand if there's any case in which I can't use crossfilter because otherwise it seems like the best option and bidirectional filters seem useless

[parry2k · Accepted solution]

@HxH I haven't came across any scenario where I couldn't use crossfilter so based on that I think it is the best option, not sure what else to tell.