It seems that pivot faceting is prone to the same slowdown in the second phase of distributed faceting. It should be investigated whether this can benefit from the same speed-up trick that simple sparse faceting uses. Even if the terms are related to different filters, it should be possible to speed the individual calls, if the number of terms/filter is high enough.
It seems that pivot faceting is prone to the same slowdown in the second phase of distributed faceting. It should be investigated whether this can benefit from the same speed-up trick that simple sparse faceting uses. Even if the terms are related to different filters, it should be possible to speed the individual calls, if the number of terms/filter is high enough.