Historics sample rate


#1

If I use a historics sample rate of less than 100%, is it possible for me to miss the only tweet that was published during the time period I specify that matches my CSDL? Or is there another potential factor at play here? I did a search on twitter and found the tweet I was looking for, but using a sample rate of 12.5% my historics query missed it.

Thanks so much.


#2

Yes, this is definitely possible. Running 1.56% or 12.5% samples means that we only search 1.56% or 12.5% of the regions in our Historics cluster, rather than the full 100% for that time period. If the only Tweet matching your filter during that time period is not stored in the regions we search for your query, we will be unable to return that Tweet.