414 Large Scale and Big Data
tools catering to massive data analytics, paved the path for efcient implementa-
tion of parallel Bloom lters and their variants. For a single data element, all the bit
positions of the Bloom lters could be accessed simultaneously, giving enormous
performance gains. The two approaches discussed in this chapter can trivially be
implemented in a parallel system.
13.3 STABLE BLOOM FILTER (SBF)
In data stream applications, as more and more elements arrive, the fraction of zeros
in the Bloom lter decreases continuously and the false-positive rate increases
nally reaching 1. Usual solutions employ random eviction policies for bits from the
Bloom lters to accommodate the new elements and keep the FPR at bay. An i ...