A Bloom Filter is a simple space-efficient randomized data structure for representing a set in order to support membership queries

which uses an m-bit array to represent a data set and queries by hashing.The representation is the payoff for allowing a small rate of false positives in membership queries;that is

queries might wrongly regard an element as member of the set.However

for many applications

especially large-scale data set systems

the space savings and the locate time constantly outweigh this drawback when the probability of an error is sufficiently low and can suffer from by the application.The paper firstly surveys Bloom Filter and its variants in detail

and gives mathematical analysis behind them about space/time/error rate tradeoffs in order to explain their practicability.Then

we present a new kind of Bloom Filter-Split Bloom Filter

which uses a s×m-bit matrix to represent a set

and give analysis in detail as the formers.In distributed systems

each network node owns a data set

which is reasonable that some nodes are large number of data while the large number of nodes are a bit of data

if all nodes uses same parameters of algorithm

then it will give rise to a memory space wholly wasted.In addition

if the number of the elements of a node date set increases continually

the error rate will increasingly make the representation nonsensically.We prove that the Split Bloom Filter can efficiently solve or weaken the two problems.