Abstract
Counting Bloom filters (CBF) and their variants are data structures that support membership or multiplicity queries with a low probabilistic error. Yet, they incur a significant memory space overhead when compared to lower bounds as well as to (plain) Bloom filters, which can only represent set membership without removals.This work presents TinyTable, an efficient hash table based algorithm that supports membership queries, removals and multiplicity queries (statistics). TinyTable improves space efficiency by as much as 28% compared to CBF variants and as much as 60% for monitoring flow statistics. When the required false positive rate is smaller than 1%, TinyTable is even slightly more space efficient than (plain) Bloom filters. Our performance study shows that TinyTable has acceptable runtime overheads.
Highlights
Space efficient approximation algorithms extensively used to approximate set membership [29], multi-set frequency [45], state machines [9], and functions [16]
TinyTable can be more space efficient than Bloom filters when the false positive ratio is less than 1%
We further prove that TinyTable can approximate additional metrics such as the distributions’ entropy and the number of distinct items which are useful within the context of network monitoring
Summary
Space efficient approximation algorithms extensively used to approximate set membership [29], multi-set frequency [45], state machines [9], and functions [16]. Bloom filters use a bit map and multiple hash functions to efficiently encode approximate set. The web browsers can store a Bloom filter that requires considerably less space than the complete list. The list servers handle only a fraction of the site accesses while the web browsers store a reasonably sized Bloom filter. TinyTable can be more space efficient than (plain) Bloom filters when the false positive ratio is less than 1%. The updates of TinyTable offer a space/speed tradeoff where we need to slightly increase space to gain faster table operation This tradeoff allows us to suggest multiple configurations that are each attractive for different applications.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
