The new data structure, called the Resilient and Efficient data Structure (REX), aims to improve widely-used hash tables.
"A hash table maps values to specific locations, labeled with indices. To find a value, the hash table performs computations to quickly identify the indices and thus, its location. The challenges are that millions of values need to be stored, and the values are generated and transmitted extremely quickly," said Vrizlynn Thing from the A*STAR's Institute for Infocomm Research, who led the study.
Traditional hash tables are also becoming inefficient as data flows get larger. As such, these hash tables fill up quickly when they are under attack, causing network performance to erode.
REX thus aims to address this through the use of both the characteristics of internet traffic and different memory technologies to track data flows.
For example, REX looks through heavy-tail data flow behaviours through the use of sub-table hierarchies that increase in size from top to bottom. It also uses Static RAM (SRAM) to process large, important data flows while Dynamic RAM (DRAM) for low priority data flows.
A*STAR claimed that REX has outperformed data structures, Cuckoo and Peacock, in analysing data in a real, recorded network traffic. However, Thing said that they will still "investigate the efficiency and scalability of this new data structure for security analysis in larger scale environments."
Sign up for Computerworld eNewsletters.