12.07.2015 Views

Département Réseau, Sécurité et Multimédia Rapport d'Activités 2008

Département Réseau, Sécurité et Multimédia Rapport d'Activités 2008

Département Réseau, Sécurité et Multimédia Rapport d'Activités 2008

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A Fast Adaptative Secure Technology for high-speed N<strong>et</strong>workResearch Staff : Sylvain GombaultKeywords : Traffic filtering, Pack<strong>et</strong> classification, Reconfigurable components, String matchingApplications : Intrusion d<strong>et</strong>ection sensor and Firewall for high speed n<strong>et</strong>workPartners & Funding : IRISA, partially funded by Région Br<strong>et</strong>agne in the framework of PRIR Fastn<strong>et</strong>IntroductionThe development of high-speedcommunication n<strong>et</strong>works led to a growinginterest for the deployment of securitymechanisms related to the specificity of theseenvironments. The Fastn<strong>et</strong> project (FastAdaptative Secure Technology for high-speedNETwork) tackles the problematic of high-raten<strong>et</strong>work traffic analysis: the proposedapproach is to combine software security ruleanalysis and hardware architectures based onreconfigurable components. N<strong>et</strong>work intrusiond<strong>et</strong>ection probe (NIDS) Snort [2] has beenchosen as a use case, and the consideredn<strong>et</strong>work link is Ethern<strong>et</strong> 10 Gb/s, up to 40Gb/s.Fastn<strong>et</strong> is a PRIR (PRoj<strong>et</strong> d’Intérêt Régional)project funded by the Britany Region (2005-<strong>2008</strong>). In this project, TELECOM Br<strong>et</strong>agne isassociated with the R2D2 project (IRISA).RealizationWe carried out a state of the art on thesoftware algorithms used or which could beused in Snort. Based on this work, a firstselection of hardware filtering and stringmatching techniques adapted to the needs ofsnort was proposed [3].We have designed a new hardwareimplementation of a string matching enginebased on a multi-character variant of the wellknownAho-Corasick algorithm [4]. Theproposed architecture model optimally exploitsthe current FPGA reconfigurable components.It combines in an efficient way, the use of thelogic and memory resources of the FPGA. It issuited for searches on very large s<strong>et</strong> of strings(tens of thousand of strings) and the proposeddesign has been validated through theimplementation of a search engine suited tothe processing of a subs<strong>et</strong> of the Snort rules.An Altera Stratix FPGA component is targ<strong>et</strong>ed.By using traffic parallelization and circuitr<strong>et</strong>iming techniques, we then show that 40Gbit/s traffic content scanning can besustained [4].fig. 1 : Preprocessor Polsec in Snort ArchitectureWe have also proposed a new functionality toSnort : a preprocessor dealing with « n<strong>et</strong>workpolicy monitoring ». To ensure the security ofits information system, an orgnisation definesa security policy which applies to all itsequipments. This policy is then translated insecurity mechanisms on each equipment. On<strong>et</strong>ranslation of this policy applies to n<strong>et</strong>worktraffic : it describes allowed and prohibitedtraffic in n<strong>et</strong>work policy rules. With these rules,we can describe both internal andinbound/outbound trafic. The role of IDS is tocheck that no policy violation occurs. As snortrules cannot be used to write n<strong>et</strong>work policyrules, thanks to its flexibility, we havedevelopped a snort software preprocessorbased on policy rules which monitors then<strong>et</strong>work trafic [5].ConclusionAs a future work, we plan to process byhardware all the param<strong>et</strong>ers of a Snort rule.This means that we have to combine pack<strong>et</strong>classification to select pack<strong>et</strong>s into whichlooking for string matching. Pack<strong>et</strong>classification is also useful to monitor n<strong>et</strong>workpolicy and we will see how to implement oursnort preprocessor in hardware.References[1] Alfred V. Aho and Margar<strong>et</strong> J.Corasick.1975, Efficient string matching: An aidto bibliographic search, Communications of the52 Pracom’s Annual Report <strong>2008</strong>

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!