On the Vulnerability of Hardware Hash Tables to Sophisticated Attacks - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2012

On the Vulnerability of Hardware Hash Tables to Sophisticated Attacks

Udi Ben-Porat
  • Fonction : Auteur
  • PersonId : 914485
Anat Bremler-Barr
  • Fonction : Auteur
  • PersonId : 914486
Hanoch Levy
  • Fonction : Auteur
  • PersonId : 914487

Résumé

Peacock and Cuckoo hashing schemes are currently the most studied hash implementations for hardware network systems (such as NIDS, Firewalls, etc.). In this work we evaluate their vulnerability to sophisticated complexity Denial of Service (DoS) attacks. We show that an attacker can use insertion of carefully selected keys to hit the Peacock and Cuckoo hashing schemes at their weakest points. For the Peacock Hashing, we show that after the attacker fills up only a fraction (typically 5% − 10%) of the buckets, the table completely loses its ability to handle collisions, causing the discard rate (of new keys) to increase dramatically (100 − 1,800 times higher). For the Cuckoo Hashing, we show an attack that can impose on the system an excessive number of memory accesses and degrade its performance. We analyze the vulnerability of the system as a function of the critical parameters and provide simulations results as well.
Fichier principal
Vignette du fichier
978-3-642-30045-5_11_Chapter.pdf (134.74 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01531139 , version 1 (01-06-2017)

Licence

Paternité

Identifiants

Citer

Udi Ben-Porat, Anat Bremler-Barr, Hanoch Levy, Bernhard Plattner. On the Vulnerability of Hardware Hash Tables to Sophisticated Attacks. 11th International Networking Conference (NETWORKING), May 2012, Prague, Czech Republic. pp.135-148, ⟨10.1007/978-3-642-30045-5_11⟩. ⟨hal-01531139⟩
310 Consultations
370 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More