Filtrowanie adresów sieciowych w sprzętowym systemie bezpieczeństwa typu Firewall

W niniejszym artykule zaprezentowano wyniki praktycznej realizacji sprzętowego klasyfikatora adresów sieciowych opartego o dedykowaną pamięć TCAM (ang. Ternary Content-Addressable Memory). Opracowana metoda implementacji pamięci TCAM charakteryzuje się dużą szybkością pracy oraz znacznie efektywniejszym wykorzystaniem zasobów układów FPGA w porównaniu do komercyjnych wersji oferowanych przez firmę Xilinx.

The paper presents the results of practical realization of a network address and protocol type classifier based on Ternary Content-Addressable Memory (TCAM). The first section deals with a subject of packet classification. The second one describes the packet classifier internal structure, characterizing in details each of the elements included in the classifier, according to the block diagram of Fig. 1. The address filter architecture (shown in Fig. 2) assumed by the authors is discussed in the third section. The fourth section contains some details concerning the TCAM cells array functionality and implementation method. The last section summarizes the results obtained. The new TCAM architecture based on RAM16X1S storage elements adopted by the authors is much more effective than the commercial solution generated by the Xilinx COREGenerator software. The device resources requirements are over two times lower than the resources required by the COREGenerator version. This significant reduction causes improvements in overall timing characteristics. The estimated maximum operating frequency for the address and protocol type filter is 160 MHz. It means that the module can analyze about 160 million packets per second. The research work is in line with the rapidly developing trend towards using reprogrammable logic for securing data transfer in information technology networks.