Extending RAM-Based TCAM with Range and Update

• Main Authors: Tzu-Chieh,Lin, 林子傑
• Other Authors: Yeim-Kuan Chang
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/ejcayn

碩士 === 國立成功大學 === 資訊工程學系 === 107 === Packet classification is an important mechanism in network infrastructure today, and it is commonly used in network applications such as firewall, intrusion detectors or Quality of Service (QOS). Ternary Content Addressable Memory (TCAM) is widely used in some hardware-based solutions for packet classification. Traditional ASIC-based TCAMs are expensive and unscalable. In addition, this type of TCAM usually cannot store rules with range fields. Although some range encoding methods can solve this problem, it will result in low TCAM utilization. Compare with ASIC, FPGA are reconfigurable and can be deployed fast. RAM-based TCAM have been proposed and implemented on FPGA. However, rules with range fields still cannot be stored directly in the RAM-based TCAM. In this thesis, we proposed the bound-match method to solve the range-match problem. Compare with the RAM-based TCAM, we use additional RAMs and circuit logic to perform bound-match. Besides, we also add the design that is suitable for traditional TCAM update scheme into our system. According to the implementation results on Xilinx Virtex-7 XC7V2000T, the proposed system can support more than 12000 5-tuple rules. Compare with the RAM-based TCAM, the clock rate of the system is not decrease too much by adopted the pipelined design. We also estimate the clock cycles needed to complete the update process of a newly inserted rule.