Sampling time-dependent parameters in high-speed network monitoring
Edurne Izkue, Eduardo Magaña · 2006
Nowadays network bandwidth is increasing continuously for end-users and network providers. Network monitoring tools have to be able to support these high-speed networks, processing a high number of packets per second. For this reason, network monitoring tools have to be improved using software or hardware techniques. In this paper we use sampling techniques, as a software technique that can provide the capacity of monitoring networks with high bandwidth, keeping a low-cost hardware platform. A new technique called mixed sampling is proposed and compared with other techniques proposed in the literature. We focus on time-dependent statistics not studied before with sampling. These statistics will require applying estimation techniques for better results