Impact of Packet Sampling on Link Dimensioning

摘要: Link dimensioning is used by network operators to properly provision the capacity of their links. Proposed methods for link often require statistics, such as traffic variance, that need be calculated from packet-level measurements. In practice, due increasing volume, deploy packet sampling techniques aiming reduce burden monitoring, but little known about how affected this paper, we make use a previously proposed and validated formula requires variance estimate required capacity. We assess impact three on dimensioning, namely, Bernoulli, $n$ -in- $N$ sFlow sampling. To account additional introduced algorithms, propose approaches better sampled data according employed technique. Results show that, depending rate load, does not negatively accuracy even at very short timescales 10 ms. Moreover, also loss inter-arrival time packets exporting process in harm estimations, given an appropriate used. Our study using large dataset consisting traces captured several locations around globe.