ExperimentResultComparison - raphaelvrosa/vnf-bench-model GitHub Wiki

Metrics / Parameter

Both tools select a whole set of different metrics, but to start with a comparison we should select a few to start with. The following table describes the columns (with parameters, e.g., used CPU cores and measured metrics) of the simple result format both tools should produce for comparison.

In this example, vnf0 is the Suricata VNF we agreed to use.

Col. name Unit/Range Description
param_vnf0_cmdstart string Shell command used to start Suricata inside the container
param_vnf0_cpubw float(0-1.0) CPU bandwidth (CPU time) allocated to Suricata container
param_vnf0_cpuset string(list of core IDs) CPU cores assigned to Suricata container
param_vnf0_memmax int(MByte) Memory assigned to Suricata container
param_agent0_cmdstart string Shell command used to start traffic generator in agent container
param_repetition int ID of the experiment repetition
param_exptimelimit int(seconds) Experiment runtime (of a single configuration)
metric_vnf0_eth_input_rx_bytes float(byte) Number of bytes received on the input interface of the Suricata container
metric_vnf0_suricata_bytes float(byte) Number of bytes processed by Suricata see here
metric_vnf0_suricata_dropped float(byte) Number of bytes dropped by Suricata see here
metric_vnf0_suricata_drops float(packets) Number of packets dropped by Suricata by Suricata see here
metric_vnf0_suricata_packets float(packets) Number pf packets processed by Suricata see here

Results (tng-bench)

  • Suricata experiment with changing number of CPU cores and memory (36 configurations x 1 repetition): CSV

  • Suricata experiment with changing CPU bandwidth (60 configuration x 10 repetitions): CSV

  • Simple Plots Jupyter Notebook

Results (Gym)

  • TODO

Comparison

  • ...