Glossaire des techniques de mesure de la performance réseau

Second of a series of articles aiming to report on the measurement techniques available to companies, service providers and used to monitor the network in real time. These techniques make it possible to guarantee, monitor and in-fineoptimize LAN, cloud and WAN infrastructures. These techniques are gathered under the acronym NPM (Network Performance Monitoring).

They can be used to:

  • Supervise Availability/ downtime of IT networks: LAN, WAN, WIFI, cloud, VPN, SDWAN, direct route
  • Detect 24-hour network problems and outages
  • Assist administrators in resolving concerns about slowdowns, bandwidth load
  • Prevent the impacts on the user, manage the evolution of uses
  • Establish network management indicators, and ensure the SLA of network links

For each measurement technique, a distinction is made between the measurement carried out and its objective, the means, the cost and the implementation requirements.


The supplier’s dashboards

This technique is based solely on consulting the dashboards provided by the network operator.

  • Advantage: no instrumentation to set up
  • Disadvantage: not necessarily in real time and gives the view from the supplier’s point of view most of the time
  • Interpretation of results: delivery manager, network manager
  • Cost: Nile


Reporting integrated with network equipment / voice solution / cloud proxy solution

This technique is based solely on the consultation of reports available in the network equipment WAN, LAN, … / voice solution / cloud proxy solution

  • Advantage: no additional instrumentation to set up for performance analysis
  • Disadvantage: reporting with low granularity and history. Not available 24 hours a day if equipment unreachable. No centralized platform.
  • Interpretation of results: network team
  • Cost: $


Collection of standardized passive measurements (SNMP, telemetry, NTA traffic analysis)

This measurement technique aims to collect indicators present on the network elements. It can be implemented via configuration on the equipment. It provides a technical view and requires a collector to centralize and visualize the information.

  • SNMP collection: configuration of a collector (poller) to centralize the information.
  • Telemetry: configuration of a collector to centralize the meters exported by the network elements.
  • Traffic analysis: configuration of a collector to centralize the information exported by the network elements with the Netflow, IPflow, JFlow, SFlow protocols. This technique provides additional visibility into the usage of network links via more detailed reports.
  • Advantage: widespread and well supported by suppliers
  • Disadvantage: necessary configuration (cost) on the operator side for the WAN equipment. Requires the use of a collector and its setting/maintenance. Large storage for telemetry and traffic analysis.
  • Analysis of results: network team
  • Cost: $ $


Collection of proprietary passive measurements (cut-off, TAP, flow gas pedal)

This measurement technique aims at collecting indicators directly from the network flows. It can be implemented through the deployment of equipment. It provides a technical vision.

  • Flow gas pedal: deployment of a box that continuously analyzes all network flows in transit and sends them or not to the WAN to a site equipped with the same type of box. This same DPI technique is used by security tools like IDS.
  • TAP box: deployment of a box on a network port (Ethernet or fiber) which will duplicate the transiting network flows and send the information to a collector.
  • Advantage: visibility into the packages (DPI)
  • Disadvantage: deployment of a proprietary box per site.
  • Analysis of results: network team
  • Cost : $$


Active measures (synthetic)

This measurement technique aims to collect a regular sample of measurements from one or more points of the network, placed as a user. It uses dedicated agents or robots that will use typical techniques (icmp, tcp connect, Cisco IP SLA, Juniper RPM probe, One Access PPA PM or Twamp) to deduce metrics such as network delay, jitter or packet loss between a transmitting device and a remote device

These technologies are the ones used by operators both to detect failures and to perform end-of-month reporting. They are also used for SDWAN link failover detection, for example.

  • Advantage: factual, continuous visibility from specific locations that simplify analysis. Very wide coverage.
  • Disadvantage: linked to equipment brands
  • Analysis of results: support, service management, IT and network team
  • Cost: $ $
  • Share:
Send a Message