Testing and analyzing network performance

Application speed and user experience depending on network speed, delays, and various interference factors. Challenges are often generously accused of ‘networking’. Really persistent updates to applications, updates to infrastructure, security-related deviations, and intermittent interaction of rogue applications that have appeared on the network cause problems online. In addition, the SD-WAN service layer, intelligent routing, and various tunneling techniques also present additional challenges for troubleshooting. Finding them is challenging and requires expertise and specialized tools.

Testing and analyzing the performance of a corporate network gives the company a good understanding of network capacity, speed, and problems affecting them. The analysis uses special tools to enable quick and accurate analysis as the basis for network development, validation of new network technologies, and assurance of the capacity and prioritization promised by the service operator (SLA and QoS). Analysis is based on packet data.

Packet data vs Netflow (or sflow)

Packet data has been the gold standard for network visibility ever since the first packet analyzer was invented. The common expression around network analysts is, “packets never lie.” Packets show what is really happening on the wire, detailing all the protocols, conversations, and timing involved in network communications. Collecting packets and getting the real wireline story enables network engineers to resolve problems that would otherwise live on in the system indefinitely.

Right analysis tools get more value from your network. Packets provide an unprecedented level of visibility, filling gaps that other technologies, such as NetFlow, leave wide open.

Phase 1

The overview gives a good idea of the quality of online services. This is recommended to be done before and again switching network technology and / or supplier. The first step of the analysis gives the following

User experience (max and average)

  • TOP20 weakest response times for users (latency and roundtrip)
  • TOP20 weakest servers (latency and roundtrip)

TOP 20 Cloud services and country 

  • Global map  (Client traffic direction)
  • Where content is really coming

TOP 20 network capacity spent(max and average)

  • Top clients  (Mbps)
  • Top servers  (Mbps)
  • WAN link  (Mbps)

Name server quality (max and average)

  • DNS servers and no of queries
  • DNS servers and reply codes

TCP-protocol (max and average)

  • TOP20 servers (out of order packets – errors)
  • TOP20 servers (packet retransmission – errors)
  • TOP protocols/applications

Figure: Example MPLS WAN office user experience before optimization. (red is a very poor service experience and dark green is excellent).

The second stage of analysis and optimization

The analysis will be continued at the service level to the protocol level, where appropriate, and efforts will be made to optimize the entity together with stakeholders.