Interface Statistics

---

• to diagnose performance issues due to congestions, bottlenecking, bandwidth, or packet loss, you must:
  • collect data and configure alerts for interface statistics

Interface statistics are metrics recorded by a host or switch that enable monitoring of link state, resets, speed, duplex setting, utilization, and error rates.

• Utilization
  • data transferred over a period
  • can be measured as:
    • amount of data traffic send and received
    • calculated as a percentage of available bandwidth
  • differentiate between:
    • average utilization
    • peak utilization
  • e.g.,
    • if around 80% avg. utilization, then bandwidth may seem sufficient
    • but if peak utilization often spikes to 100%, then need to upgrade link
    • monitor queue link can help determine if link is a bottleneck
• per-protocol utilization
  • packet or byte counts for a specific protocol
  • useful to monitor:
    • packet counts
    • bandwidth consumption
  • high packet counts incur processing load on CPU and system memory resources of appliance
    • even if size of each packet is small
• Error rate
  • number of packets per second that cause errors
  • errors may occur as result of:
    • interference
    • poor link quality causing data corruption in frames
  • error rates should be < 1%
  • high error rates indicate:
    1. network media problem
    2. or a driver problem
• Discards/drops
  • interface may discard frames for several reasons
  • each interface will class the type of discard or drop separately
    • assist with troubleshooting precise cause
  • some vendors use
    • “discard” for frames that are rejected for errors or security policies
    • “drop” for frames that are lost due to high load
  • terms are often used interchangeably
• Retransmissions
  • retransmitted data due to errors/discards/drops
  • if high levels of retransmission,
    • analyze and troubleshoot underlying packet loss
      • could involve multiple aspects of network configuration and connectivity