Specifications

Introduction 1-13

NAM / Traffic Analyzer v3.5 Tutorial

Introduction 1-13

NAM / Traffic Analyzer v3.5 Tutorial

Network Performance Monitoring

What Data to Collect

Network Performance Monitoring

What Data to Collect

• Port level statistics—utilization, collisions, fragments

– Basic physical stats good for usage trending and baselining

– Useful anywhere in the network

– Not necessary for all user ports

• Detailed physical, network, and application layer data

– Collect layers 2-7 statistics for understanding traffic breakdown

– Valuable for WAN aggregation links

– Valuable for LAN aggregation links (building to building, distribution to core,

server farm to core)

• What collection interval?

– Shorter intervals for real-time monitoring and troubleshooting (5–30 sec)

– Longer intervals for historical trending (5 min–15 min)

What Data to Collect, continue …

Port level or interface statistics is the first alarm when issues arise. These statistics are available most of the

time by simply querying the interface MIB. It may only be necessary to monitor these statistics at critical

points in the network and not at all access points.

Collecting statistics at upper layer protocols (network through application) would require the use of an RMON

II probe or analyzer, such as the NAM. RMON II would provide visibility into who is (applications, hosts,

conversations) using valuable WAN or LAN resources at the core or distribution layers and at the WAN edge

or access layer.

Always an important consideration is how often these statistics should be collected. Rule of thumb: if the

data is needed for real-time troubleshooting then the polling frequency should be often; whereas, if the data

is needed for long term trending and placed into a database, then the polling frequency should be shorter

and average over a longer period of time.

Let’s look more in depth at the data collected at the RMON I and RMON II standards.