Network Performance Management

The network has become the lifeblood of most enterprises. As the world adopts the Internet Protocol as the primary means to conduct business and to share data with customers, partners, and employees, more and more of your work day revolves around your ability to access network resources. Pressure is increasing on IT Operations teams to ensure that employees can successfully execute their revenue generating activities. The result is that downtime is not accepted and latency in the user experience causes immediate widespread pain.

Best Practice Recommendations

The top 10 attributes of effective network performance management to help meet customer’s demands and ensure the performance of your IT infrastructure:

  1. Helps avoid service interruptions with proactive alerts.
  2. Gives you system wide visibility in real-time.
  3. Provides a complete view of the raw data with no roll ups or averaging.
  4. Has open access to data via a web-based console and portals.
  5. Provides easy integration with other tools for effective correlation and ticketing/help desk services
  6. Is easy to use and deploy quickly providing “quick time to value”  
  7. Has flexible collection technology support  (from SNMP, NetFlow, and IP SLA to vCenter API)
  8. Calculates accurate baselines for establishing the normal levels of utilization.
  9. Alerts the user to deviations from normal levels
  10. Gives easy access to your data through a standards compliant open API

In summary, effective network performance management must help avoid problems that can be detected with thresholds, it must give me a real-time view into my data when troubleshooting and it must give me a direct route to resolve what type of traffic was associated with an event with NetFlow reporting in a single click.   

"Clients should look for network performance management products that not only track performance, but also automatically establish a baseline measurement of "normal" behavior for time of day and day of week, dynamically set warning and critical thresholds as standard deviations off the baseline, and notify the network manager only when an exception condition occurs. A simple static threshold based on an industry average or a "rule of thumb" will generate false alarms. Clients who are looking for the utmost efficiency should link network performance management processes to network CM processes so that bandwidth allocation and traffic prioritization settings are automatically updated based on changing business demands and SLAs."

Source: Gartner's Hype Cycle July 2010