CiscoWorks Service Level Manager

Performance Modeling

Document ID: 15182

Updated: Oct 01, 2009


Related Information
Related Cisco Support Community Discussions

Q: How is routing update overhead calculated in the Performance Service Manager?

A: Routing update overhead considers update packet size and update interval for distance vectored protocols and hello interval for link state protocols. It is calculated during data collection.

Q: How is average delay calculated in the Performance Service Manager?

A: Average delay is averaged over all possible paths from end system to end system.

The following factors are used to calculate average delay:

  • delay caused by FIFO-buffer-type queuing

  • user-configured delay on a given interface

  • switching-based delay

The Average Delay report reflects a one-way delay from the source to the destination, and will only report on the 100 worst (highest) values.

Q: When calculating delay caused by an access list, does Netsys Service Manager (NSM) look at the order of the statements in the access list?

A: The length of the access list on an interface is taken into account when formulating the delay factor. An on-average delay factor is taken into account, independently of where the rule occurs in the list.

Q: How do I take into consideration the effect of traffic that is not included in the baseline model on links and routers that are in the model?

A: Traffic that has a source or destination end system (or both) that is not in the baseline model is displayed in the Remote Networks Report. To consider the effects of such traffic on the existing model, edit that traffic so it references two known end systems which would most closely follow the path of the original unmapped traffic.

To accomplish this:

  1. Display the Communicating Pairs Report and filter for the Remote Network. This gives you a list of traffic that originates or ends from an unknown end system.

  2. Sort the resulting report by source address.

  3. Select as many entries as possible that can be mapped into any one known subnet.

  4. Click on the "Edit" button.

  5. The "Task Manager" window appears with the Edit Traffic task highlighted.

  6. Follow the task steps to change the IP address of the selected report entries.

  7. Repeat steps 3 through 6 for all unmapped traffic that you want considered.

Q: Why does changing burst rate not have any effect on performance statistics? In this case, I want to change the committed information rate (CIR) on a Rrame Relay circuit and see the effect.

A: The Performance Service Manager computes utilization and delay at steady state. The resulting packet rate on a Frame Relay circuit will be some (constant) number, and utilization will be relative to bandwidth. In this situation, changing CIR is equivalent to changing bandwidth, and utilization will vary inversely with bandwidth.

Because NSM does not model for transient analysis, you have to model both the "average" as well as the "worst" case in different scenarios to see whether CIR is exceeded.

NSM will provide information with regards to an average, whether you are within your subscription, and how much is available for bursts. It is recommended that this value be used as a measure of how much freedom you have for bursting rather than to change the traffic inputs and recompute the "averages."

Q: Does NSM take into account the impact of stack compression when configured on a router?

A: NSM does not currently model the effect of stack compression when configured on a router.

Related Information

Updated: Oct 01, 2009
Document ID: 15182