There are two key pieces of information learned through response time testing:
- identifying the normal response time for a specific application over a specific path
- whether a specific path is still operational
Both are important to solving problems once they happen. (See the “Know Thy Network - Baseline. Don't Wait Until a Problem Strikes” tech tip.)
Most response time testing is performed at the Network Layer (such as the ubiquitous “ping”); however, did you know that some functionality has been defined for the Data Link Layer? The IEEE 802.2 LLC defines “XID Command PDU and Response PDU” frames (referred to here as another form of ping) for the LLC sublayer. This type of ping is effective only within a broadcast domain. Unfortunately, most Ethernet products don’t appear to respond to the LLC ping frame even though 802.2 requires it in Clause 6.6. So much for standards!
While a layer 2 ping may have its uses, it is more important to test response time at the Network Layer than at the Data Link Layer anyway because Network Layer traffic crosses the end-to-end path. If a problem exists, this will be revealed quickly without taking time to test each segment separately, including WAN links. As the path between stations becomes more complex, greater variations in the response time can be expected. Knowing the normal amount of time it takes to cross a specific path, or whether a specific path is still operational, is vital to network operations staff. Without this information, it is extremely difficult to locate the source of network slowdowns and failures.
- Network Maintenance and Troubleshooting Guide. Second Edition
- Cisco – Measuring IP Network Performance