Information Technology Reference
In-Depth Information
to remove them. This case study focuses on providing empirical data to support the
convergence-tuning recommendations provided earlier in this chapter.
Test Scenario
The equipment used in performing the convergence testing is
Cisco 7206VXR
NPE 300 network processor
256 MB DRAM
Two versions of Cisco IOS are used in the testing. All the tests except the update enhance-
ment test are performed using Release 12.0(15)S1. The update enhancement test is per-
formed using Release 12.0(23)S. This change in code is required for update packing and
BGP read-only mode.
The Unit Under Test (UUT) is configured with a single eBGP peering session to supply the
prefix information. This information is then sent to 50 iBGP peers. The prefix count ranges
from 70,000 to 140,000 in 10,000-prefix increments.
The BGP tables used are specially built to represent the standard Internet routing tables in
prefix length distribution and attribute combination distribution. The same BGP table is
used for each test for a specific prefix count.
The BGP process on the UUT is said to have converged when the BGP table version for all
peers is equal to the BGP table version for the router and the BGP InQ and OutQ for every
peer is 0. The output from show ip bgp summary is used to determine when a converged
state has been reached. The convergence time for the UUT is the amount of time from the
first BGP peering session's initiating until the BGP router reaches a converged state.
Baseline Convergence
The baseline convergence is performed using a default configuration. The BGP feature
status is as follows:
Peer groups are not enabled.
Path MTU Discovery is disabled (TCP MSS 536).
The input hold queue is 75.
Update packing is not supported.
Figure 3-7 shows the results of the testing.
Search WWH ::




Custom Search