Testing Routing in a WAN Environment

Follow

Summary

This knowledgebase article contains a series of tests that should be performed prior to starting the Neverfail Heartbeat install. If any of the following tests fail then the Customer’s Network Engineers should be consulted and asked to rectify the routing.


More Information

The tests bellow assume a Primary - Secondary configuration, but similar tests can be performed for Secondary - Tertiary and Primary - Tertiary in a server trio.

Determine the server IP configuration

From a command prompt, run:

ipconfig /all

Routing tests

From the Primary Server console perform the following two tests:

  • Test 1 – Primary Channel to Secondary Channel connectivity test:

    ping <Secondary Channel IP Address>
  • Test 2 - Primary Channel to Secondary Channel routing test:

    pathping –n <Secondary Channel IP Address>

Note: The –n flag instructs pathping not to perform name look up as this will increase the time to return results.

Example:

Tracing route to 192.168.28.1 over a maximum of 30 hops

0 192.168.27.1
1 192.168.25.254
2 192.168.28.1

Computing statistics for 50 seconds...
Source to Here   This Node/Link
Hop  RTT    Lost/Sent = Pct  Lost/Sent = Pct  Address
0                                           192.168.27.1
0/ 100 =  0%   |
1    0ms     0/ 100 =  0%     0/ 100 =  0%  192.168.25.254
2    0ms     1/ 100 = 1%      0/ 100 = 0%   192.168.28.1

From the Secondary Server console perform the following two tests:

  • Test 3 – Secondary Channel to Primary Channel connectivity test:

    ping <Primary Channel IP Address>
  • Test 4 - Secondary Channel to Primary Channel routing test:

    pathping –n <Primary Channel IP Address>

Analysis

  • Tests 1 & 3

    These tests are straightforward to identify success. Check that response times are in the order of <10ms. Times in excess of 100ms suggest undue latency on the channel network and should be investigated.
  • Tests 3 & 4

These are analyzed in two parts:

  1. The addresses returned along the path should be checked against the IP configuration for the server.
  2. The statistics returned for the journey itself should be analyzed.

Path Analysis

The first IP address (identified by the ID zero) returned must be the channel IP address. If it is the public network IP address, then packets destined for the remote channel are being routed incorrectly out over the public interface. The routing tables or the subnet mask may be wrong.

The intermediary IP addresses, IDs 1 to n-1, are the routers through which the packets pass to reach the destination. In the example there is a single router with the IP address of 192.168.25.254; typically there would only be a single router. In more complex corporate WANs there may be several routers involved. If this is the case, confirm with the Network Engineers that the route taken is the most efficient.

The final IP address - ID 3 in the example - should be the destination channel (IP 192.168.28.1 in the example).

If you see asterisks in place of IP addresses this indicates that there was no route beyond the last IP address displayed.

Example:

Tracing route to 192.168.26.1 over a maximum of 30 hops

0 192.168.27.1
1 192.168.25.254
2 * * *

In the example above pathping could not find a route between the 192.168.25.254 router and the final destination. Check that the IP configuration is correct (particularly the router and subnet masks used), before consulting Network Engineers.

Statistical Analysis

For each segment of the journey a percentage of lost packets will be displayed. In WANs with large amounts of traffic a percentage loss of <5% is probably acceptable, however figures in excess of 5% should be cause for concern. A really healthy network will show no lost packets at any single step. Very large figures can suggest severely congested routers, poor routing algorithms / configurations or faulty hardware. These cases must be rectified before installing Neverfail Heartbeat. A 100% loss indicates that the pathping program failed to reach the destination beyond the router; the most likely cause is an incorrect routing configuration.

Example:

Computing statistics for 50 seconds...
Source to Here   This Node/Link
Hop  RTT    Lost/Sent = Pct  Lost/Sent = Pct  Address
0                                           192.168.27.1
0/ 100 =  0%   |
1    0ms     0/ 100 =  0%     15/ 100 =  15%  192.168.25.254
2    0ms     1/ 100 = 1%      0/ 100 = 0%   192.168.28.1

Pathping will alert you to router overloading and link congestion. The column ‘This Node/Link Pct’ indicates the Lost/Sent percentage of packets. Those percentages next to an IP address indicate a router or physical device such as a server network card. Percentages next to the ‘|’ indicate the loss on a link.

In the example the router 192.168.25.254 is experiencing a 15% loss, this may well indicate overloading or poor routing, incorrect metrics. This example is definitely a cause for concern.

A 1% loss is present on the link between 192.168.25.254 and 192.168.28.1, this is not really a concern, and it may be worth repeating the test again to see if it increases.

WARNING : Do NOT attempt to rectify these problems; they should be dealt with by the Neverfail Network Engineers or the Consultant Company that is responsible for the network.

Testing the Bandwidth

This can be done using Neverfail SCOPE Data Collector reports. Please refer to Neverfail SCOPE documentation on how to set up bandwidth measurement or Knowledgebase article #1000 for more detail on how SCOPE measures the available bandwidth.

On Windows 2003 and 2008 open up Task Manager and select the networking tab to view the channel statistics. Alternatively you can use the Windows Component Networking Monitoring Tools or Perfmon to watch network statistics in particular network utilization. An Ethernet network will typically operate at a maximum utilization of 35% - 40%.


Applies To

All Versions


Related Information

None

KBID-342

0 out of 0 found this helpful

Comments

0 comments

Please sign in to leave a comment.