Author: David McGee

Link Polling is an important principle and mechanism in Network Management. It is a core feature of any Network Management System (NMS), and the IDMS SNMP Manager. Put simply, link polling involves frequently checking-in on a network endpoint/device to ensure it is reachable, or alive. In the DAS (Distributed Antenna System) world, this is often primarily performed against controllers with a Ping on any device with an IP Address.

Ping is a simple network diagnostic utility used to determine the availability of a host on an Internet Protocol (IP) network. Round-trip time and packet loss is measured. There are numerous implementations of ping and a variety of options [1] that can be passed in.

It is also commonly performed against any interface that has a web page or an agent that responds to simple native SNMP (Simple Network Management Protocol) GET requests, such as get 'sysUpTime'. We will go into further detail on this below.

In the Telecoms DAS arena, it is often assumed that by NOCs (Network Operation Centers) that a system is entirely down if management/visibility is lost to it. Whilst this assumption is not always correct, being blind to managing a system requires a fast response to resolve in the event of power outages impacting an entire controller's functionality. A link being down could mean an entire hub has lost power. If an entire hub has lost power, cellular voice and data services will be impacted.

The main link poll types implemented within the SNMP Manager are:

• ICMP (Internet Control Message Protocol): Ping Message against an IP address. Packet loss and whether it is reachable or not are taken into consideration.
• HTTP Get Request (Simple Web Page Scrape) against an IP address or supported technology URL: Check for a HTTP 200 (OK) response.
• SNMP GET sysUpTime: Varients of SNMP(v1) GET, SNMP(v2) GET requests are attempted. Further details at [3]

Errigal has developed and evolved a very in-depth management process for link polling over the last decade. This process is well-suited to large scale DAS telecoms environments. Please study the diagram below, and reach out to a colleague if you have any thoughts, or questions. Maybe there is room for process improvement? Different technology types may not respond to ping, SNMP GetRequests or may not have a web-based controller. For this reason, elements can be configured differently. We won't give too much away here, but you should take a look at the network_element table in the SNMP Manager database (Hint: Search for pollable keywords).

Please use a QA environment to answer these questions:

1. How do you determine what types of polls are performed against a network element?
2. What are the distinct types of polls?
3. How many network elements have each poll type enabled? (Try a group count by type)
4. What other types of SNMP requests can be performed against an SNMP agent?
5. What is a trace route?

Resources

[1] Wikipedia - Ping (Networking Utility)

[2] HTTP - Request Methods

[3] Simple Network Management Protocol (SNMP) - Protocol details