Hi,
I am currently looking at adding SNMP agent functionality to some existing telecoms mobile infrastructure software running under Linux. This would be done by defining and implementing some enterprise- specific MIBs, and using SNMPv3.
Linux has its own standard SNMP agent (running as the snmpd daemon) that handles standard MIB variables, such as MIB-II (RFC 1213).
For a mobile telecoms operator's point of view, is it better to have two separate SNMP agents, one for the standard basic Linux node and one for the telecoms software, or is it preferable to incorporate both into a single SNMP agent?
A single SNMP agent could run using the standard SNMP UDP ports, i.e.
161 for operations such as gets and sets and 162 for operations like traps, whereas different ports could be used if there is more than one agent.I know that it is possible to have sub agents, with communication between sub agents and a master agent using a protocol such as AgentX or SNMP.
Are there any views, recommendations or best practices on this, particularly for mobile telecoms management?
Thanks in advance for any info or advice.