Hi,
When I configure switch, add it for monitoring on Nagios XI, by default, if notifications are enabled, I will receive notification about any change on the switch.
The sole purpose of SNMP trap is to send notification like that passively.
Since that function is default in Nagios XI, and Nagios will always inform us about any changes, what is the purpose of SNMP trap in configuration wizard?
Thanks.
SNMP traps
Re: SNMP traps
Hi Flender,
Redundancy is a good thing, let me say that again, redundancy is a good thing ; -).
There are a couple of reasons we have traps available in addition to active SNMP polling. Some environments do not allow active polling for security, technical, or policy reasons and traps can be a useful way to work around those restrictions. For example devices at a remote site behind a host using NAT ( often a firewall ) can send traps for events rather than be polled directly. Also there are some devices and situations where the snmptrap mechanism allows for more flexibility in messaging.
Regardless in most cases people prefer to use active polling for it's predictability.
Hope this is useful.
Thanks and Best Regards,
Keith
Redundancy is a good thing, let me say that again, redundancy is a good thing ; -).
There are a couple of reasons we have traps available in addition to active SNMP polling. Some environments do not allow active polling for security, technical, or policy reasons and traps can be a useful way to work around those restrictions. For example devices at a remote site behind a host using NAT ( often a firewall ) can send traps for events rather than be polled directly. Also there are some devices and situations where the snmptrap mechanism allows for more flexibility in messaging.
Regardless in most cases people prefer to use active polling for it's predictability.
Hope this is useful.
Thanks and Best Regards,
Keith
Re: SNMP traps
SNMP polling (ie, using check_snmp) is useful for things that are predictable. That's counters, timers, and numbers in general (memory, cpu, counts of things, etc). Traps are asynchronous. That means they are not predictable as to when they will occur. They're useful for security events, hardware warnings (like power supply failure, temperature too high, etc) and other things that you cannot predict when they will occur.
Traps and polling are complementary to each other, not duplicative or opposite. You can use them both for different things (or the same things, looked at separately). Remember that traps all go into a single bucket though, so if you want to get more complex trap management other than "here's a trap, you should alert on it" then you need to get more complicated with your trap handling. SNMP polling however, is just "monitor a new OID with a new Nagios service" and be done with it.
Traps and polling are complementary to each other, not duplicative or opposite. You can use them both for different things (or the same things, looked at separately). Remember that traps all go into a single bucket though, so if you want to get more complex trap management other than "here's a trap, you should alert on it" then you need to get more complicated with your trap handling. SNMP polling however, is just "monitor a new OID with a new Nagios service" and be done with it.
Re: SNMP traps
Hi Eric,
Thank you, that's an excellent and clear description of the differences and some of the use cases. Much appreciated.
Thanks and Best Regards,
Keith
Thank you, that's an excellent and clear description of the differences and some of the use cases. Much appreciated.
Thanks and Best Regards,
Keith