SNMP was created to simplify network device management by enabling centralized discovery, monitoring, and configuration of devices on the network. It has had a long run for sure and it is hard to let it go as it had been implemented into millions of hardware and software systems over the past three decades. However, it never really achieved all of its goals especially in the area of configuration.
So, is SNMP Dead?
The answer really is yes and no, but overall the real answer is yes. We have seen over the past few years, Microsoft and Google say as much. If you look today at what is being discussed in the industry, it is clear that SNMP has no future. Another clear sign that SNMP has no future is that aren’t any new SNMP standards being worked on or invested in it. There are not even any active IETF working groups chartered to do SNMP work.
This isn’t anything new, to be honest, we have seen this coming for some time now. When RFC 3535 “Overview of the 2002 IAB Network Management Workshop“ was published in 2003, it was concluded that SNMP had failed for use in configuration but was acceptable for monitoring. Thus, SNMP was not really a solution for anyone deploying automation. The findings in RFC 3535 laid the foundation for the requirements for the next generation of configuration management and went on to give birth to NETCONF. I truly believe that NETCONF was the final nail in SNMP’s coffin.
While SNMP has seen widespread use in network monitoring, even in that area various shortcomings have been found which next-generation monitoring systems are looking to solve. If you look into this area, there are recognized and much-discussed shortcomings for SNMP when it comes to monitoring. For instance, SNMP doesn’t have source time stamps for logging events. During heavy load times, you can get some anomalies in the SNMP sampling which makes diagnosing and fixing performance issues very hard. Additionally, some of the negatives about SNMP include the fact that it is pull driven (i.e. polling), it doesn’t really automate well, and it doesn’t scale well into large or busy networks.
SNMP did its job a long time ago, but it isn’t really relevant today except for legacy systems that were designed around it. Network monitoring professionals are looking at streaming telemetry as the future of their networks. This is just another reason why we can say SNMP is dead and has no future. It won’t be used in the next greatest thing being innovated in today’s ever-growing networks.
Today, both the IETF and the OpenConfig community are focusing on telemetry solutions and working on developing the best standards to run with; YANG Push and gNMI respectively. The question is which of these will we be the industry choice in the future? One thing for sure is that it won’t be SNMP.
I recently read a great article by Benoît Claise from Cisco where he shares his thoughts and a comparison of the different standards from the IETF and Openconfig for streaming telemetry. It will be very interesting to watch what happens from here. Mr. Claise states, “Now that the IETF specifications are finally published, the market will decide which approach will survive between … if not both. Or maybe the market has already decided?” Either way, SNMP is clearly dead.
In summary, I think it is time for us to just say SNMP is dead. Right now, it is on life support and nothing new is being developed. But one thing that remains the same in the network is how long-lived it can be. SNMP may still be in use over the next decade, but it will be replaced as legacy networks become modernized. SNMP is dead; LONG LIVE NETWORK PROGAMMABILITY.