Network link monitoring is clumsy. It typically relies on the Simple Network Management Protocol (SNMP) which has been around for decades and has an architecture which is cumbersome to deploy especially in the cloud. Additionally, the network management platforms that do SNMP also rely heavily on the Internet Control Message Protocol (ICMP) which is nothing more than ping. The protocols used in these solutions are heavily reliant on polling and are processor and database intensive especially in service provider backbones. They are mostly centralized and do not distribute very well in a structured manner. They are also very difficult to configure in customer networks as either special virtual private networks (VPN) or firewall rules need to be created. The result is that the legacy manner of monitoring a link is problematic as well as costs and arm and a leg.
There is no intelligence in the network device being polled, as the polls themselves are numerous and not optimized. A large amount of raw metrics is extracted which is then processed to generate a valid metric to be stored, analyzed and graphed. This mechanism is dated, instead inspiration can be found in the methods used by the Internet of Things (IoT).
Internet of Things (IoT) devices operate on a different principle to polling and instead the things use a mechanism of push to a cluster of cloud based web services. Also, all the processing to generate a metric is executed on a microcontroller unit (MCU) of the IoT device meaning that the tasks are now distributed across a whole computing ecosystem. The requirement to frequently poll by a network management system (NMS) is also mitigated by the architecture as in IoT devices the metric push is less. So where a NMS might poll every 1 to 5 minutes, an IoT device might push data about every 10 to 30 minutes, as it has the aggregated metrics.
But by far the significant improvement that the methods used by IoT have over a legacy NMS is security. That is because IoT use an outbound encrypted push to a cloud server while an NMS requires an inbound, often non-encrypted poll. In firewall terms it means that IoT can be handled by Network Address Translation (NAT) as a singular rule while an NMS requires more advanced filtering as well as port forwarded. The latter actions are more complex and less secure and associated with an unmanageable number of firewall rules. A simple wireshark exposes all the info being polled by an NMS.
In a software defined wide area network (SD-WAN) environment the methods are further optimizes as there is no requirement for each SD-WAN customer premise equipment (CPE) to upload metrics. As each CPE connects to an aggregator in a data centre, the aggregator has all the required metrics. All that is required is for the aggregators to push these metrics using API calls to cloud web services which publish the analytics. This is quick and has low overhead which results in it being able to scale far larger than any legacy system.
Any decent SD-WAN solution has these metrics built-in and available at no extra cost. So the mere act of implementing SD-WAN kills off the legacy link monitoring system, Good riddance!
Ronald Bartels can demonstrate an excellent SD-WAN solution with all the required metrics and link monitoring abilities.
Ronald Bartels ensures that Internet inhabiting things are connected reliably online at Nepean Networks - the leading specialized SD-WAN provider in South Africa. Learn more about the best SD-WAN provider in the world! 👉 Contact Nepean
