Private networks present a number of challenges for testing and assurance. By their nature, private networks are delivering some type of service aspect that cannot be achieved via the public network, whether that be more secure access, targeted coverage, or more sophisticated features such as high reliability or low latency. But how are those service levels ensured and maintained, and what are some of the factors in play? Here are three.
–The use case determines the network performance aspects that must be assured. In a conversation during the virtual Private Networks Global Forum event, Jagadeesh Dantuluri, GM of private and dedicated networks and Industry 4.0 for Keysight Technologies, gave the example of a California school district that implemented a private network in order to provide home connectivity to students. Having a reliable connection to the student is the primary use case, he noted; the key performance indicator (KPI) then, is what percentage of time that a student can successfully connect, and latency is a secondary concern. In the scenario of a private network for an auto manufacturer, he offered two hypothetical use cases: Constant connectivity for the human-machine interface (with extra attention paid to RF monitoring, to ensure high reliability within an interference-prone environment) and operating AGVs, which would need low latency of around 15-20 milliseconds. The uses ultimately shape how the system is assured.
-The depth of available data visibility changes your view, and ability to find faults. “The KPIs and SLAs are quite different when you take the different use cases in the private 5G. And what the network should actually give you is the visibility into these KPIs and SLAs,” said Sree Nandan Atur, technical director of Symcloud at Rakuten Symphony. “Which means that you should have the data from the ground up: infrastructure, cluster, and the applications get the fault management data, performance management data so as to collect store process and dissect whether we are actually meeting these SLAs.” Atur gave an example of automated self-healing based on such data, saying that that in instances where an application and the network are cloud-native, Rakuten Symphony has seen glitches in the radio access network add milliseconds of latency—but the Kubernetes-based system recovered because was “designed in such a way that it gives you closed loop automation without the intervention of the network operator or the application. Things were fixed at a very, very fast pace. … These are some of the things that we are actually seeing to say that yes, KPAs and SLAs are important; [and] it depends on the different use cases and how you collect them and dissect them and use them.”
–Domain KPIs have to be translated to network KPIs, and that process is still taking shape. Private network customers have a different domain expertise—not telecom—and they think about KPIs accordingly, Dantuluri said. “What they care about is their own domain knowledge KPIs. And the school district is about, is the kid able to connect? Or, in the assembly line, am I able to connect my machine, my AGV, et cetera, [all the time]? … The domain KPIs have to be mapped into the network KPIs. That’s the number one thing. That is an industry-wide challenge that we are going through, and it’ll take some time. Having said that, the ICT players like us in industry, who [provide] the end-to-end connectivity and everything, we are also helping in educating … [and] building that expertise.”
View this full conversation and more sessions on private networks on-demand from Private Networks Global Forum.
