前瞻的力量：主动保障用例

The Power of Proactive Active Assurance Use Cases We need a fundamental shift in our approach to assuring the network. The network ismoving to the cloud. Functions are now virtualized. And next-gen networks are too dynamicfor manual troubleshooting to handle. There is a compelling argument for employingActive Assurance in tandem with Passive Assurance. As 5G and SD-WAN drive us to adoptvirtualization, Active Assurance will be a critical enabler of the automation needed tosuccessfully ensure differentiated performance and quality for these next-gen services. Active Plus Passive Assurance: Getting the Best of Both Worlds Passive Assurance is the traditional method for determiningthe health of the network. It collects data from virtual orphysical network functions and other sources once thefunction, slice, or service is up and running. Passive Assuranceis good for identifying severe, customer-impacting problemsas traffic runs through the network. But, as we pointed outin the previous paper, Passive Assurance has its limitationswhen it comes to turn-up, monitoring, and troubleshooting. Active Assurance fills the gaps left by Passive Assurance byemulating network functions, devices, and users to createhighly realistic synthetic traffic (via virtual test agents, orVTAs). Active assurance inserts this synthetic traffic across theend-to-end network and evaluates performance, making itideal for pinpointing the root cause of a problem. It enablesservice providers to evaluate performance at turn up,check critical services and links regardless of traffic levels,proactively identify issues by using defined traffic so minorfluctuations are discernable, and isolate problems anywherein the network cost-effectively. This makes active assuranceideal for when the service provider first turns up a function,makes a change to a function, wants to ensure a public safetynetwork is functioning properly, or needs to isolate a complexproblem. To fully grasp the benefits—and necessity—of ActiveAssurance in a real-world context, it would help to examineseveral specific examples. The following examples are basedon our experience implementing assurance in a wide varietyof networks and environments for both fixed and mobileservice providers: •Passive Assurance is ideal to: – Monitor performance once the network is up and running–Track services and links that have consistent traffic flows– Detect issues in high-priority parts of the network– Determine how many users are impacted by these issues •Turn-up, monitoring and troubleshooting for 5G slicing•Turn-up, monitoring and troubleshooting for SD-WAN•Monitoring of an IP mesh network•Turn-up and change management of LTE mobile networks •Active Assurance lets you: – Evaluate performance at turn-up before customerusage starts–Continuously check services & links even when trafficlevels are low– Proactively identify issues before they become majorproblems– Troubleshoot complex issues in any part of the networkincluding customer premises and over-the-air / RF Use Case: Active Assurance for 5G Slicing Expectations have been set that 5G will be a complete gamechanger, so the stakes are higher than ever for operators todeliver differentiated experiences and guaranteed qualityfor a range of new applications, from connected vehiclesto IoT sensors and beyond. To this end, 5G revolutionizesnetwork slicing with improved control over end-to-endperformance and enhanced scalability to support thedeployment of large numbers of slices. However, thepotential for a multitude of slices, each with differentiatedperformance expectations, creates real challenges forassurance. In this example, everything from the UPF in the core networkto the data networks is “green,” or operating correctly. Fromthe premises to the UPF, however, test agents report errorsincluding the on-premises agent. The problem, therefore,is isolated to the UPF, its infrastructure and the associatednetwork connections. We can then find the root causeso it can be fixed before the customer notices a servicedisruption and before any SLAs are violated. Once the fix isimplemented to the UPF we can test the new UPF before itgoes live, ensuring the problem is truly fixed. Because it’s dependent on actual user traffic, passiveassurance, in contrast, has a limited ability to detectperformance issues that occur during periods when usageis low, such as the middle of the night. In addition, after theturn-up of a new UPF, passive assurance isn’t able to assessthe UPF before it goes live. And while passive assurance canmonitor latency, jitter and packet loss for the user-plane, dueto the high cost of decryption, it isn’t cost effective to directlyexamine user-plane packets, so quality measures such asMean Opinion Score (MOS) are statistical estimates. ActiveAssurance inserts known synthetic traffic into the networkand directly measures performance and perceptual quality(MOS) of user-plane packets at end-points and across eachnetwork segment. To