White Papers

Understanding OPNFV

Issue link: https://read.uberflip.com/i/1344850

Contents of this Issue

Navigation

Page 27 of 144

Understanding OPNFV 27 decelerate its deployment. Therefore, a key requirement for NFV to succeed is operational simplicity. From "day-1" initial deployment to "day-2" post-deployment operations such as changing configurations, adding capacity, adding functionality, modifying reference architectures, updates and upgrades, all need to be simplified dramatically for NFV to be deployed at a large scale. And by the way, these operations need to be performed without interrupting existing VNF workloads. Additionally, log monitoring, metrics and alerts need to be provided for an entire cluster in aggregate, in a simple to consume manner. NFV-Specific Requirements NFV has numerous unique requirements that are distinct from, say, enterprise requirements. Let us look at a couple of the most important ones: ● Service assurance: Arguably the availability demands on a VNF are higher than a web or mobile cloud application. After all, an OnStar customer relying on the LTE network to request emergency services in response to an accident cannot tolerate any loss of availability of the network! So how are we to detect the root cause of a failure and address it efficiently? This is the crux of service assurance, where we want to be able to trace the root cause of a failure to the actual infrastructure fault and then address it in the most optimal manner. Over time, we want to predict potential failures before they happen, and be proactive rather than reactive. The holy grail is to use artificial intelligence and machine learning to predict and respond to issues. ● Service function chaining: This is a concept where a service provider needs to apply services, consisting of multiple chained VNFs, to data traffic between two endpoints. These services could be anything from authentication to security to transcoding. What makes the requirement even more interesting is that the insertion of VNFs is policy- based and dynamic. So, two users requesting a video from the same website could have different VNFs inserted in the middle depending on the device they are using, their payment plan, and, say, their location. Next Steps NFV is gaining momentum both in the telecom and enterprise fields, which is leading to new service models and use cases, as evidenced by a study conducted by Heavy Reading in August, 2016, where 98% of the service providers surveyed indicated that NFV is part of their strategy. More generally speaking, NFV is nothing but a network-heavy workload (as opposed to compute-heavy). With the rise of IOT, machine-to-machine communication, intelligent edge devices (wearables, drones, augmented reality, and so on) NFV will see widespread adoption in enterprises as well. Net-net, everyone needs higher quality, more dynamic, increasingly elastic, and better tuned connectivity to their applications and services.

Articles in this issue

Links on this page

view archives of White Papers - Understanding OPNFV