Does the NFV service require big data, small data, or both ?, Both nfv and nfv

Does the NFV service require big data, small data, or both ?, Both nfv and nfv

Operating NFV-based services and networks is the next service focus of progressive communication service providers (CSPs). However, it is not easy to achieve this goal. In fact, CSP indicates that it takes a lot of time and effort to build VNF and run vnf in the nfv environment.

NFV is switching from technology to operation, from how to implement NFV to how to operate NFV. One of the key challenges is how to integrate NFV network and service management into existing operations so that CSP can effectively run the current network and NFV cloud network.

How can we run NFV on networks and services after going online? How to ensure on-demand service? How to provide dynamic SLA? If these problems cannot be solved, service providers will not be able to provide these services as commercial services, and they will not be able to transform to NFV and cloud network business cases.

A big risk is that NFV will become a technical isolated island, a good lab project, but eventually excluded from the operator's service system. What's embarrassing is that the main driver of SDN and NFV is the key business orientation of service providers, that is, the need to quickly start personalized services and run standardized and scalable networks.

More than 100 network operators and service providers around the world participated in the NFV Service Guarantee and Analysis Research completed by Heavy Reading in fourth quarter of 2017. CSP said they are working to implement NFV and are working to break away from the internal impact of the organization, including the internal knowledge and software skills gap; differences of opinion between the network and IT team on new requirements; how to fill in gaps in existing IT systems and lack clear guidance on the industry direction required for NFV Service Assurance.

Almost every issue surrounding Service Assurance is rated as a huge/major challenge by at least 40% of CSPs. The top five challenges that CSP rated as "large-scale" or "important" in its NFV operations include:

For CSPs that have already deployed NFV in the network, the key challenges include managing the interoperability and performance between multiple VNF vendors. Traditional interfaces, such as mobile signaling, management system, and component Manager (EMS)/configuration, are lagging behind. on the cloud, only one VIM manager is needed to process the performance of a single VNF. However, when CSPs provide user-oriented services with VNF from different vendors, there is a gap in implementation. These services need to be configured, guaranteed, and monitored to provide end-to-end service quality. CSP needs a centralized platform with real-time operational data to actively manage multiple networks and service layers.

General CPE platform has gradually become a real service guarantee problems, operators need to surpass a single manufacturer of SD-WAN VNF, deploy multiple lightweight VNF (such as firewall, ip pbx, Server Load balancer, and application acceleration) from different vendors ). In this case, CSP will take the initiative to test and monitor as a management service quality, solve user problems and ensure that the service works accurately after configuration or reconfiguration and meet dynamic SLA.

Service providers also want virtual detectors or test agents to have lightweight CPU and memory usage space, which can be containerized, therefore, you can perform an active test without interrupting the real-time traffic. CSP indicates that once VNF is deployed, a highly automated virtual circle or lifecycle covering Order Management to re-execution is required. This must be carried out in a very orderly and automated manner, and is a very good engine without any obvious interference to end users.

62% of CSPs believe that active testing is very valuable, especially when automated and driven by VNF orchestrations, many providers (such as Netrounds) provide orchestration and closed-loop assurance through APIS.

This reflects the industry's desire to improve network programmability and automation. Service providers also hope to implement service orchestration with as few manual interventions as possible, so as to promote automation and process improvement and provide services.

Dr. Stefan Vallin, director of product strategy at Netrounds, said that data from active tests and monitoring produced detailed real-time service KPIs, which are called "small data ". The data itself provides great value, but it is also a driving factor for the successful application of big data and AI. Small data obtained from active testing and monitoring directly answers many important service assurance questions.

Stefan Vallin said: "If we can directly measure the quality of service, why do we need to reverse engineer it with incomplete data at the resource Layer ?"

For customer-centric service assurance, service providers need to visualize their end-to-end services, prioritize issues and avoid errors that affect users, and reduce meaningless data overload. CSP, which focuses on Selling Enterprise Services, worries that they may be unable to correctly determine priority data due to user or service impact. They hope that in the future, they should stop paying attention to device alarms in real time, instead, you only need to care about service alarms and service monitoring, because MANO will process policies/re-route decisions. They will use big data analysis for association between different layers for troubleshooting, forensics, and root cause analysis.

This allows you to manage faults in a more effective way, get rid of manual faults and non-time sensitive alarms, and focus on making the network domain remedy and solving problems, at the same time, it increases the service-level fault prevention or repair required to automatically fix problems that affect users.

A level-1 operator in the Asia Pacific region said: "The NFV environment needs to restore user services, service awareness, and manage user services, and be responsible for these services ."

In short, service providers urgently need big data and small data to effectively operate NFV networks. One of the most notable features of big data analysis, machine learning, and AI is based on high-quality data input, rather than a large amount of low-level data from the resource layer. To this end, service providers began to cooperate with VNF vendors to manage data volumes from VNF in real time and put forward stronger requirements on devices to provide less high-quality data, so as to better answer the above questions.

High-quality data will help train algorithms to predict what will happen to prevent downtime or service degradation. The second focus is to use new data sources to measure the service quality of actual delivery in real time. From the customer's perspective, the preceding small data can directly provide key service indicators for user service providers. This is the best way to understand the service quality and experience in the user's eyes, rather than trying to use resource data to generate service KPIs at a higher level based on lower-level information. It is important to configure services, roll back services quickly, and experience user needs in real time at critical moments or on a continuous basis.