Real-time analytics and dynamic policies are shaping the way subscribers and interest groups are served by Communication Service Providers (CSPs) networks. However, to keep pace with the increasing variance of QoS attributes driven by the specialization of digital streams across industries, CSPs must extend their approach and make their network platform more reactive and tunable in order to respond to granular demand. To achieve this goal, they must exploit new UX (User Experience) data sources to optimize the operation of their Carrier Cloud while reducing churn and opex. In this blog, we are discussing the integration of Enterprise APM (Application Performance Management) UX KPIs with CSPs’ policy frameworks to drive operational efficiencies and increased CSAT.
This blog present’s a point of view on the capacity challenges facing CSPs and how they can better utilize their assets along with their customers’ intelligence to meet growing demand. The blog will also be used to gather external feedback on the proposed framework that Aricent is actively developing with its partners.
The ability to determine dynamically the dimensioning rules associated with a given VNF is possible with Aricent’s SOC² (Self-Optimizing Carrier Cloud) framework. The complex rule engine can machine learn the actual performance and engineering limits of the VNF based on various NFVI configurations. As OEMs and ISVs deliver VNFs to CSPs based on specific NFVI configurations, it is now possible with SOC² to automatically optimize the engineering rules based on the specific CSP context. The same framework can also be used to characterize more complex service chains. SOC² continuously converges towards an optimal network operation state that maximizes telecom UX and optimizes resource utilization. Extending the framework to the end-to-end value chain, where OTT content and applications deliver part of the user experience, is desirable. However, this is currently limited by the lack of visibility and perception of the actual UX by the CSP. One way to address this is to exploit the UX intelligence readily available from enterprise APM tools. For example, this would address situations where the network is apparently performing flawlessly from an NFVI/NFV/Telecom Service Chain perspective but where the streaming of a viral video in a given market is affecting x% of the users with very poor connectivity and media quality. We are therefore proposing to exploit the APM analytics so that the Ux is optimized across the stack including the application layer and user device. This means that application KPIs like time to download web page, latency of certain web services and video quality would be analyzed and correlated with network KPIs by SOC² analytic engine to deliver the best end user experience.
The current network performance KPIs capture major trends at the L1/L2/L3 layers. Today CSPs are also in a position to monitor the most important OTT applications using URL, protocol types and various analytics. However, spontaneous usage bursts of Tier 2/3 apps in various markets may be only marginally addressed by current policy and orchestration implementations. Furthermore new AR/VR applications, the steady increase in video usage and unpredictable social responses to various events will result in unprecedented QOS requirements variance for individual users, groups or markets that may severely impact the quality of experience (QoE).
CSPs need to come up with better data models and APIs
The main problem with our approach is that there are many APM tools used across the various industries and enterprise verticals, including IoT applications. Furthermore, the APM tools will have to interface with all CSPs rendering their applications. Therefore the CSPs need to come up with data models and APIs that can be easily used by the enterprises seeking better performance. The subscription model is up for discussion but it is likely that in the long run CSPs would offer that service for free to be at par with their competitors. Otherwise we see the following challenges for Enterprise APM tool integration with the CSP policy framework.
Lack of third party application UX dimensions that is critical to enterprises (e.g. web service latency may limit eCommerce transaction’s effectiveness). Enterprises cannot get consistent UX across markets and countries with different operators.
Enterprise APM tools address only the application layer. They cannot exercise control over the network despite the wealth of information they produce for the vertical owner.
- Thousands of applications and customers with evolving needs and mission critical dependencies
- Digitization and IoT forces enterprises across industries to work towards the improvement of their digital value streams and actual QoE visibility
Closer partnership with their CSPs will open up new business models and promote digital durability
- Enable more control by the enterprise to define how the network shall be conditioned to render the application at a given time for a given context
- CSP’s network transformation and operation teams are not sufficient for granular network tuning to render applications
- They need to exploit SOC² like components (analytics, rule engine, unified data model) and maximize automation
- Current or planned systems (OSS, orchestrator, workflow and policy engines) may not easily be open to outside world
- Need for agile partner that can develop in the CSP specific SDN/NFV–APM integration architecture and minimize incremental investments
CSPs must exploit intelligence from APM tools
To cope with the increasing QOS requirement variance between applications, we believe CSPs should consider exploiting the intelligence available from their enterprise customers’ APM tools. This can be achieved using optimized data models that can easily feed into the analytics and workflow automation frameworks controlling their network. Aricent has readily demonstrated such an implementation using its SOC² framework along with the APM tool from its partner Avekshaa.
Aricent orchestration engineering services can streamline the APM integration process and deliver quick operational improvements for key enterprise customers.
At the early stage it is possible to use a synthetic UX index per application and market so that benefits can quickly be observed as the model gets extended to other applications.
Aricent is vendor agnostic and will deliver a program plan to the CSP that addresses the immediate requirements through a future proof architecture that leverages existing assets and will pave the way for more sophisticated scenarios. (e.g. AR/VR, eMBMS, eCommerce specific KPIs optimizations).