Industry Solutions and Trends
Technology is more than just networking and Juniper experts share their views on all the trends affecting IT
Juniper Employee , Juniper Employee Juniper Employee
Industry Solutions and Trends
Bringing Predictive Insights to IP/MPLS Core Networking
May 4, 2018

Imagine for a second that you’re operating an industrial plant. Giant machinery enveloped by metal pipes and thickets of wire that hums along while steam billows from above. All of a sudden, it all comes to a screetching halt.


What’s the problem? How long until it’s fixed? Without visibility into the status of each critical component of your operation, it’s going to be a very long process tyring to get everything back online. Wouldn’t it be great if you had an all-seeing eye and an automatic way of resolving these issues, many before they ever happen?


Something similar is happening in the network.


Ubiquitous internet is redefining and accelerating the pace of change in both emerging applications and customer expectations. As services move to the cloud and become more distributed to support emerging 5G, edge computing and IOT trends, the networking infrastructure not only needs to deliver greater capacity, but also offer a highly reliable, high quality user experience.


These demands can place additional strain on the underlying infrastructure and lead to performance degradation, service disruption and, ultimately, system-wide failures. In order to balance availability and reliable SLAs for emerging applications with optimal economics, network operators must rethink operations and deliver remediation in a predictive versus reactive manner.


As traffic volumes increase in core networks, maintaining high availability and meeting SLAs in a predictable manner with optimal economics becomes critical. In traditional networks, problems occur when controllers are too slow to make changes without impacting traffic flow and SLA compliance. Operators can take a threshold-based approach to detect degradations earlier, but with massive amounts of telemetry it’s harder for humans to recognize and implement policies to identify those degradations.


As part of The Self-Driving Network, Juniper Networks brings together powerful synergy among efficient Junos streaming telemetry, a flexible NorthStar Controller and predictive AppFormix insights that work together to bring rapid, predictive remediation to IP/MPLS core networks.


“Synergy – the bonus that is achieved when things work together harmoniously.” – Mark Twain         


Juniper’s AppFormix makes it possible to insert high-volume network telemetry that anticipates and provides insights into degradations and faults before they ever occur. These predictive insights turn into closed-loop control by alerting the NorthStar Controller to take corrective action at a network level. With this solution, service providers can now bring rapid network tuning, predictive, high responsive remediation and reliable service assurance to core networks.


Available now, this solution is currently in trials at major Tier-1 service providers and was demonstrated at MPLS World Congress 2018



In this solution, the AppFormix analytics platform ingests Junos streaming telemetry and third-party device telemetry information in a highly scalable manner, then applies machine learning techniques to predict trouble spots before they even occur. Once trouble spots are identified, notifications are sent to the NorthStar WAN SDN Controller to take immediate corrective action. AppFormix also offers scalability through its highly distributed agent architecture.




Efficient Junos Streaming Telemetry

Traditional management methods like SNMP and CLI polling models are limited in terms of scalability and efficiency. Juniper’s streaming telemetry solution, Junos Telemetry Interface (JTI), overcomes these issues by employing mechanisms such as an asynchronous push model to eliminate traditional polling. JTI is highly scalable and can monitor thousands of sensors in Juniper’s physical and virtual nodes. JTI also supports real-time operational data streaming to synchronize operational state with external controllers and analytics platforms, leading to faster decision making compared to more traditional approaches.


Flexible NorthStar Controller

Juniper’s NorthStar WAN SDN Controller is a powerful yet flexible traffic engineering solution that allows operators to proactively monitor, plan and traffic engineer their core network with global visibility and control. For service providers, the result is higher utilization, predictability, resilience and SLA assurance at the network level. NorthStar leverages IETF and web protocol standards to ensure seamless multivendor integration and OSS/BSS integration. It works across optical and IP/MPLS layers to optimize multi-layer control, with standards-based traffic engineering using Path Computation Element Protocol (PCEP) and Source Routing In NetworkinG (SPRING)/segment routing.


Predictive AppFormix Insights

Juniper AppFormix is a cloud operations automation and artificial intelligence platform that gives cloud operations teams a global policy framework for defining a set of abstracted intent-based policies. It does all of the heavy lifting, transforming intent-based business policies into infrastructure-level constructs and distributing policy enforcement for monitoring and predictive analysis.


Distributed smart agents collect and analyze device-level telemetry data relevant to the predefined distributed policy constructs, allowing AppFormix platforms to scale with the cloud infrastructure via a multiple agent model. AppFormix can perform unsupervised machine learning techniques, such as clustering to dynamically learn and baseline network infrastructure resources while they operate, generating alarms and notifications to external systems when real-time metrics deviate from trends that exceed tolerance levels.


At Juniper, we are on a mission to realize the The Self-Driving Network in various network domains by employing SDN, automation, streaming telemetry, big data analytics and machine learning technologies. As high bandwidth applications emerge in a distributed cloud delivery model, the traffic volume in IP/MPLS core networks increases—all the more reason to maintain a stricter, more reliable user experience. In realizing The Self-Driving Network in the IP/MPLS core domain, we are creating synergy between Junos streaming telemetry, the NorthStar WAN SDN Controller and the AppFormix analytics platform to enable responsive, predictive remediation and control.

0 Kudos
May 4, 2018
Anup Kumar Mishra

Informative article, looking forward to see more use cases.

May 5, 2018
Distinguished Expert

The embedded video link is broken.