Solutions Showcase

Operators currently manage these networks and protocols using a variety of ad-hoc tools and scripts; however, the unpredictable and complex interactions between network conditions and workloads make such manual tuning difficult. What is needed is an autonomous approach to network management, i.e., networks should self-drive: sensing, inferring, and deciding without human intervention.

In this work, we present the design and implementation of Taurus, an intelligent data plane capable of running machine learning inference at line rate. Taurus modifies existing programmable network devices, such as switches and NICs, by adding custom hardware based on a map-reduce abstraction; the new block uses SIMD and pipeline parallelism for fast inference.

Our evaluation shows that Taurus achieves an average latency of 270 ns—three orders of magnitude better than traditional control-plane, server-based approaches—while adding at most 24% more chip area using anomaly detection and traffic scheduling models. We believe Taurus is the first step toward realizing a self-driving network, where data-plane devices make intelligent decisions and operate autonomously using machine learning.

Such a pool is instantiated over a large number of administrative and technological domains, which brings a bigger challenge for the management and orchestration of allocated resources. Resource monitoring in this new entity so-called slice is extremely important to address these operations.

Therefore, in this work we propose and implement an architecture for the monitoring of physical and virtual resources in cloud network slices, considering multiple administrative and technological domains. In this paper we focus on the vertical and horizontal elasticity operations in the monitoring context, which means how our architecture dynamically detects the addition or removal of slice resources.

From our preliminary results, we are able to demonstrate the functionality of the architecture components by evaluating the performance of instantiating new monitoring components exercised in operations for start, stop and update monitoring.

The flexibility on how and when to deploy network services in order to attune to the market needs allows network operators to increase or decrease network performance on the basis of the user’s consumption and also provide specific resources that cater the need of a specific Tenant.

Technology to achieve this functionality relies on the use of Software Defined Networking (SDN) and Network Function Virtualization (NFV) as the means to have more control of the behavior and components of a network. Due to this, a mechanism for managing the complex provision and operation of a Mobile Network System is required.

This research proposes a three-layered Virtual Mobile Network System where the top layer includes an Intent-Based (IBN) Tool for policy definition of Data Plane Slices QoS. The middle layer focus on Management of the Virtual Network Functions (VNF) and Policy implementation with the use of the M-CORD Platform.

The lower level is the Physical Layer where the VNFs that represent the EPC and eNodeB reside. The upper layer (Application Layer) main functionality is to provide an ease of configuration for the creation of network slices and the definition of QoS for each of the slices that are created. The IBN Tool is composed of a User Interface (GUI), Architecture Catalog, Resource Manager and a Policy Configurator.

The GUI provides a clean interface where a user can define network slices as “Contracts”. Each contract has an associated slice with specific QoS that are defined based on the type of service that the user is consuming from the mobile network.

The contracts also have information of the type of architecture for what the slices are going to be defined, this information exists on a database called Architecture Catalog, which includes specific deployment information of VNF, e.g. LTE-A has slice QoS definitions for SPGWU and 5G may have definitions specific to a UPF.

This information lets the Policy Configurator know how to properly translate the contract policies into TOSCA configuration files which may differ from one network architecture to another. The TOSCA files have the information of which data plane slices are going to be instantiated as VNF with the proper definition of QoS, before passing the TOSCA configuration to the Management Layer, the Resource Manager has the task of verifying that appropriate RAM and CPU resources exist on the Physical Layer for the instantiation of the VNFs.

This proposal aims to abstract the complex operation and configuration of a Network Sliced Mobile Network System, it also contains enhanced functionality for Network Slice Selection based on previous research and developments.

Then, it shows a specific ONOS and SDN IP application configuration to help a multi-site networking control between the sites, especially if several ONOS controllers control different sites. Importantly, it demonstrates the unique application development to provide limited networking control between ONOS controllers for controlling the (route prefix)-based forwarding by manipulating the ONOS Intent.

The first application is trying to block specific IP prefix if it sent a huge amount of traffic which is assisted by flow visibility for collecting and analyzing the route prefix-based traffic aggregation.

The second application is to provide an automatic mechanism for changing the upstream link in the provider router based on BGP community attributes which are sent from the client router through the downstream link. Both use case applications will be demonstrated either through a live demo or video recorded demo.

Additionally, it will elaborate on how ONOS is being utilized and what are the contributions made to ONOS base code during those applications development such as abstracting the complete multi-to-single-point intent into REST API and extending the BGP route entry for processing BGP community attributes.