Accelerate NFV Migration with Probe-based Service
Network function virtualization (NFV) is radically changing the communications deployment model. Traditional deployments take a waterfall approach, investing first in network deployment, then in network management systems and service fulfillment, and finally in service assurance and customer experience management. NFV and software-defined networking (SDN) turn this model around, taking a top-down approach by molding the network to service requirements.
The new model requires service orchestration – in conjunction with various network management and orchestration elements – to automate the steps required to build and provision end-to-end service. By making service assurance monitoring and customer experience management an integral part of this service orchestration, service providers can deliver right-sized solutions from the start. They can also identify when to scale with new virtual elements to ensure high networ performance and quality of experience.
The missing components in this transition to NFV are probe solutions that can address hybrid environments encompassing both virtual and physical networks. Such a hybrid world requires probes that deliver true end-to-end network visibility (Figure 1).
Figure 1. Hybrid networks with NFV elements require a virtual probe (vProbe) solution to deliver true end-to-end network visibility.
Taps or splitters, for instance, cannot monitor logical interfaces that use internal virtual machine-to-virtual machine (VM-to-VM) communication between functions hosted on the same server. Monitoring such virtual network elements requires a virtual probe (vProbe) function that monitors all traffic in the virtualization platform – both external traffic on the virtualization platform’s physical interfaces, as well as VM-to-VM communications.
One of the first vendors to launch such a vProbe for NFV is RADCOM. In 2014, the company introduced its MaveriQ service assurance and customer experience management solution. With probes that support both legacy network traffic and virtual traffic in its product portfolio, the company serves as a comprehensive probe provider for hybrid networks.
NFV-ready, the MaveriQ solution helps service providers handle today’s massive traffic explosion. The solution is composed of three elements:
- MaveriQ vProbes deploy in an NFV network as a Virtual Network Function (VNF) or in a legacy network with the same software running on commercial-off-the-shelf (COTS) hardware. No matter how deployed, they provide the same functionality and can be mixed in the same network under a single MaveriQ management system. These vProbes scrutinize multiple services such as mobile data, voice, SMS, and video that run over a wide range of technologies including LTE, VoLTE, UMTS, GPRS, and IMS, all in one box (or as a VNF).
- The MaveriQ Service Assurance Application Suite delivers solutions for roaming, interconnection and service-level agreement (SLA) monitoring.
- The MaveriQ Management System provides network analytics and a presentation layer.
MaveriQ vProbes can be seamlessly deployed on third-party virtualization platforms with scaling elasticity and high availability (Figure 2). They use virtual network interface card (vNIC) and virtual switch (vSwitch) capabilities to non-intrusively monitor traffic on external network interfaces as well as internal VM-to-VM communication.
Figure 2. MaveriQ vProbe is a virtual network function that can be deployed on third-party virtualization platforms.
Using MaveriQ vProbe, operators can rapidly increase network analytics capacity when needed, by adding vProbe instances on additional VMs on the same hardware, keeping CapEX and OpEX under control. The MaveriQ vProbe can be easily migrated between hosts which can be used to increase redundancy and eliminate application downtime during planned server maintenance.
MaveriQ can simultaneously monitor both virtual interfaces on cloud-based networks and physical interfaces. Data collected from both physical and vProbes can be analyzed, stored, and presented by a single central management and database solution which may be installed on either physical or virtual environment.
MaveriQ addresses multiple service provider needs in an NFV architecture, including customer experience monitoring, network performance monitoring, service optimization, marketing analytics, network planning, audits, and subscriber and network troubleshooting in a unified operational flow (Figure 3). MaveriQ employs a comprehensive array of performance and measurement methodologies to continuously analyze service and network performance, big data, and quality. Enhanced correlation capabilities give service providers full visibility of the network across technologies, while also providing the ability to drill down and troubleshoot to the session level and to see all the details of an operator’s traffic.
Figure 3. Example of MaveriQ in a comprehensive service assurance and customer experience monitoring solution for NFV architecture.
MaveriQ is a unique solution optimized for terabit networks. Using advanced packet-processing technologies, MaveriQ vProbes offer low footprint and high capacity on a cost-effective platform. They have unlimited scalability, utilizing the ever-increasing CPU power, as well as built-in load balancing capability. One MaveriQ probe can simultaneously monitor LTE, 3G, 2G, VoLTE/IMS, VoIP, mobile core and wireline data signaling and user plane interfaces, dramatically reducing the number of probes required for monitoring multi-technology networks.
MaveriQ delivers the capacity to analyze hundreds or even thousands of Gigabits per second (Gbps) in real-time by utilizing technologies such as big data analytics engines, column-based data warehousing, massively parallel processing (MPP) and ultra-fast packet-processing techniques. The latter includes specifically Intel’s intelligent offload technology employing the latest Intel® Xeon® processors, accelerations in certain network controllers and the Data Plane Developer Kit (DPDK).
The Ideal MaveriQ Commodity Hardware Platform
Service providers looking to take advantage of the performance boost of Intel’s intelligent offload technology will find a number of solutions from members of the Intel® Internet of Things Solutions Alliance. The most powerful of these use the latest Intel® Xeon® processors.
For example, the Intel® Xeon® processor E5-2600 v4 family features up to 36 cores (and up to 72 threads) in a dual-socket configuration – 12 more cores than the previous generation. These processors are up to 44 percent faster with a 23 percent average generational performance gain across key industry-standard workloads and applications. In addition, the Intel Xeon processor E5-2600 v4 family delivers key improvements in virtualization efficiency that enhance application performance.
The processors deliver the greatest performance when paired with the Intel® Ethernet Controller XL710 product family (Figure 4). These 10- and 40-gigabit Ethernet controllers extend Intel® Virtualization Technology (Intel® VT) hardware assistance to network virtualization to reduce I/O bottlenecks by intelligently offloading networking traffic per VM, enabling near-native performance and VM scalability. For example, the controllers support standards-based offload overlays like Virtual Extensible LAN (VXLAN) and Network Virtualization using Generic Routing Encapsulation (NVGRE).
Figure 4. The Intel® Ethernet Controller XL710 product family provides a comprehensive feature set for extending Intel® Virtualization Technology (Intel® VT) hardware assistance to network virtualization.
An innovative offload feature, Intel® Ethernet Flow Director, directs packets to the core and application that requires them by observing outgoing flows and creating a connection between sources and destinations. The Intel Ethernet Flow Director supports up to 8,000 perfect match values extracted from sampled outgoing packets and stored on die to provide an efficient method for classifying packets and setting affinity of flows to cores.
The Intel Ethernet Controller XL710 product family is also optimized for DPDK. DPDK’s community-driven, open source packet-processing libraries maximize performance on Intel Xeon processors and the Intel Ethernet Controller XL710 product family. Running in the Linux* user space environment, DPDK assists in managing memory, buffers, queues, and flow classifications. Together, Intel Xeon processors, Intel Ethernet Controller XL710s and DPDK create a platform that can intelligently and rapidly process different data flows for NFV and the MaveriQ service assurance and customer experience management solution.
Put Service Assurance Front and Center in Your NFV Implementation.
Modernize probe functions and better support NFV architecture by transitioning from proprietary appliances to MaveriQ. For systems based on the Intel® Xeon® Processor E5 v4 product family, visit the Alliance’s Solution Directory.