Software-centric networks represent the next evolution for service providers (SP). With the introduction of Network Functions Virtualization (NFV) and Software Defined Network (SDN) technologies, these networks will become more agile and elastic, better managed and orchestrated with automation, and eventually enable SPs to lower the total cost of ownership. Moving to software-centric networks will enable SPs to gain control of service creation, innovate faster and meet competitive challenges.
This evolution will take place over many years as SPs wrestle with a myriad of technological, operational and organizational challenges associated with virtualization. “Hybrid” networks composed of both physical and virtual components, will exist for some time. This blend of networks makes it increasingly important for SPs to have a single source of information from both physical and virtual network elements. NFV and SDN will add plenty of complexity, and if SPs must rely upon several disparate solutions to manage both physical and virtual components, network operation and management will be even more complex and even less efficient.
Monitoring and assuring hybrid networks must be seamless and disruption-free, especially when user traffic spans both physical and virtual infrastructures. These sessions must be correctly “stitched” together at the application layer so that end-to-end KPIs and session traces can be properly constructed. In a mobile network, as session handovers between networks take place, the session identities at the physical and virtual probes will have to be properly correlated to provide a single session trace to the monitoring application.
In a virtualized network, traffic can traverse multiple Virtualized Network Functions (VNFs) between the source and destination. “Service chaining” involves stringing together a series of VNFs to build an end-to-end service. Some VNFs will reside on the same host machine and others will be on different host machines. This virtual construct necessitates virtual probing at the VNF layer to view each instantiation and all hops involved in the service flow.
The ability to spin up new VNFs is a key advantage of NFV; one that enables SPs to build, launch and ramp up (and ramp down) services “on-the-fly”. Any negative impact to customer experience during this process must clearly be avoided. Having visibility and early warning to errors and failures associated with the automatic spinning up (and down) of VNFs, can reduce mean time to repair (MTTR), churn and allow SPs to protect the customer’s experience.
North-south VNF traffic (enters and exits the host machine) can be monitored using physical probes. However, monitoring of “service chain” traffic requires a virtual, software-based probe to be placed on the same hypervisor as the VNFs being monitored. Many purpose-built probes rely upon hardware-assist from custom silicon and cannot be easily abstracted into a software-only component that can run inside a Virtual Machine (VM). Probe software that runs on general-purpose processors and operating systems can be readily ported to run in a VM while maintaining equivalent performance levels.
One concern with “service chain” monitoring involves the throughput limitations of existing vSwitches, which are responsible for connecting VMs. As east-west traffic (within the same host) is tapped by the virtual probe and transmitted upstream to the application platform, the vSwitch resources will be constrained by this additional traffic. Implementing smart, virtual probes that perform intelligent processing of the monitored traffic at the source, enables extraction of the “signal” from the “noise”. This significantly reduces the amount of data sent upstream to the monitoring application while also preventing congestion and increasing efficiency.
Software-centric networks will encourage CSPs to unify previously disparate network ecosystems. For example, employing a common metadata across all physical and virtual probes will facilitate seamless integration at the monitoring application level. In addition, software-based probes can now be easily deployed in places where it was previously not possible or economically feasible using physical probes. Cost-effective monitoring can be extended to the network’s edge, where so much content is now consumed, enabling true end-to-end, 360° visibility and seamless service assurance across physical and virtual infrastructures.
NETSCOUT’s NFV Service Assurance solution provides 360° visibility for a seamless view of what’s really happening across the entire network, and allows SPs to leverage physical infrastructure and still experience the cost savings and service agility associated with NFV. This single, unified view enables end-to-end troubleshooting and performance monitoring, and extends the same user experience across hybrid networks, with no disruption to existing business processes. NETSCOUT’s patented Adaptive Service Intelligence (ASI) technology extracts the most critical information from IP packets with minimum impact on resources. ASI renders rich, scalable metadata for additional insights into subscriber analytics, CEM, Big Data, cyber security and many other use cases.
Visit serviceprovider.netscout.com/nfv to learn more.