Customer Experience Management (CEM)

NetScout: Service Assurance Still a Must for Virtual Realm

DENVER -- Cable Next-Gen Technologies and Strategies -- Service assurance becomes even more essential in virtualized networks, and it must be expanded to cover both physical and virtual resources, NetScout's Vikram Saksena explained to an audience of cable executives here on Tuesday.

Saksena, who is chief solutions architect in the office of the CTO at NetScout Systems Inc. (Nasdaq: NTCT), laid out the case for why it is more important than ever to have an end-to-end view at the service level as some functions that comprise services are virtualized. But, he added, there are significant challenges to creating service assurance in a way that doesn't overburden key network elements such as hypervisors, which aren't designed as high-performance vehicles.

The move to a software-centric network can make services "more maneuverable, manageable and malleable," which improves service velocity, he notes. Additionally, virtualization can lower the total cost of ownership for network operators.

"Service assurance is a very foundational capability for this, because as networks become more agile and dynamic, you still have to give the customer what they want and you can't compromise on user experience and what you deliver," Saksena said. Service assurance actually becomes a bit more complex in the near-term because networks will be a combination of physical functions and virtualized network functions "for a long period of time."

"Service assurance will be more real-time on applications and resource management," he added. "As the networks are becoming more agile and changing more dynamically, you need something that provides feedback so you can do reconfiguration in real time, even as the functions become more distributed."

Saksena called assurance one of the four foundational capabilities of a software-centric network, the other three being virtualization, abstraction and orchestration.

This next generation of service assurance will mean a combination of probes, the monitoring devices deployed in the network to provide feedback, he said. More traditional physical probes will still be needed to monitor what he called "north-south traffic," or what is moving between physical boxes and the hypervisor and virtual switches, which are software based. Virtual probes, or vProbes, are software-based clients that are used to monitor "east-west traffic" as it moves among the virtualized elements that are part of a service chain.

NetScout's View of Combined Physical, Virtual Probes
Source: NetScout
Source: NetScout

There are challenges to this process that include not allowing the insertion of probes and their data collection/reporting processes to affect service performance, Saksena said. There is a need for real-time analytics in this process so the probe must not only collect data but compute local performance parameters, requiring intelligence of its own, he said.

"The challenge is, how do you extract meta-data and compute KPIs [key performance indicators] locally and only send a fraction of the data northbound?" he said. "Having distributed processing and stateful failover becomes very important. Monitoring and service assurance become a very significant challenge to provide the quality your customer expects."

Want to know more about vendor NFV strategies? This will be just one of the many topics covered at Light Reading's second Big Telecom Event on June 9-10 in Chicago. Get yourself registered today or get left behind!

For services such as Carrier WiFi, business voice and multi-tenant cloud service, the combination of physical and virtual probes knitted together in an end-to-end service assurance process becomes essential, Saksena said. As service providers offer hosted VoIP to businesses, for example, they can virtualize the IMS components on the back end and then combine probes of the CPE, network and IMS components in a service assurance layer to monitor end-to-end performance of the overall service. For multi-tenant cloud services, the probes need to be set up on a per-customer basis so that even traffic on the same virtualized server can be monitored in a separate, secure fashion.

Saksena showed NetScout's involvement in a large "network on demand" deployment -- assumed to be AT&T Inc. (NYSE: T)'s rollout in Austin, Texas but not identified as such by the NetScout exec. In this case, virtual probes are deployed in the operator cloud and in virtualized CPE, and physical probes are used in IP network elements. The service is used to adjust bandwidth up and down as needed by applications.

— Carol Wilson, Editor-at-Large, Light Reading

Be the first to post a comment regarding this story.
Sign In