Gnome Meeting

Satisfying the New Service Delivery Normal with SDN Management

At Packet Design our main mission is to help operators better manage the complexity of running multi-service networks. Since applications have different performance requirements, growth rates, and fault-tolerance characteristics, running multiple ones on a converged network is especially difficult. SDN applied to the WAN can help address these challenges, as I’ve written about on this blog and at SDxCentral.

Another challenge network operators face is an increase in the rate of service activation and deactivation requests combined with a new normal in acceptable service provisioning times – from weeks to hours, and even seconds. For example, many service providers offer self-service portals that allow customers to request more bandwidth. An SDN controller can automate these requests by issuing commands to the network devices without human intervention.

There are countless other SDN use cases. Before long, automated network provisioning will be done by machine-to-machine communications, using API calls. For example, when virtual machines must be migrated to a different data center, a data center orchestrator will communicate with the WAN orchestrator requesting paths with sufficient capacity to support the move and the communications patterns after the move. Many “Internet of Things” devices will operate this way, and the volume and rate at which services are provisioned will skyrocket.

However, running a software defined network presents management challenges, including loss of visibility into changes taking place in the network and the need to capture engineering know-how in SDN applications. What’s needed is the ability to provision network resources dynamically to accommodate different service types, variable demands, and failures, using the right intelligence. As we wrote in a blog post last year:

“Real-time SDN analytics are critical to enabling engineers to make good decisions. They are also vital to allowing the network software itself to make good ‘decisions.’ If a link performs poorly, an SDN network can route around it – if it knows that the link is indeed performing poorly and what the next best route is. But if the information is incorrect or misleading, the computer will blithely go through its programming, making the ‘right’ decisions for the wrong scenario. Truth be told, a human being could also make the same mistake, given the same data, but computers have the ability to make billions of mistakes per second.”

The Packet Design SDN Platform provides the real-time telemetry, analytics, optimization, and policy to intelligently provision network services. The platform is built on the Explorer suite of products, which network teams have been using for more than a decade to run their networks more efficiently. At the heart of the platform is Route Explorer, capturing real-time routing telemetry from the devices in the network as well as the SDN controllers. As a result, the platform knows the exact network topology and paths used by the services in the network, as well as each service’s traffic flow and performance metrics.

The platform maintains a real-time model of the network, which is important for adapting to unforeseen events in the network. For example, if a link fails and causes congestion, the model will reflect the change immediately, and a risk mitigation application could automatically change the paths for some flows to alleviate the issue. The platform also records all telemetry in its database, enabling network events to be replayed for back-in-time forensics, and also for establishing historical baselines.

However, a real-time model alone is not sufficient, since many applications need predictive network models. For example, if a large amount of data is to be moved across the network, a time slot must be identified when the network will have sufficient available capacity. Predictive models of network behavior use historical point-to-point traffic matrices and baselines of traffic and performance to calculate network behavior under expected conditions in the future.

A path computation and optimization component finds shortest as well as constrained paths in the network. It employs multiple algorithms depending on the service’s requirements. For instance, for a high-revenue premium service, it may use an algorithm to find shortest delay paths. For streaming video, it may use an algorithm that minimizes jitter. Also, when the network capacity starts to run out, it may spread the network traffic to under-utilized links using an artificial intelligence-based algorithm.

The policy component is used to establish the constraints and path characteristics that may be recommended under normal and exception conditions. For example, the policy for a streaming video service might be to allow a percentage of under-provisioning during heavy load periods, because streaming video is adaptive and the video quality may still be acceptable.

Service providers are banking on SDN to deliver accelerated service activation, differentiate their services, provide more resilient networks, enhance customer satisfaction, maximize the return on infrastructure assets, and other benefits. Only by moving today’s management practices into the automation realm will this be possible. We’re already seeing customers use our SDN platform to not only enable intelligent provisioning of network services via SDN controllers, but also to develop their own SDN applications.

For more information, download these white papers:

Editorial Staff

Add comment

Follow us

Don't be shy, get in touch. We love meeting interesting people and making new friends.

Most popular

Most discussed