Service providers tend to be early adopters who lead the way deploying new networking technologies. In particular, their creative and resourceful approaches to the new frontier of SDN are very interesting. We are working with numerous service providers to deliver SDN management solutions and have been surprised at how many unique use cases they have brought to us. We’d like to share these with you all over the next several months.
Running networks hotter is certainly not a new goal. Traditionally, service providers run their networks at about 45 percent. This provides enough capacity in case there is a failure. This also represents billions of dollars in underused assets. One way to reduce costs, which is a high priority for many of our customers, is to unlock the investments they’ve already made in their backbones and networks and thereby save on additional capital expenditures.
The goal for many service providers is to run their networks at around 70 percent. SDN can make this possible with self-reliant, self-healing networks. However, keeping utilization high can only be accomplished with real-time and predictive analytics. Using SDN controllers fed by this intelligence, service providers can ensure that any automatic changes will keep network applications and services running smoothly.
This is why so many service providers are testing and deploying the Packet Design SDN Analytics and Automation Platform. Usually, this is as part of a multi-vendor ecosystem, sometimes involving multi-layer orchestration, but more on that in a later blog. With our Explorer Suite as its foundation, the SDN Platform captures real-time network telemetry from the devices in the network as well as SDN controllers. As a result, our platform knows the network topology and paths used by the services in the network, as well as each service’s traffic and performance.
From this, a real-time model of the network is built, which is important for adapting to unforeseen events while keeping utilization high. For example, if a link fails and causes congestion, the model would 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 for back-in-time forensics and for establishing historical baselines, which are important when setting SDN automation policies.
The Platform has another fundamental building block: its path computation engine (PCE). It can simulate modifications to the network topology, service traffic, and performance measurements and then analyze their impact. This way, the network can be configured without risk, either manually or programmatically. For example, it is simple to simulate moving a service running on VMs from one data center to another and see the impact in the WAN (e.g., Whether or not the service’s workload caused any congestion and impact on other services).
Using the intelligence provided by its real-time telemetry, historical and predictive analytics, PCE, optimization algorithms, and user-established policies, the SDN Platform makes network configuration recommendations to accommodate application requests. The Platform’s provisioning component then instructs one or more SDN controllers via their northbound APIs to configure the recommended changes. This process can be fully automated, or operators can view the recommendations, select or deselect certain ones, and then make the changes with a single mouse click.
The first application we’ve built upon the Platform – the SDN Traffic Engineering (SDN-TE) application – uses this management intelligence to help service providers run their networks hotter. The application automates the difficult and time-consuming task of balancing network loads. It does this by creating Resource Reservation Protocol (RSVP-TE) or Segment Routing (SR-TE) tunnels to shift traffic from heavily congested links to lightly used links. This results in better overall use of network resources and smoother service delivery.
Everyone is looking for the business case for SDN. Running networks hotter is a good one. Instead of spending time and money padding in network layers and infrastructure, service providers can use SDN orchestration technologies to bring utilization up across their networks – but only with the management intelligence needed to compute the right changes.