New Perspectives on SD-WAN: An Interview with Stefano Gridelli

SD-WAN deployment has accelerated in recent years as organizations extend SDN benefits across wide area networks. It’s a pretty transformative process, producing new management and cost benefits along with new user experience and performance demands.

A few weeks ago I discovered NetBeez, a network monitoring company with a unique, proactive, user-centric approach to monitoring these more dynamic networks. Their hardware and software sensors are deployed at the edge, including before SD-WAN deployment,  to assess MPLS vs internet tradeoffs, from a user’s perspective.

A recent SDX exchange on the future of SD-WAN late last year prompted me to ask Stefano Gridelli, founder and CEO of NetBeez, three questions about SD-WAN monitoring. His perspective has been shaped out of network engineering roles in health care, which inspired him and his team of founders to introduce a better way to monitor SDNs and SD-WANS.

If anything, industry speculation at the end of 2019 about the “cloudy” future of SD-WAN brings new questions about gaps between new forms of dynamic network infrastructure, existing tools and practices and the evolution of careers in networking:

Q) Why is SD-WAN different when it comes to monitoring?

[Stefano] SD-WAN is a game changer in terms of network management. Benefits of SD-WAN include ease of configuration and operation, cost reduction from the use of Direct Internet Access (DIA) mixed with traditional transport technologies (e.g. MPLS), and centralized management. 

In terms of monitoring, most SD-WAN solutions have network and application visibility tools that provide statistics about top users and top applications. These statistics are collected by analyzing traffic traversing the SD-WAN router’s interfaces. The problem with this “passive monitoring” approach is that it doesn’t really build a network and application performance baseline (no user traffic, no data), and also reduced proactiveness on performance issues.

Another challenge of monitoring SD-WAN installation is that it makes use of tunneling, split tunneling, and virtualization. Since user traffic is dynamically routed, sometimes on a per-packet basis, across multiple lines, it is more difficult to pinpoint the root cause of performance issues. With split tunneling, users may use the Internet connection to browse public or SaS applications, reducing visibility into the end-user experience from the centralized NOC.

Q) Do you think SD-WAN will be commoditized by the cloud or become more strategic?

[Stefano] I don’t believe public clouds will completely replace private data centers. There is no doubt more companies today are running a fraction of their compute workloads in AWS, Azure, or Google Cloud. Yet, I don’t see the future being run 100% on public clouds. I believe the hybrid multi-cloud model is the future. For that reason, I see SD-WAN supporting hybrid multi-clouds, and we will see cross pollination between networking vendors and public cloud providers. We’ll also see more startups in this space than before, thanks to the decoupling of hardware and software, and software companies like VMware, which has been mostly playing in the virtualization market, tapping this opportunity. To conclude, I still believe SD-WAN will become more strategic, so I differ with others.

Q) How will SD-WAN change how networks are managed?

[Stefano] SD-WAN simplifies WAN configuration and management. In traditional WANs, configuration and troubleshooting was mostly done via a command line interface, one device at the time. SD-WAN equipment are centrally managed from a web interface, and applying consistent network and security policies is much easier. This advancement requires less skilled network engineers to operate SD-WANs, and I am sure that the larger the network, the higher the savings. AT&T for example is planning to cut over $1.5B in labor costs in the next few years.

Will network engineers be the casualty of software-defined networks, similar to what happened to switchboard operators last century? Network engineers are here to stay, at least for a while, but their job descriptions will change. Their roles will evolve into NetOps, and it will require a basic knowledge of the Linux operating system, of the Python programming language, and of APIs in general.

 

Thank you Stefano! For more information on network monitoring for SD-WAN, check out NetBeez. They were recently called out for having a top blog in network monitoring and management.