Demystifying Software-defined Networks: A Decade Later, Where Are We Now (Part II)?
In part IV of this SDN/NFV blog series, I talked about how we arrived to the present situation of SDN, a bit of its history, as well as future plans. In addition, I cited the key implementation of Google’s SDN as an example.
Deployment Example #2
Another company that is making use of SDN is Gap Inc. – yes, it’s not a technology company but the American fashion icon and parent company of Old Navy, Banana Republic and Intermix. They’re using an application of SDN to connect its Internet stores to one another in the corporate network. In the words of the Mr. Patel, Senior Network Architect and CTO of the SD-WAN-SDN project at GAP, “This software approach is about 50% less expensive than the conventional method of connecting stores together in a wide area network. Companies with many stores or branch offices are beginning to look at SDN networking to connect their stores together, but Gap is one of the first to implement this technology at scale and make public its efforts.”
Deployment Example #3
Microsoft Azure is my third example. The Redmond giant is actively implementing SDN for its Azure Public Cloud. In the words of Albert Greenberg, Microsoft lead technologist: “One of the key principles behind Azure’s SDN is its ‘Virtual Layer 2 Architecture,’ a Layer 3 Cross-Fabric that spans the entire data center.” He continues, “Automation is key to managing a massive, high-bandwidth network built with commodity components. The network state service that Azure uses as its control plane abstracts away the individual networks.”
To be able to mix and match the best network element hardware, Microsoft followed a very similar approach to Facebook, using the open source Switch Abstraction Interface (SAI) API to program the ASICs of the switches.
We also have major carriers like AT&T with its Domain 2.0 initiative and Telefonica with Unica. AT&T had a data traffic increase of 100,000% on its wireless network (not a typo) since 2009 to present day. That’s why they needed SDN and NFV implemented across their network because it’s the only technology that allows adding capacity faster with automated deployment and pushes out fast upgrades at the speed of the Internet. AT&T is still planning to have 75% of their network virtualized by 2020.
So What’s the Conclusion?
SDN and NFV are both a bit of hype and disruptive reality. I concur on the view that if SDN and NFV were just bubbles, they would have burst by now. It’s because there’s a real need for both of these technologies that the industry has kept investing in them.
If we think about it, we have been using CLI to configure L2-L3 Network elements for the last 35+ years without many major changes. It is true that we have new protocols, the bandwidth, processing power, and capacity has skyrocketed, as well as the complexity of the networks, but the job of network engineers didn’t change much until we started pushing for SDN/NFV. The paradigm changed – centralized control, separation of planes, abstraction, generic hardware, open source, etc. It is also worth noting that part of the difficulty of its implementation is not just technical, but cultural: In most organizations, networking is typically siloed from the rest of the IT organization. The new approach to networking, SDN and DevOps, requires a different mindset and it takes convincing, training, effort and time.
According to a report from BCC research, the estimation for SDN global revenues will jump to over $56 billion in 2020, plus, in the near future, 100G will be the norm and manual control won’t be enough. We will need to have automation all across the network from the top to the bottom. The agility that SDN and NFV technologies provide will be key soon.
I believe these technologies have spearheaded the biggest leap in networking technology over the last 20 years, it’s just taking a bit longer to completely take over. Once we have the problematic interoperability and standardization figured out, we’ll have massive implementations across the board.
We have many initiatives, like ONAP and OPNFV, just around the corner. ONAP (a project combining AT&T’s ECOMP and Open-O) provides for automatic, policy-driven interaction of these functions and services in a dynamic, real-time cloud environment. It’s not just a run-time platform; it includes graphical design tools for function/service creation, delivering capabilities for the design, creation, orchestration, monitoring, and lifecycle management of VNFs, SDNs and high level services. OPNFV is focusing on the higher layers with quality open source software for the virtualization layer, specifically on the ETSI NFV interfaces VI-Ha, Vn-Nf, Nf-Vi, Vi-Vnfm, and Or-Vi in the diagram above (from ETSI architecture).
As a closing thought, I’d say 2020 will be the date for the massive implementation of SDN/NFV. We’ve come a long way in just 10 years but work needs to be done on the higher layers in particular; on orchestration and consolidation of the platforms we have in place, and improving the interoperability on automation features.
Part of the Blog Series