As with all technologies, big developments are often accompanied by equally big levels of hype. Gartner calls this the “hype cycle.”
Several Technologies such as IPV6 received huge projections and praises during their “hype-cycle”, only for the industry to see little or no reward. This reminds me of a common saying in the consulting circle that goes like this: “everything works on Power point, some things work in the lab and most things seldom work in production.”
Software Defined Networking SDN is currently going through its own “hype cycle”. The good news is that we are slowly approaching the peak of the hype. Afterwards, the true value of SDN will emerge in the next phase called the “Reality phase”.
To prepare for the “Reality phase”, it is important to review the fundamental challenges in networking that the SDN inventors intended to solve. In this two-part blog, I will review those fundamental challenges and how we have managed to live with them during the pre-SDN era.
Gartner’s chart
Over the last decades, there have been several innovations in networking. Innovations such as MPLS-Traffic Engineering, NAT, VxLAN etc have changed the way we design networks.
On the positive side, these innovations have been impressive and solved real business challenges in the field of networking. Using MPLS-TE as an example, I can influence the path taken by traffic through my network thus offering new services with extra guarantees and lowering the investment in new network resources as a result of an improved utilization of network resources.
In contrast, these innovations have also come at a cost. The increase in cost of the network (specialized equipment needed for MPLS-TE, NAT, VxLAN etc.) and the complexity associated with deploying and managing these technologies are some of the downside. From the operational side, the average network engineer has become an expert in mastering complexity rather than extracting simplicity to solve real business challenges.
Ironically, most of these creative innovations, over the last decade were really meant to mask (not solve) the fundamental problem in networking. The fundamental problems that prompted innovations such as NAT, MPLS-TE,VxLAN etc are still present in todays network. This is because the foundations of networking are weak and rather than address the weak foundation, additional layers of complexity have been added to keep our shops open.
The foundational problem is primarily described as the lack of abstraction. To put it differently, network is too rigid and cannot move at the same pace with the business.
So what exactly is abstraction? Why do we need it after doing just fine in more than two decades? How do we go about implementing it? These are fundamental challenges that SDN inventors came together to solve. I will review the questions in part two of the blog.