April 18, 2017 | IT & Telecom
Back in 2011, when the Open Networking Foundation (ONF) had just come to the fore, I had a discussion with the senior leadership of a premier networking organization. I specifically happened to ask them about the likely impact on their product portfolio, margins and GTM strategies given that enterprise buyers had a possible choice emerging to do away with something they had been long complaining about − vendor lock-ins and high CAPEX. The reaction from leadership was among expected lines. They didn’t want to admit that the open source community had any business in the enterprise networking domain, which was considered a sacred arena for the closed-platform behemoths such as Cisco, HP, Juniper and Huawei, among others. Fast-forward six years later, and they will hardly be able to sport the same attitude towards the concept. Thanks to ONF and Open Compute Projects (OCPs), the enterprise hardware landscape will never be the same again!
We want choices as a buyer – whether it’s basics and essentials, lifestyle or luxury purchases, or business purchases. Yet, the enterprise IT buyer has compromised for far too long when it comes to purchasing networking equipment and has paid a rather hefty price (literally!) for doing so.
A few select suppliers dictating what, when and how enterprise buyers purchase, deploy and upgrade networking hardware is simply not acceptable. I have long resented (earlier as a focused networking analyst with a leading analyst firm, and now as a business advisor to enterprise buyers) the fact that limited suppliers have had too much control in dictating our choices over the years. But thankfully, such supplier dominance over buying choices are numbered with the advent of initiatives such ‘Software-Defined Networking (SDN)’ and ‘Open Compute Projects (OCP)’ which has resulted into white-box and brite-box equipment.
White-box and brite-box switches are welcome departures from traditional layer 2 and 3 ASCII switches which had the control plane, data plane and the network operating system (NOS) as part of one switching device offered by a supplier. The suppliers’ control over the three aspects – control and data planes, and the NOS – allowed them to dictate choice and prices, and led to the very large premiums that the buyers had to shell out.
However, with the advent of the open networking forum / open source community-backed SDN and OCP initiatives, white-boxes and brite-boxes have started challenging the dominance of traditional switches. White-box switches are a by-product of SDN, which decouples the hardware (box) from the network operating system (NOS), allowing buyers to procure network switching equipment and pair it with their preferred NOS. This approach has led to the networking hardware being considered a ‘dumb’ commodity and has significantly lowered the cost of procuring network equipment.
White-boxes coming from traditional networking operators (Cisco, HP, Juniper, etc.) who have had to tweak their approach in wake of the increasing popularity of the white-box approach proposed by the hyper scale operators (Google, Facebook, Amazon) are labeled ‘brite’ (branded + white) boxes. Again, the savings prospects resonate closely with that of its cousin’s (white-box).
The future of networking lies within the realm of ‘white’ and ‘brite.’ I certainly see buyers progressively switching away from the traditional, integrated ASCII hardware to white and brite boxes. However, the transition is going to take time and would take shape over the course of next two to five years. The opportunity to generate hard savings with capital cost reduction in the range of 5X-7X per port, reduced supplier lock-in and increased flexibility and programmability are certainly too appealing to turn away from.
Now, between the white and brite approaches, I certainly see the latter emerging as the preferred choice for enterprise buyers. The white boxes are no panacea and there are massive barriers for mainstream adoption: managing the acquisition process, integration challenges with existing infrastructure, support, to name a few. This has restricted the adoption of white-boxes to hyper-scale cloud service providers. The brite-box approach has certainly enabled hyper-scale switching concepts to transition to the mainstream, with buyers having an array of deployment models to choose from:
I certainly believe that the future of networking technology is in brite-boxes! Network ecosystems will increasingly transition toward a hybrid culture, including both traditional integrated switching and the brite-box approach, with brite-box gaining increasing relevance by the day.
While a potential savings range of 30% - 50% comes across as very appealing, procurement does need to adopt a measured approach to adopting the white / brite box approach. This is not the sort of change to be undertaken for sheer novelty. Procurement should adopt a staggered approach toward introducing brite-boxes into the network architecture.
Hardware category managers will certainly need to adopt a very thorough approach in calculating the TCO for the entire solution. While the white box hardware switch will come cheap (when compared with traditional branded switches), the NOS that runs on the hardware will vary in price and buying model – certain NOS have subscription plans, support contracts, staff training and integration issues that need careful analysis and consideration.
The approach to brite-box adoption needs to be a staggered one rather than one that is big-bang. Having said that, however, there are enough reasons for procurement to start warming their suppliers to the prospect of the brite / white approaches (not to mention the potential to leverage additional cost savings on contract renegotiations).
Here’s to a BRI(GH)TE tomorrow!