 | Issue: | Europe II 2014 |
| Article no.: | 11 | |
| Topic: | Software reboots the networking industry letting users innovate again | |
| Author: | Jeff Paine | |
| Title: | VP of Marketing | |
| Organisation: | Vello Systems | |
| PDF size: | 365KB |
About author
Jeff Paine, Vice President of Marketing at Vello Systems, has more than 20 years of experience as a high-tech marketing and business development executive. He began work in the networking industry in 1989, when he joined pre-IPO Cisco Systems as its initial marketing hire in preparation of the public offering. After six years at Cisco Jeff went on to hold a number of marketing, product management and business development executive roles at both start-ups and large companies in the networking and telecommunications industries. Most recently he was Vice President of Product Marketing for New Zealand-based Endace (recently acquired by Emulex). He has also served as Vice President, Global Marketing, for Vividas and Vice President of Strategic Marketing for UTStarcom.
Jeff holds a BA degree in Communications from Brown University and an MA from the University of Texas at Austin.
Article abstract
The promise of SDN is not in the core of the data centre, but at the data centre edge, where interconnect issues are painful and high bandwidth is costly. If enterprises can self-provision what they need, when they need it, and service providers can boost their networks according to real demand, both sides win. The new software automates SLAs and allows predictive methods to provision capacity, leaving staff to focus on innovation and business. The new white-box switches also come with an ‘open’ approach – self-sparing of common components, which reduces both the CAPEX for SDN-enabled equipment, and the OPEX for its maintenance.
Full Article
2013 was a watershed year for the broad group of companies in the overall software-defined networking (SDN) space that have been promising – and, often, overpromising – solutions for most of this decade. Now, a small number of those companies have successfully leveraged the overall SDN framework and associated architectural concepts to produce solutions that are in full commercial deployment at large enterprises and service providers. Deployments that are solving painful problems are left unaddressed by legacy networking technologies. No more slide-ware… no more POC (Proof of Concept) deployments… only real solutions.
While this may come as a bit of a surprise to some, to be fair, the widely held perception that SDN is a technology that is not quite ready for primetime is largely a self-inflicted wound on the part of the overall community of SDN vendors. We have a saying inside Vello: when even the smartest person in the room is confused by what SDN means, then you know that the SDN hype machine has been left on in overdrive for far too long.
The real market requirement for software-defined solutions has come about almost by accident. Because the legacy L2/L3 networking vendors have been so wildly successful in their sales efforts, the top companies in most market verticals, literally have the same network with the same capabilities as all of their competitors. So differentiation between competitors based on technology has, well, fossilized.
Also, adding to the confusion about SDN has been the dive to the bottom on price among traditional L2/L3 core data centre switch vendors that wrap themselves in a patina of SDN or ‘openness’. In reality, they have entered into an epic price war and are offering ever lower price-point solutions than their competitors (with a few new networking software capabilities) in the heart of the data centre. Even start-ups have now entered this game with a cut-rate version of what the traditional vendors offer, only this time based on commodity white-box Ethernet switches.
The promise of SDN, which has manifested itself into real solutions offering new capabilities, is not in the core of the data centre, but rather at the data centre edge. This is where the action is for a number of reasons.
First, the issues revolving around data centre interconnect are typically more complicated and more expensive than those in the data centre core. In the core, bandwidth is relatively cheap and overprovisioning is the standard answer to most problems brought about by the growth of Big Data. However, throwing additional bandwidth to solve the problem at the data centre edge can be excruciatingly painful and costly. Bandwidth between countries can be extremely expensive, and deploying additional inter-country WAN links is simply prohibitive.
Next-generation solutions that are available now utilize an open set of software-defined products, where enterprise customers can actually self-provision what bandwidth they need from a network service provider, based on time of day or other parameters. In this example, ‘by the drink’ pricing automates the entire SLA process. It allows enterprises to be released from long-term bandwidth contracts that give them more bandwidth than they need, while concurrently letting the service provider intelligently overprovision the WAN links and fully monetize them 24×7, through tiered SLAs. What’s more, these new services can be set up and provisioned in a matter of minutes now, instead of weeks. This is a win for both the service provider and the enterprise.
The software-defined movement has brought innovation back to the table and benefits the service provider and enterprise alike. This new breed of software automates SLAs and allows for new, predictive methods to provision capacity, and provides foresight into the revenue stream and capacity needs of providers, while guaranteeing enterprises the bandwidth that they need, at the price they want pay. Networking IT staffs are shrinking as DevOps staffing increases, placing headcount in places that will enable even more business innovation. Previously, these new capabilities were simply not practical to offer, based on legacy networking solutions. Recently a Global 200 retailer in Europe, a Top ten US bank and the #1 provider of inter-country bandwidth and data centres in Asia-Pac have all started deploying solutions based on this new software-oriented paradigm.
Second, with the rising availability of networking software and of inexpensive white-box Ethernet switches (that are fundamental to the cost-reduction part of the SDN promise) the technology has matured to the point that these ODM (Original Design Manufacturer) devices can handle the level of data flows found at the data centre edge today. Data-flow traffic levels between data centres are well within the sweet spot of current OpenFlow-enabled ODM Ethernet switches. In contrast, when these switches are deployed as TOR (Top of Rack) devices in the data centre core, they are configured as standard L2/L3 switches, with no flow-based capabilities. They will get there as the technology curve continues… just not now.
There is nothing wrong with the old, legacy L2/L3 networking equipment powering most data centres today. It is simply that these models — founded on good engineering and practices in their day — have now been surpassed by modern manufacturing and new business practices. The fact is that most networking devices have almost 100% common components, as the low cost of these components is now impossible to match by any single system vendor when building them themselves. This is why the merchant hardware that is used by ODMs to build white-box Ethernet switches, produces solutions with US$50-per-port pricing versus the marked up US$500-per-port for the same components on a branded platform.
So the promise of software-defined networking not only resuscitates innovation and differentiation among technology consumers, but it also radically disrupts both CapEx and OpEx business models. For example, white-box Ethernet switches – and, soon, equivalent optical systems based on standardized components – are so inexpensive that self-sparing becomes both easy and desirable. With self-sparing, the requirement for expensive hardware maintenance contracts quite literally disappears.
In a sense the movement toward this new open model is irresistible, and ‘open’ is the pivotal word in that sentence. It allows for self-sparing and interchangeable sparing – both of hardware and software – as well as significant operational improvements. It provides a real platform for innovation as it lets companies focus on customer services through standardized and simplified hardware and software.
Data centres will never be the same — nor should they be – as the ability of these new SDN-inspired software solutions to rapidly deploy innovative services – often in a matter of days — is turning the existing data centre business model into an expensive evolutionary cul-de-sac.
