I noticed an interesting phenomenon at Interop, which sparked my theory on new network technologies. New network technology maturity and its adoption correlate directly to the five stages of loss: 1) denial; 2) anger; 3) bargaining; 4) depression; and 5) acceptance. For example, Interop break-out sessions on cloud and bring your own device (BYOD) now mostly seemed to be mainstream initiatives compared to other technologies, such as software defined network or network functions virtualized. In the mainstream initiative sessions, an aura of acceptance and even tinges of optimism reverberated throughout the room. Presenters spoke passionately and positively about their topics and reinforced the importance of:
Teamwork.Courtney Kissler, Vice President of E-Commerce & Store Technologies at Nordstrom, shared with the audience that the new world is made up of a team of business product managers and mobile app and networking professionals, to name just a few groups working together under the initiative. There was the mentality that everyone is accountable and must work together as a team, helping each other to roll out a great application that will benefit the business.
Historically I have not been a big fan of Interop for a variety of reasons. However Greg Ferro, George Stefanick, Ivan Pepelnjak, and Harvard Business Review IT Director Ken Griffin — to name a few — changed the breakout sessions experience for me. During Greg’s data center network session, he said something that was priceless. He was giving some guidance around refresh cycles and told the audience to not worry about investment protection. It was so refreshing to hear it. I wasn’t the only one nodding my head: Investment protection is hogwash.
Greg made the case that moving to a 3-year replacement cycle changes a customer’s buying and design criteria. Right now, customers have to guess what is going to happen over an 8- to 10-year cycle; this long term guess creates the desire to protect that amount of spending with “investment protection” features. By considering flexibility, growth, and scalability of the network over that period, customers lean towards chassis switches with ports, which can cost 5 to 10 times as much as a pair of 1RU switches with the same type of ports. By selecting chassis switches, Greg says customers have doubled or tripled their project cost.
However on a 3-year replacement cycle, customers can choose right-sized equipment (which is probably a 1RU switch), do less maintenance, and gain faster access to new features. In addition, the risk could be lower. For example if bad decision was made, a company is only stuck with selection for 3 years. Or, the company can choose to replace the network earlier and take smaller hit on capital expense line than if the compay bought chassis switches.
I’m getting a lot calls from clients who are using “white box” and “bare metal” interchangeably when discussing network switches. It might not seem like a big deal, but it is when customers are trying to accomplish a certain task and are examining the wrong products or don’t see the full picture. This is especially true when it comes to assuming that the hardware cost of an Accton 5712 switch is significantly lower than the hardware cost of Broadcom-based switches from Arista, Cisco, or HP. The reality is that pundits are mixing up terms and products.
Fundamentally, they are not making an apples-to-apples comparison and therefore are setting up the wrong expectations for customers. Forrester’s research document The Myth Of White-Box Network Switches dissects the cost of the Accton Edge-Core 5712 (Broadcom Trident II ASIC) switch layered with Cumulus Linux OS and compares this combination against a generic vendor switch built on Broadcom II ASIC, such as Cisco Nexus 3172PQ. We built a model showing the cost of goods sold from the components up and found less than a 5% difference between the switch from the original design manufacturer and traditional network vendor. This holds true when comparing other Accton Edge-Core switches against other traditional vendor Broadcom Trident II switches. The real cost is in the software, global distribution channel, compatibility testing between hardware and software, and global support. “The Myth Of White-Box Network Switches” gets into a lot more detail regarding actual costs.
Wow. Certain networking vendors have started to declare they are winners, while others say software defined network (SDN) is over. All I have to say (in my best George Takei voice) is, “Oh, my!” I’m lucky enough to spend most of my day interacting with many end users to know that those statements clearly show how out of touch some vendors are with customers. Let me make this clear: In today’s environment, only customers can make those statements, and this is probably why some vendors don’t get it. It is a foreign concept and vendors are in the denial stage of loss, losing power to customers.
I’m getting inundated with briefing requests from vendors either coming out with their own white box offerings or somehow supporting the white box market. While white box network solutions provide great value for some industries, more than likely, they are not for your infrastructure but for specific industries such as web scale or high-frequency trading companies. The network world is fragmenting into industry-specific solutions, and the era of Swiss Army knife network hardware is over (see figure below). Mainstream vendors are freaking out because that was their bread and butter. Now they have to figure out who they want to serve. Some uncertain vendors are placing chips on all the squares of the network roulette table; this strategy is a losing proposition for everyone.
Don’t get me wrong. White boxes/bare-metal solutions have their place, but be cautious of the irrational exuberance over this new trend. Resources are finite. The vendors chasing tail lights will at some point have to give up and lock down on a particular path. Activist firm Elliott Management has rattled the cages of some high tech firms and has basically said, “You are killing investors’ return by not simplifying and focusing.” Personally, this might not mean much if you aren’t an investor. However, as an infrastructure and operations professional, you should be concerned about solutions existing a year or two down the road from shotgun-approach vendors; worse yet, you get the effects of a mile-wide, inch-deep investment, which means the solution lags on getting the investments needed to help your company succeed today.
The recent business articles about customers screaming for change, such as Bloomberg’s recently published article about Goldman Sachs’ CIO threatening Cisco, conjures up images of Dee Snider busting through the wall and screaming, “OH, WE'RE NOT GONNA TAKE IT ANYMORE! WE'VE GOT THE RIGHT TO CHOOSE, AND THERE AIN'T NO WAY WE'LL LOSE IT! THIS IS OUR LIFE.” Connecting customers, employees, and business resources has become a life-or-death element for businesses (see The Enterprise Network Enables Business Innovation).
Am I being overly dramatic? I would like you to name a technology that the entire market openly voiced their displeasure about and forced a market leader to come up with a new strategy like Cisco’s Application Centric Infrastructure. Sure, the market has gone through transitions like the movement from fat access points to controller-based access points and the implementation of server virtualization, but the difference between those transitions and the current one is that these technologies were created before customers demanded them.
Now we have customers defining what they want before the technology exists or even creating their solutions, such as:
If a network vendor representative starts off with any of these three phrases — software-defined networking (SDN),bring-your-own-device (BYOD), or lower total cost of ownership (TCO) — I would ask them to leave and come back when they have done their homework on your business. Why? Because clearly they don’t know what your business does and aren’t prepared to help you improve revenue, add new clients, or delight current customers in The Age Of The Customer. The company is treating your team and infrastructure as just a number.
These phrases are all vendor-led marketing initiatives, not customer pain points. Fundamentally, networks should be more than packets delivering PowerPoint slides, connecting users to SAP, or enabling a voice call. Networks touch every part of the business and have significant impact on changing the way business can be done. And the business is expecting to get some business value of out the platform. Therefore you shouldn’t be ok getting a generic networking pitch. You are the customer —make them work for your dollars by making them demonstrate how they can help your business. If you work for a:
Cisco released its 1st quarter financial statement last week, and the numbers weren’t pretty. But this shouldn’t surprise too many, since the company warned the financial community that the revenue growth was going to be below their expectations. Unlike most, I see this as more of an inflection point in an undulation that swings back into a growth mode that comes with a change in strategy than a parabolic upside-down curve. While there are multiple transformations starting to occur in the networking domain, the Cisco Doomsday-ers seem to solely focus on software-defined networking and the creation of cloud infrastructures; they assume the data center of the future will look like Google’s data centers, even though no one truly, outside of Google, knows how it really runs or what the components are.
For argument’s sake, let’s assume every data center (private or XaaS platforms) will be a Google data center full of white-box components and Cisco’s high margin/feature switches will disappear. Does this mean Cisco becomes irrelevant or loses its position as the 800 lb. gorilla in the networking industry? Heck no. What clearly is being missed by most of the world is the incredible transformation starting to materialize outside the data center. And no, it isn’t the presence of mobile devices. That is today’s transformation that changed the consumer. The business will catch up. Tomorrow’s emergence of Internet of Things (IoT) will enable the business to meet its consumers’ desirers, and Cisco sees it already. Cisco could lose every port in the data centers and still be ahead if you look at where the amount of port growth and network revenue will come over the next 10 years.
Ten years ago, if I had stood up in front of IT professionals and said that their company would allow employees to bring their own devices to work in lieu of corporate-owned devices, I would have been heckled out of the room, but look at where we are today. Well, I am here to say that it won’t stop at personal devices or applications. The user edge of the network (where users and mobile devices connect, not servers or storage) is slowly shifting under the control of business and is an integral part of the ecosystem that shapes a customer’s experience. Already, non-IT employees are doing traditional networking tasks like:
Granting wireless network access. Controlling who gets on the network had always been an IT function, until wireless came out. Assistants, business greeters, and other employees can give guests Internet access with all the wireless solutions on the market today.
Setting up networks. Today, manufacturing engineers design manufacturing lines and deploy automation equipment with built-in Ethernet/IP capabilities, such as motion sensors, energy monitors, and logic boards. The design and management of that part of the network falls under their domain.