I remember my first time surfing — Ken Hook took me out to Linda Mar in Pacifica, California early on a chilly Sunday. I bought a wetsuit, borrowed a 9’6” ‘soul board’ from Ken and proceeded to inhale enough water to fill a bathtub. I got pummeled on the swim out and turned into a washing machine, never understood how to ‘turtle’ to take a big wave, and maybe got to a knee coming back in two hours later. I didn’t surf again for twelve years.
The swim out is the hardest part for me and I have always been a fairly decent swimmer, but swimming with waves crashing into you, getting pushed back 2-3 lengths for every 3-4 you go forward can beat you down both mentally and physically. There is a magic spot though, once you get past the waves, where you can sit on your board and feel the burgeoning swells grow under you and pass by without crashing on top of you. Whether you meditate, do yoga, or just find a shady spot and daydream — that spot past the blast zone is a similar place, one that lets you experience the power and the tranquility at the same time. It is a true gift.
I spent quite a few hours this week talking about the metaphorical swim out, what are the big waves coming in the networking world and how can companies swim with them or avoid them, and which ones may just take you out. Waves come in sets, and we are seeing a few sets in networking — the question that should be on every entrepreneur’s and investor’s mind is how do we catch the wave, avoid it, or can we survive being hit by it? Some are obvious, some are just developing and we’re not sure if they are going to be a rogue wave or not — time will tell, but all are worth some of our attention.
The wave set we’re surfing with or fighting against right now is comprised of three close waves: Cloud, Software Defined, and Network Virtualization. These are transforming IT and Infrastructure and it is not uncommon to hear a CIO announce a ‘cloud only’ or ‘cloud first’ strategy today; however, at the same point we are seeing the largest SaaS companies and other firms with ‘production engineering’ start to bring some core applications and functions that were born in the cloud back in-house. While I don’t think bringing everything back in-house is for everyone, it is interesting to watch and observe as company’s determine if infrastructure ownership can equate to competitive business advantage.
The First Wave: Disaggregation
Disaggregation refers to the concept of deploying switching hardware from Vendor A and Software from Vendor X. A common thing I hear from server/OS types is, “We’ve had this for decades in servers and Linux so why is it taking so long in networking?” This often has a follow-on epithet with colorful phrases implying us network types are some obscure form of Luddite/troglodyte.
What I see happening in disaggregation is the large hyperscale customers are making support for disaggregation a purchasing criteria, thus we are seeing customer-appeasement as a strategy from the major networking vendors. To call the support for disaggregation from a company with a blended HW/SW commercial model ‘begrudging’ would be like calling a Hollywood divorce ‘amicable’. The networking vendors currently have a leg-up though — they have extremely well vetted and proven software quality that customers do require — so the vendors are disaggregating by ascribing 80-90% of the system value to the software making the net-net of disaggregation result in no appreciable pricing difference for the customer.
I see the big hyperscale customers launching phase two soon — look for investments and structured spending on network OS vendors. By voting with their wallets the hyper-scale buyers will spend and invest to create viable software alternatives to today’s incumbent HW/SW. This is not dissimilar to what we saw when the large telco providers invested early in Juniper in the late 1990s in order to create a viable competitor to Cisco’s routing business — smart procurement and CIO/CTO types can recognize an opportunity to take down 5-10% of their CAPEX and apply the savings to further OPEX refinements that can deliver lasting shareholder value.
Wave 2: Point and Click Millennials
Ever go windsurfing? Look around at the other windsurfers and notice that they are all ‘our age’ while the cool kids launching 50’ aerials off of 2’ waves on a kite are 20-years younger than us? The kids kiteboarding are millennials, and they don’t like CLI. (Note: I learned to kiteboard last year, it is &%#$% hard, it is also wickedly fun.)
However we are building our products today, we need to understand that this is a market that has a discrete skills gap and a generational gap in how people want to manage and operate their infrastructure. As my friend Greg Ferro @EtherealMind posted on Twitter, which is part of my inspiration for writing this piece. Thanks, Greg!
Greg told me he would tell me how to do this for $2k a day (which is far more than I can afford, even as much as I love his company!) so I am taking a guess that he wants to deliver a network that self-operates. It is my opinion that he is advocating for a network where the management, operations, and automation are built into the construction method of the network itself and not a ‘bolted on’ approach. This is going to require vendors to work together, not be isolationist, and stick to their stated values when it comes to ‘being open’ and building APIs that you actually want used.
The type of engineers who are entering the jobs market from colleges today are decidedly unfamiliar with networking, with CLIs, and with rather archaic things like SNMP. A network OS designed for and by engineers and operators today will be one that embraces the need for both a human-to-machine interface and an equally capable machine-to-machine interface while supporting the distributed and federated systems operation that made networks reliable and resilient in the first place.
A smart system would also recognize the governance and regulatory compliance models requiring discrete separation of roles and responsibilities has obsoleted 90% of the AAA models we built into our networked systems. I believe that a networking vendor who can embrace a roles based administrative model will create a competitive advantage for themselves in any regulated industry.
Embrace the skills gap and build products and systems that help regular network engineers do cloud-like hyper-scale awesomeness. Embrace the generation gap and build systems that are open enough to become platforms for transforming the business and transcending the cost-center of infrastructure to play in the wildly profitable and wonderful world of business relevancy.
Wave 3: Buyer Shifting
I have spent the last twenty years of my life, starting shortly before I learned to surf (or swim while swallowing gallons of saltwater depending on your perspective), building infrastructure that was primarily purchased and consumed by ‘the network gal, ‘the server guy, ‘the storage gal’, etc. Over the past five years I have watched a sea-change (pun intended) in new technology insertion: as developers were able to mimic infrastructure capabilities through accelerated software, faster CPUs with more cores that could be dedicated to these functions, and richer programmatic interfaces we have seen the developer emerge as the key buyer of new technology in the enterprise.
If the developer buys early but is fickler, less brand conscious, and eye-balling short-term development agility gains over longer term enterprise vendor viability, what does the selling motion and product absorption look like within a traditional enterprise infrastructure? How should a new company embrace this change, and achieve both insertion and enterprise scale deployment? I don’t know the answer today, but this is something that I feel every new company looking to deliver their fancy new software-defined, virtualized, cloud-native and ready, container-based, bad-ass networking thing to market needs to consider.
On the flip side of this argument my friend Guido pointed out to me last night that new waves ALWAYS start outside of IT in the business — the PC revolution for instance. Then the ‘business’ makes the same mistakes they have always made in forgetting security, governance, change control, etc and IT is called in to support it and add process, to bring order from chaos.
The other question is simply, who is my best buyer? If the answer is, “everyone needs this” then unfortunately, ‘everyone’ has the ability to vote No to your widget too. I advise early-stage companies to pick a single buyer and make that person really happy and an absolute winner with your solution. The more people who your product involves, the less likely you are to win any given opportunity.
As an entrepreneur, go with a ‘high and narrow’ insertion — deliver a lot of value to one person or organization. As a technology buyer, always challenge vendors to prove to you that the product or capability will make your life better.
In 2005, Yvon Chouinard, famed alpinist, and Founder of Patagonia (great wetsuits btw, and I loved their Gore-Tex parka in the Himalayas last year and on The Great Ski Race from Tahoe to Truckee two weeks ago) wrote a book titled, ‘Let My People Go Surfing’. When devising a product or corporate strategy this is an important concept — let your people go surfing. Look at the market landscape and what the big wave sets are, and plan your product design to surf the wave rather than swim out or upstream. If you can truly capture a wave, you are capturing a market in transition. That’s when you go from chasing mavericks to riding them.