I recently read a blog post that Greg Ferro (@etherealmind) had posted up on Twitter, and it stirred something up inside me. Joe Howard is the Technology Market Builder at Brocade (I’m not 100% sure what this means, but sounds like sales to me), and while we’ve never met – you’ve got it all wrong, my friend.
Here is a link to the article Joe published. Give it a read before you continue further.
I’ve been a network engineer for the majority of my career, and while not as seasoned as some of my peers who are pushing 30+ years, I am quickly approaching the 20 year mark. I first cut my teeth on Novell systems administration circa NetWare 3.12, installed (and later ripped out) ArcNet, Token Ring, and 10-Base-2 ethernet, so I’m not what you’d call a ‘millennial engineer’. Disclosure: These days I work for a Cisco partner, but I was a customer for a long number of years. I’m not in sales, and I’ve got no ulterior motives – these are my opinions as an engineer.
In the past 18 months, I’ve migrated my personal IT environment to a MacBook, an iPhone, and an iPad (away from Windows and Android OSs). Why? The usual reasons – better reliability and security, a superior user experience, near seamless interoperability, and they “just work.” I really like my Apple products. No wonder Apple has been the most important technology company, by a long shot, for several years.
I couldn’t agree more, Joe. My first attempt at making the switch to Apple was when the mini first came out (I actually returned it), but it just didn’t work for me – that is, not until the Intel processor. Add a couple years of product evolution and virtualization to the equation and I was sold and have never looked back. Now I own 3 or more of just about every device Apple has made, and I pretty much love them all. You’ve nailed it with ‘it just works’ – I’ve been referring to this as ‘the apple effect’. However I’ve got to disagree with much of what you have said in the rest of your article, and I’d like to call out some quotes, if you don’t mind, and argue a few points. No offense, but I see things a bit differently.
Ethernet and IP networking is embarrassingly complex, unreliable, arcane, and parochial. That results in very high operational costs, poor security/high vulnerability, and nothing close to five nines reliability.
I’d have to disagree entirely. In comparison to legacy technologies, Ethernet and TCP/IP networking has been some of the most reliable technology driving global communication over the past 30+ years. It’s gone through a consistent evolution (both speeds and feeds, as well as upper layer protocols), but one thing has been relatively constant – the concept of architecture. Designing enterprise-class networks for much of my career, I’ve seen the exact opposite. The networks I supported weren’t embarrassingly complex. They weren’t unreliable, in fact, they had some of the highest levels of availability I’ve ever seen. It wasn’t because I was some rockstar engineer, or because I bought the latest and greatest whiz-bang product – it was because of architecture and strategy. A solid design, based on solid fundamental principles, with a well-defined scope of operation and service-level. Did I use Cisco gear – absolutely (among others), but not because they were the 800b gorilla. I used it because they had an end to end architecture strategy to go with it, and following that design philosophy allowed me to get the most out of the equipment I was purchasing.
Cisco has no incentive (in any serious way) to innovate and make the technology inherently more automated, standardized, and easier to operate. Their market dominance is dependent on complexity.
Cisco’s market dominance isn’t dependent on complexity – it’s the result of decades of product development, and furthering of the architecture strategy I mentioned previously. The technologies we’re using today (and I’m not talking proprietary Cisco technologies, I’m talking industry-standard protocols) weren’t made up overnight; they underwent continual development over decades, sometimes stumbling and failing, but ultimately resulted in the production of many of the core networking technologies we rely on today to keep our networks running. Highly-available enterprise architectures aren’t ‘overly complex’ by the sheer nature of being big – in fact, its quite the opposite. Simplicity is often a dominating factor in the most reliable of networks. Simplicity in itself IS a strategy (I’m sure you’ve heard of the KISS approach) and I’ve found that when you use simplicity as one of the building blocks in your enterprise, you can reap the rewards from it.
Most network engineers have achieved Cisco certifications. Cisco trained engineers will be heavily biased towards Cisco products.
How is this any different from what we agreed upon in the opening of this post – we both love Apple products because they ‘just work’. We aren’t biased towards them because Apple makes them (not directly); we’re biased towards them because we’ve used them, and the experience we had was a good one – again, ‘the apple effect’. I’ve had that same experience with enterprise Cisco networking gear. I’ve also spent a lot of time ripping my hair out trying to make things work on gear made by those other manufacturers who only have that 3-4% market share.
Furthermore, Cisco trained engineers embrace the complexity, for the obvious reason that their skills become special, valuable and well compensated.
No, actually we embrace the reliability and consistency we experience in deploying Cisco enterprise networking solutions. When something works well, and we can do it over and over again in a dozen locations without having to worry that it “won’t work this time” THAT’s what we embrace. None of us want to spend our weekends sitting on the floor of the data center troubleshooting a production-impacting issue, we all prefer technologies that ‘just work’.
In fact, I really don’t think your article is about networking technology much at all. You’re using it as a guise for calling their baby ugly, and I’m calling bullshit. People use Cisco because it works. Am I a fan-boy – ABSOLUTELY; because time and again, as a customer and as an end user, they’ve earned my loyalty – but they didn’t do that by being a marketing machine, they did it by making available to me mature, time-tested and field-tested solutions that in most situations ‘just worked’. I haven’t given them a free pass because they were the incumbent (although it did earn them a seat at the table for conversations when the time to purchase new equipment came up), and on at least a couple occasions, I walked away from some of their solutions that seemed a bit half-baked, but those occasions were few and far between in comparison to the number of successes they’ve afforded me.
Now, you do have a few points that I agree with you on.
As SDN technology matures with time and investment, resulting automation will lower OpEx, improve operations, and enable new network services and capabilities.
I see a mixed future for SDN. In the right environment, most importantly, when the right people care about the right things and really want to take things to the extreme, I think that SDN can offer a lot. It can definitely simplify deployment of end-to-end architectures (to my point above that it’s all about the architecture, not about the whiz-bang feature of the month), but only certain audiences call for that. There will be a certain subset of customers (of which I’m not ready to speculate the percentage of) that will always be better served by a ‘KISS’ style network architecture. They don’t need automation; they don’t need orchestration – what they need is something to provide basic, highly-reliable connectivity and communications.
Insist on engineers that are network professionals, not Cisco clones. A true professional is all about the protocols and technology and operational excellence, not about a specific vendor’s products.
I couldn’t agree more – true engineering is about more than a single vendor, in fact, I’m going to leave _every_ vendor out of my response to this – vendors aside, it’s the architectures and design, the underlying technologies that enable us to build the networks of today, and will enable the networks of the future. Understanding these fundamental concepts will apply to every network you build moving forward, just like many of the skills I learned back in the days of 10-Base-2 Ethernet still apply today.
Now is the time to move beyond the 1980s and start towards a modern network world. Every serious technology executive can contribute to the advancement of the state-of-the-art, and the first step is to not automatically give that next big order to Cisco.
I think I’ve sufficiently demonstrated my point above that we’re not in the 1980’s networking world, and as a manufacturer, if you want 60% market share instead of 3-4%, you’re going to have to create something, disrupt the industry, and change the world. Begging people to stop buying from the companies who have been innovative, created disruption and changed the world – well, I’m afraid that’s not going to get you very far.