Anticipating the future, assessing the present, judging the past - An industry visionary seeks gold among the chaos of change
Date: Thu, 11/25/2010 - 20:26
The trouble with "living in interesting times" is that we need some sort of road map to keep them interesting, not just confusing. So John McHugh – counted among Network World's "top 50 most powerful people in networking" – was an inspired choice for our NetEvents 2006 keynote.
Based on his success to date in predicting industry trends, we asked him to look forward four years to 2010, and predict what major changes would have taken place – which then current trends would fizzle out and which would progress. We also asked him to hazard his opinion on surprises in store
Well, 2010 is here. Things have become even more interesting, and it's time to put his predictions to the test. John McHugh is now Vice President and CMO, Brocade (one advance he did NOT predict in 2006!) and he is back to face the music.
What are the key trends in, and impacting, our industry now? How well did John forsee them in 2006? Above all: where are they taking us? what more is on the cards? and what can we expect for 2015?
We all need that roadmap, and John McHugh, President & Chief Marketing Officer, Brocade is the ideal person to draw it.
NetEvents EMEA Press Summit Istanbul 2010
Outstanding. Thank you for the kind welcome and applause. It is probably -- a lot of my PR people said it was a questionable decision to do this particular presentation five years ago and they have equal concerns about it today. But it is interesting that it's the one presentation that I've done in my 15, 20 years of presenting that sticks with me and more people -- this podium is just about as much of a challenge as one could have for presenting -- it's the one presentation I get asked about more and people point out what I said and the kind of remarks that were implied there.
So maybe the really questionable judgement is actually coming back here and answering to those and worst of all, doing it again. I'm going to wade back into the water. I will try to touch on several different aspects of the technology and social aspects of networking and put myself out on a limb with some very assertive and bold statements that I can be held accountable. But I am a business manager at heart and so the first thing you have to do is you have to answer to your results before you get to go in and start requesting people's indulgence for the next plan and then going forward.
The one thing I want to re-stress -- and thank you to NetEvents for bringing together such a graceful treatment of my last presentation -- the one thing I want to stress that I put in there is that this predicting the future is not like a cocktail party game. If you're in an industry where customers buy your equipment and expect to use it for five to 10 years, where development cycles take an additional three years or two to three years, you have to regularly understand where the industry is going to be in five to 10 years. If you're not good at being able to understand major trends, customer buying decisions and various other elements of the ecosystem you're selling into, you'll be a rounding error within five to 10 years. It really is a critical part of the basic ecology of what you have to do in these long lead, engineering-driven and innovative markets, such as networking,
So I'm going to go through each one of them. And there's a quote from Arthur C Clarke that I love very much which is that we tend to overestimate change in the short term and underestimate change in the long term. And I think that is -- coming up on 28 or 29 years in this industry I think the one thing that I've come to know and love that very strenuously. In other words, the hype cycle that we as an industry constantly churn upon and constantly throw these images in front of customers saying that here's tomorrow and it's today, you need to buy now, you need to go through all this churn. This is stuff that just if you're not doing it you're going to be out of competition. It really is a basic norm of our industry. And frankly I think it's specifically a defective norm. That for the most part -- and you'll see a lot of my predictions are contrarian predictions, not necessarily just getting on the boat and going in the same direction. My experience is that in general the large direction of the industry is toward status quo and toward customers buying based on solid business values and processes.
So let me try to stay on track here and not get too bad. So collaborative information. In 2006 I talked about a world where information was readily accessible in all sorts of places and the common places that you had gone for information, the big houses of stacks and stacks of hard copy and hard research data was no longer going to define the information landscape. The information landscape would be defined by collaborative, dynamic information churn pools that will constantly create new types of information and feed this information-hungry ecosystem. I think I got that one pretty well. To be honest, to pick the level and gravity of what's going on you’re your smartphones, the accessibility of information that's applicable to where you are right then frankly was a little bit off the chart. But I certainly believe we've gotten to that point.
Innovation versus commoditisation. Since 1989 I've been hearing that the networking industry is commoditising. Back in my days working at HP I remember a consultant advising HP's Executive Committee back then to get out of the networking business because networking was just connectivity, had no relevance to the general innovative process in the industry. In fact I believe in my career that was the worst piece of advice that anybody ever paid for. Because what we've seen in the last 20 years is that in fact networking has been the engine I would argue of innovation, has been the most critical non-commoditised technology space. Today -- even today, look at the dominant position networking has in the industry and how many people are investing and creating and driving new networking-based architectures and innovations.
Complexity of security. This is the argument about the necessary evil. What I said was that basically I believe most adoption trends would be driven by security or in fact the inability to secure them and that the challenges associated with any architecture were really for mainstream customers who are going to come back to how they embraced security or did not.
Okay, there's another -- you'll love this treatment of the English language. There's a common thing in terms of predicting and making forecasts in the difference between precision and accuracy. Okay. In forecasting in the process that we use we tend to be pretty accurate but we tend to not be very precise. In other words, you can make a forecast saying x is going to happen and what happens is there is this synergistic dance that happens between innovation and customer evolution. In other words customers get a bit of innovative technology, they don't just consume it the way you gave it to them but they actually change the way they do their job. They start seeing connections between other things they do, they stop doing some things and start doing other things. And so when you push this ball down the hill believing it's going to end up right where you pushed it at the bottom of the hill, in fact what happens is it does go down the hill, it just ends up in a very different place than what you expected it to do.
So what I said back then is I thought 10 gig was a prevalent technology that would be the centre of deployment, one of the most critical deployed technologies in this timeframe and I will freely admit the expectation was high definition video to the desktop, a lot of applications that were driving the need to push through the one gig barrier. I got the basic trend right, I got the destination wrong. As it turns out, bladed servers, virtualisation, the fact that most servers which back in this timeframe had about 15% utilisation on average, now we're pushing 50-70% utilisation when they're virtualised and have numerous applications running on them. Suddenly the IO performance of a server architecture is what the fundamental limiting performance requirement is. And hence 10 gig Ethernet in the data centre is the most important deployment issue that most customers are addressing today.
So like I said, not precise in the prediction but I'll argue still pretty accurate. Wage parity. This is probably one of the most edgy and I've got to tell you as I watched outsourcing happen -- and this really is about outsourcing -- as I've watched outsourcing happen I've been surprised at how people believe that if an economic value exists for outsourcing it will immediately become the norm. There's an assumption that if you can get something cheaper it inherently is better. And I've got to argue there's a lot of examples that that's in fact not the case in many, many situations. In the outsourcing example and the move towards China and India as being the hallowed ground of all engineering, all call centre, all service provisioning, I started applying the brakes actually as early as about 2001, 2002 and saying yes, the economics are great but two fundamental problems. Service level not necessarily comparable. You've got to factor in overall productivity into the equation, not just is it cheaper on a per transaction basis.
And the other is watch out for the economics. You can move an entire strategy to this more immature development area, find that you've got some problems with output and quality and productivity but also watch the economies get out of control. My prediction was that the fastest-growing wages in the world would be in China and in India today. And in fact that's pretty much exactly the case. In fact I just came from India yesterday and I had one of the customers tell me he's a big service provider there in India and provides various solutions for his customers and he told me a stunning story that I just -- I was so happy to get it before I came to this event. He said yes, we've got this call centre and service centre and when we engage with customers we create a very customised support arrangement with them. And I said how's your business right now. He goes oh we're dealing with some transitional stuff. He goes I'm outsourcing my call centre. I said you mean to somebody else here in India and he said no the Philippines. He said wages are too high in India, we can't afford to do it here.
This is the -- in physics nature abhors a vacuum. In economics the industry abhors a low cost region. It will flood in, over consume it and drive wages and costs up. And so literally there might be some first mover advantage but then you're suffering the deal of actually trying to mature the market at the same time you're trying to get economies out of it. Very touchy thing but certainly it really goes to how this is a global economy and everybody participates that way.
I made some comments about mobility. Mobility certainly is much more prevalent today, is much more part of our lives, our expectations of our smartphone, when we can't get something to work I think we're more surprised than when we can get it to work. The service level and the ease of doing it is not seamless by any stance but I think this trend is solid and will continue to go.
I talked a little bit about virtualisation but let me throw that up in this two-part step. Because there's really two things going on with this massive deployment of network and cloud accessible applications. There's really two processes at work and they need to be separated because they're fundamentally independent. One is what's happening with virtualisation in the data centre. This first bullet is really about how the data centres today are evolving. Fundamentally today's data centres by and large are incapable of dealing with what virtualisation requires them to do. And so customers who are loving virtualisation in a way unseen frankly in any trend I've been exposed to in my career are finding that once they deploy this technology that produces great value for them the complexity of managing it in their static rigidly defined network in their data centre doesn't work at all. In fact it increases complexity.
So another way to think about it is data centres that were designed to support 100 apps or less are having to support instances of 1,000 or 2,000 applications. And their staff, and their complexity of management of that just doesn't stay together.
But there's another piece of it. So virtualisation and being able to virtualise an application makes something mobile, makes it much more dynamic and able to be extrapolated to anywhere in the network but it doesn't enable the cloud. Right. I can virtualise a data centre and still not be able to take advantage of the cloud, still fundamental issues exist with being able to deploy and make cloud work. Both technical but also business controls and our good old friend security.
Alright. So let me talk about where we're going. Let me start putting my neck on the line to 2015. So first one. Cloud computing. Remember I said one of my strengths was shooting holes in the hype. Cloud computing is over-hyped right now. Cloud computing is going to get adopted and embraced much more slowly than people believe. And let me define the ideal state of cloud computing. The ideal state of cloud computing is I as a data centre IT operator have a bunch of applications and performance I want to run at some particular speed, I need some extra capacity next week because I know there's going to be very high demand. I want to be able to go out to multiple service providers and very rapidly be able to get a rock solid contract for service level agreements, feel comfortable with the predictability of their delivery, the protection of my information and my critical business, and that I can go ahead and choose from multiple suppliers.
That vision is probably 10 years or more away. What's going to happen in the next five years is what I call extended private clouds. In other words somebody who really wants to embrace the cloud, wants to have capacity on demand, disaster recovery, wants to have a virtualised location outside of their own, in most cases is going to have to work with one provider to develop a customised if you will public cloud environment that they uniquely own that has metal cages around it and assures the integrity of their business information.
We really don't have both the technical and the legal capabilities to fully engage in the public -- in the so-called hybrid cloud with a fully enabled cloud in the next five years.
Home PC sales. The home PC is a dinosaur. Remember how many of you have typewriters floating around your house. I think very closely in 2015 the classic desk-side, desktop home PC will be feeling a lot like your typewriter. I don't think we will necessarily see a rapid drop-off of traditional clamshell portables in this timeframe but we could very well as well, certainly to individual users. I think the number of those that will be purchased and deployed will be beginning to roll off.
The world of the smartphone, of the highly mobile interactive appliance is taking over. And the other thing that's happening in the home is that the role that the PC has been playing in a generalised fashion is being custom-built very efficiently into existing devices or so-called appliances that provide real value. So I know this one is -- a couple of people have taken a swing at this but I really believe in the next five years this will begin happening.
Cisco enters the storage market. I would love to say that Cisco's business strategy is very thoughtful and thinks about evolution and things like that. And in fact they've been masters of driving innovation and driving new customer models throughout the industry. There's no question about it. But you have to always look through the actual implementation of their strategy to understand what the true strategy is. Cisco has been since their inception a growth company. With their market share, with their margin structure, with the size of the overall addressable markets that they participate in they can no longer be a growth company without going into adjacencies. Voice, now into data centre and computing, storage is too big, too rich in margin to avoid their focus in the next five years. So I believe that they will continue to redefine the data centre by becoming frankly more of a systems company than a networking company.
Land markets transition. So like I said I've been fighting the networking as commoditising since 20 years ago and discounting it. I'm now going to change a 20-year held opinion. The traditional campus land area of the networking market I think will have a hard time resisting commoditisation in the next five years. I think the beginnings of a much more generic deployment and connectivity structure, the seeds have already been laid. I mentioned back in 2006 that Cisco would be under very new assault and significant assault from directions that they hadn't in the past. I think that world is happening and part of it will be -- one of the victims of it will be the highly differentiated campus will become much more genericised.
Smart conferencing. So this is in the gadget area. I believe in the next five years that your smartphone and various other small devices, recording devices, will actually become the tools that you take notes with. But you won't be involved in trying to quickly move your thumbs around a very tiny keyboard or a touch-screen keyboard. In fact the ability to take spoken word or video and translate that into discrete searchable and translatable elements will be -- will certainly be in the exotic prototype phase. But many companies will install these things in conference rooms and the idea of taking notes and having to be responsible for tracking what went on in the meeting real time will go away. You'll get back to your desk and you'll have a transcription of the meeting, maybe even with highlighted elements, which were action items or other pieces.
So in other words the ability to search, capture and analyse spoken and video will be much more common in this timeframe, increasing of course the amount of raw data that's flowing through all of these networks, more information, more accessibility.
Global companies. Right now because of financial constraints and some constraints in terms of logistics we still ultimately, even with the internet such as it is, we still largely are made up of a world of very local companies. I believe in fact that in the next five years the world will shift to the point where the ability for somebody anywhere in the world to set up a global company and do it with incredibly low entry costs and initial capital costs will change dramatically to the point that most small entities that form will begin thinking about their organisation as a global enterprise.
One of the things that I presented that they didn't highlight in the video was I talked about the lessons that we learned and we continue to learn and some people never learn. These actually have not changed at all because they are the foundation to this process of trying to figure out what trends are real versus which ones are not going to be successful. Technology does not drive adoption. Just because you can make a technology work, you've taken step one of about a three or four-step process.
The persistence of infrastructure. Brocade as a company is an example. Industry leader for the last 15 years in storage area networks and data centre networking, is now in transition as a company from that lineage long strong history and leadership in storage area networking to becoming a networking provider, networking leader. We're doing that and having to have two feet in both worlds. One is we recognise that our solutions and the products that we've been deploying over the last 15 years are going to be around for the next 15 years. In fact that market despite the fact that most people can see its ultimate demise some time in the future will still be a healthy growing industry for many, many years. To this point is people underestimate how valuable customers believe their existing infrastructure and capital is. Especially when it's working fine. So whenever we predict massive changes in customer buying behaviour we totally discount their existing infrastructure to our mistake, to our error.
Proprietary infrastructures limit deployment. It's very simple. Look for this in cloud computing especially, in virtualisation and cloud computing. Because the first solutions that are being brought to market are typically designed by several companies getting together or one company saying do it this way, I'm going to make this rack and it's going to have a server, it's going to have connectivity, it's going to have storage. That's a virtualisation environment, it's the best virtualisation environment in the world. It's the best virtualisation environment in the world but it's utterly useless. We can see these -- some of these have been shipping for over a year and if I had five friends I could count the number of units that have shipped on all of our fingers put together.
The reason - cloud computing is about unbounded interaction and choice and flexibility. It's not about buying a rack of equipment that defines what a virtualisation environment. You know what that's called? That's a mainframe. We rejected those things 20 years ago. And so some people are using virtualisation as an opportunity to define this world around this rigid structured environment system that allegedly is great for virtualisation. The world of virtualisation is multi-vendor open standard and customers get to make the choice, not the technology providers that are trying to force them in one direction.
Rapid advances in technology in one space unlock potential in other spaces. So in other words, I've talked about this accuracy versus precision. This is what causes the disruption. Once you create genuine value in innovation, like virtualisation right now, the ability to virtualise an application is being taken up by customers in record numbers, about 300% year over year. Even as server sales overall declined last year the number of applications running on those servers went up dramatically. And what that says is that customers are loving virtualisation for what it does in terms of using up their capacity of their existing server. Awesome. But the problem is it hit the wall when it got to the data centre architecture and so now the innovations happening in that space very discontinuous innovation, brand new things are coming out that are going to change the world but frankly virtualisation is going to be held at bay until those new technologies actually come to fruition.
And finally applications can only move in unison with security. We see this with cloud computing right now. Most customers I talk to frankly technically think they can do more with cloud computing than they can. But what they freely will admit is I don't like the cloud because I know my compliance officer, my security officer, my CFO will not allow me as the CIO to take and move information anywhere out into the world and not be able to control it, not be able to secure it and not be able to demonstrate compliance to security standards.
And so that's a fundamental issue. We need to realise as I talk about cloud computing's going to go slower than we expect. That's what the next five years -- we're going to come face to face with that issue and realise frankly how far we are from where we need to be.
So let me wrap up and talk about when I think about trends in adoption and what's changing the world, the kind of things that I think are driving and have continued to drive customer vision all along. Simplicity. Another way to think of this is reducing OpEx. Whether it was two years ago, five years ago, 10 years ago, customers always when they spend a dollar they want to reduce their overall costs. Simplicity will be characteristic in anything that's successful. Investment protection. I need to know what I'm buying today is going to be valuable in 10 years and I also want to know that what I bought five years ago is going to continue to maintain its value or improve in value in whatever you're selling me.
Non-stop. Ethernet has had a long rich tradition in best effort networking. Right. Ethernet's economies, openness, inter-operability is all driven by the fact that Ethernet frankly didn't worry too much about some of the little subtleties of robustness, line rate deterministic performance. Good enough. I've been in this industry for 30 years almost and good enough is really what we always design to and it's addressed 90% of applications. I think the standard has finally got here that Ethernet and the IP world need to start deploying in a way that feels like non-stop to the customer. I've got a lot of customers in healthcare and when we talk one-hour maintenance cycle once a month they go, your mum, your sister, your dad is in patient room number 23. What time of the day or night or weekend do you want us to shut off their monitors in their room for an hour to be able to upgrade the code on the system. It doesn't work. In the world where networks have become a fundamental part of the business model that doesn't work anymore.
And finally application optimisation. It's all about the apps. Apps and information. When we boil our industry down to the very basics what do we do for customers. We give them the ability to store, synthesise and retrieve information and we give them the ability to run applications. And so being able to execute those applications anywhere in the so-called cloud in a predictable, safe and reliable manner is what the business is all about.
So with that, let me put my stakes in the ground and let me do a little one-on-one with Manek.
Manek Dubash, Editorial Director, NetEventsTV
Thank you John. Take a seat. I'm just going to ask John two or three questions and then throw it open to the floor. So get your pencils sharpened.
I think there was quite a few themes running through there but I have to say one of the things that occurred to me was that you spoke whether specifically or implicitly a lot about Cisco. Cisco was in there a lot. About proprietary standards obviously, proprietary non-standards if you like, open standards, all those things. We talked a lot about open standards, we all like open standards, those of us who don't work for Cisco. So how does Cisco survive?
Actually the threat, the biggest threat that I believe exists to Cisco is one I don't think most people would predict. Because Cisco has the classic leader's dilemma. Which is Cisco has got so big in their home turf and their home market they simply can't grow there any more. And their choice and their strategy has been increasingly to be less of a networking company year by year. As I said I was just in India and I saw a Vice President of Cisco presenting their technology strategy and it was surreal actually, the presentation. Because the presentation was all about applications, the modern home of the future, the modern work environment and driving in your car and sending e-mails and all this other stuff. And it had nothing to do with networking at all. It was completely user -- use cases. And then at the end of it they stopped and said oh by the way pull this little curtain off, here's the new 4500E that we're introducing. And there's this big giant (inaudible) chassis. And it's like oh yeah, that's right, they still are a networking company.
The reason I say that that's so fundamental is because over the last 10 years and having been part of several of the major players in this space, the companies that have taken market share in networking have been the companies that have been absolutely and universally focused on networking. And those companies that thought networking was an accessory, was part of an integrated IT solution play, have lost market share or have been non-players in the networking space. Cisco is messing with the fundamental formula of what made leaders in the networking space. When Cisco was absolutely focused on networking and nobody else in this industry was, that's when they went from a small router company to the dominant player they were.
So now you've got IBM once again re-entering networking, you've got HP finally thinking that they actually want to have networking. But wrapping it up into this overall IT solutions space. It was tried in the 90s. IBM, HP, DEC, Compaq, they all did it, they all failed. So now Cisco's becoming a server company, maybe a storage company. Is this success? Is this what the next level of growth for Cisco is or is this what's going to fundamentally undermine their business? I think that's a very important philosophical question to look at.
I believe the companies that are going to succeed the most over the next five years are those that are focused on networking as experts in networking and recognising that networking is kind of the Switzerland of IT. You've got to work with everybody, you've got to approach customers by embracing all these pieces. The two pure plays. There's a company down the street that's trying to get from SPs into enterprise and Brocade.
So you're predicting the end of the Roman empire basically?
I'm not sure that predicting at this point the end of Cisco is a very wise or thoughtful -- Cisco is -- I just think in terms of if you look at market expansion and you look at their what seemed to be an unquestioned leadership position, I think they're messing with the formula.
Okay. Switching to mobile. I think this is an issue close to a lot of people's hearts, we've talked a lot about mobile data and we're going to be talking later on today about how mobile data is important for networks and the impact it has and so on and so forth. And we hear a lot about long-term evolution and high speed bandwidth and all this kind of stuff. But ultimately we still don't have 2G working properly. Why not? And where was that in your prediction?
I know this is an issue, you always think that there's a pretty big gap on mobility, and I tend to see it as less -- it's less [poignant]. I think there are some fundamental capacity issues, I think there are some fundamental compliance and inter-operability issues. I really believe that because of the customer demand ultimately the economic demand should get figured out as well, that it's going to continue to mature. I don't think there's any turning back because people have tasted the taste of this and they're not going to let it go, they're not going to back away from their demands. I think rising demand is going to pull the quality, the compliance and inter-operability issues along.
I think I agree. Mobility is what everybody wants and obviously mobility even within the home, just to pick up that point you made about home PCs. So what happens to gaming? This is a big market and growing for which a lot of PCs do that. So what will happen to that? Is that all going to go to consoles and all that stuff?
I think about the numbers. I have two kids who are in the prime centre of that particular market and I think of the number of them that are running PC games versus dedicated gaming system games and PC is a rounding error for them. And even at that, they don't want to necessarily sit in one place in the house. The next generation is a very mobile, active multi-feed kind of individuals. And so they're redefining the use model. Social media, the way people communicate. I believe all those factors are basically leaving the traditional bounded even today PC a bit in the dust.
Well I don't know. I'm not going to go into [some depth], my understanding is the last stats I saw actually gaming is growing in the 25-45 age range and actually it's changing quite a lot, it's not kids any more. So that implies what?
Well, what it implies is that we use the internet as more and more of our entertainment consumption. And when you look at the amount of information that's being processed and stored, some people I believe this year there's something like 1.5b devices connected to the internet. The expectation is in 2015 that's 15b devices connected to the internet. The amount of information that's being stored and moved around continues to go up by an order of magnitude every three or so years. I think that's just part and parcel. It's one of the elements and engines that's driving it. At the end of the day I think it's one of the forcing functions for the infrastructure build-out.
Okay. I've got a number of issues so I'm just going to raise one more then throw it open to the floor because we're going to run out of time otherwise. I think you just -- I think you by the back door said you think that voice recognition is finally going to happen?
Well it doesn't seem to be any closer now than it was 10 years ago. What do you think is going to happen in the next five years?
So the -- I think the things that are starting to come together is the processing, together with the connectivity. It's a classic networking thing is that what networking does it put processing power and storage power at the scene of the application. But I think the combination of high performance mobility, the density of processing, the sophistication of software and the capacity to move and store information is all aligning, the stars are aligning to let that application actually start clicking.
So I believe with some of the early start-ups that are starting to happen now we're finally going to get if you will push through that final level to where it's no longer a tool which creates an inaudible or inedible garble but in facts starts producing something that's coherent and predictable as a business function.
To your quality standard we might still be like mobility is today. It works most of the time, it works enough that we really want it to work but it's not perfect enough that we would trust it with our lives.
Well this is the same technology that does face recognition effectively, those kinds of algorithms. So that implies a huge invasion of privacy if nothing else.
I'm not sure how I make the leap in terms of invasion of privacy.
CCTV, you can pull people's faces from wherever.
Sure. I think the business and personal application, I think this is a self-selecting thing. Carrying a cell phone around in your pocket represents an invasion of privacy. You have a choice.
There's lots of other issues though. Questions?
I'm Bob Emerson, freelance writer. Also looking for work like many. You've said applications need to be in synch with security but smartphones are making their way rapidly into enterprises and they are wide open for security. I did a little bit of research in this area and there are solutions for smartphones but the iPhone, most of them do not apply to the iPhone and that is the one that is really entering the enterprise. So I wondered if you could comment on that.
This is an issue that's pretty well understood by customers and to be honest with you it's an issue that's pretty well understood by manufacturers as well. And the two or three organisations I've been associated with that ran enterprise applications and either permitted or didn't permit the iPhones forced the iPhones into certain states, the same way with Droids and other smartphones is that there are very specific standards that they have to pass. So are they flawless? No, but no technology at the end of the day is. But I actually believe that this is one where enterprises have reasonable risk assessments that they take and I know my iPhone is operating under a set of security parameters that my corporation enforces upon my use of it. So I think that's how most companies are doing it.
The typical things they require are some sort of encryption and remote wipe, the ability to erase it remotely.
From the floor
I just want to say first, thank you. You always know if someone's done a good presentation if you've written down a few lines you want to steal and there are a few I'd like to steal from you. And one of the things that I liked was your discussion about the pragmatic adoption of cloud. Analysts like myself, my peers in the room no doubt, we're all struggling about what the adoption rate's going to be and discussing with enterprises on this. But one of the statements that you made was expressing some level of concern about proprietary architectures and constellations of vendors. And I know what you're talking about but at the same time at the early stage of adoption of a particular technology you have to have certain constellations to bring the -- to move the market along. So how else would you suggest that one move to cloud environments, if not having certain constellations of vendors working together?
It's a perfect question and it's a perfect observation as well. And to be honest with you I think you're absolutely correct in your assessment that the starting point of any new technology is customise. I often somewhat derogatorily refer to them as science fair projects. I can make this work and I can demonstrate its capability but it's not scaleable. That's the phase we're in right now is that most of the solutions that are coming out that require customers to choose this thing from column A, this thing from column B and this thing from column C, it shows the concept, it has value in terms of being able to implement on a limited scale. It doesn't provide the mainstream economies and ultimate scaleability that the industry will adopt in a widespread manner. The things that are going to drive that are standards, inter-operability and multi-vendor support. Dealing with reality.
So I think to your point is we're just in the first phase where people are doing proof of concept type of implementations, trying to sell them as this is the best solution. And there are certain customers that that will meet their requirements perfectly, it will give them a pre-packaged solution. It's just that those things are going to look strange in two to three years when everything else is designed based on I get to make the choices as the customer, I expect widespread inter-operability.
Back to the cloud adoption and my pragmatism there. One of the things that I think's the best example that's limiting the cloud deployment and will constrain it a bit is that nobody's talking about the most difficult part of virtualising and extending a data centre. And I describe a data centre and I've described this to literally hundreds of customers and got affirmation time and time again. And it's what most customers who are way into virtualisation have discovered. Is that first of all messing around with somebody's data centres and their apps and their architecture for information is like doing heart surgery on their IT. So first of all, they don't want somebody coming into the operating room who's doing their first heart surgery or just got into heart surgery in the last year or six months. And secondly, the part of the heart surgery that nobody's really facing head on yet is storage. Moving an app from here to a data centre in Singapore is very straightforward. It's two gig, you can do it in milliseconds and you can do it without a service interruption that most people would be able to notice. Moving the associated storage, the terabit of database and other information that supports that app and makes it provide something of useful value, I can't move that in 10 minutes if I'm lucky through most internet streams.
So you can't virtualise it. You need some -- you've got basic issues with the raw bandwidth, with the latency going through there. It's really storage. I like to describe virtualisation as like getting this yacht. So you can hop on the yacht and sail anywhere you want around the world and have this wonderful experience. The problem is the yacht is tethered to a three-ton anchor that's stuck in the mud of the harbour. And that is how do you make storage zip around and appear to move in milliseconds anywhere in the world.
So as a surgeon that's been doing surgery for 15 years Brocade is a company that's had an expertise in this area, specifically in what we call remote extension. How do you make storage appear to be various places at the same time without it actually physically being there. This is technology that's going to be critical to how truly you go from a virtualised data centre, easy, to a fully implemented hybrid cloud, difficult.
And so a lot of that stuff is going to start I think developing over the next year or so.
Sales pitch over. So --
Oh that was [cold]. Come on.
So the [glowing] company then that you talked about, that's all very well but by the same token they still have to shift stuff around unless they just -- unless we're going to a totally services economy we still have to have stuff don't we. So how's that going to work?
This is all about the virtual supply chain as well and the ability -- I have some friends who are open market entrepreneurs and literally they sit down at their desk in their home office and they are working with suppliers in Asia, they are working with logistics firms in Europe, and they're looking at how to build a global infrastructure, being able to deliver. So it never comes to their location. Inherently it's being manufactured halfway around the world. Just because they happen to sell it in a certain area of the US region, the fact of the matter is it's already come halfway around the world to get to them. So the ability for them to deal with the logistics that are going in a different direction are very straightforward.
Steve Broadhead, Broadband Testing. It's customary for me to ask a question on the first session. So I've got loads of networking questions John but they're all pretty dull so we'll do those separately. Like if we move to 15b connected internet devices does that mean IPB6 becomes finally global? And if so what does that mean in the real world of IPB6 ready devices? But we can do that separately.
But just to continue from what (inaudible) was saying and basically what you've been saying. We have a world at the moment, if it's moving into the cloud etc. and networking just becomes the plumbing which it really is, are we going to continue with the separation of physical -- not just physical and virtual but applications and application-building is very separate still to the world of storage networking etc. What you're talking about here, you're painting a picture whereby I don't see how that separation can continue. So in theory does someone like Google become the next Cisco? In other words, can we continue to have separate application companies and separate networking/storage companies?
I thought actually Google was becoming the next Microsoft. And I don't mean that entirely flip. I really believe that -- I disagree with your assertion that the -- well actually I think that the assertion you made is one of the engines of this transformation. In fact applications shouldn't have to know and shouldn't be engaged at all in the infrastructure that they run upon. I think the model of the industry leader, who I'll not mention again, has been defective. Who I buy my unified communications solution from should have nothing to do with who I buy the infrastructure it runs on. I think that's a totally defective position. And I think increasingly this is my point about these customised one-off solutions for cloud or anything else, is they don't scale. It ultimately does not appeal the mass market. The solutions that ultimately dominate will be open multi-vendor and that the infrastructure customers deploy should be able to run the applications that they choose independently to develop and deploy.
Let me highlight what's going on here. The advent of virtualisation. One of the things that -- just to help you understand how virtualisation, why it's so compelling to customers. Historically if you had a situation where you expected 100 customers at a time to all be trying to hit on you for information, for ordering and various other things, you develop one program that can handle the complexity with all of this multi-tasking and you have this long multi-year development cycle to create this massive engine that could handle that and keep data coherent and deal with other things. In a world of virtualisation, instead of that you design an application which can support one person talking to it, a single instance application. If you talk to any software developer the difference between designing an application that can handle multiples of people versus one is an order of magnitude. It's huge. It's very straightforward to develop an application around this single instance thing. Virtualisation lets you do all your app development basically assuming you're only going to have one user. And then every time a new user comes in you spin up a new virtual instance and run the application in there. So you scale virtual instances not create these multi-user environments. So the application is inherent, it's fundamentally inherent. The application cannot have any visibility to what the network is, even what the virtualisation engine is, what the storage is, what the device it's running on is. If it's truly virtualisable it has no clue what's out there other than it has resources and the performance and the customer interaction that it needs.
One more question?
From the floor
Are you sure what you just said is true for real time applications like voice and video, this network or infrastructure agnostic position?
Absolutely. If you need to know what infrastructure running on then you are -- then the infrastructure is not the correct infrastructure. And by definition --
From the floor
Do you mean from a specific vendor or do you mean anything about the infrastructure and how it's connected and operating?
It's all around quality of service and that it's achieving the service level requirements that were specced for that application. So at the end of the day a single vendor does not have any magic capabilities to be able to implement a solution in our industry. I think all of well-architected solutions can be -- with the right feature set and the right performance characteristics can support the applications or none of them can. And so the period of time where a proprietary and single vendor solution is necessary I think is transitory. It's a period of time before standards emerge, before real economy emerges and certainly before customer choice emerges. And those have always been brief periods of time, from the advent of Voice over IP to the deployment of video today.
From the floor
You spoke about extra private cloud, but isn't it a new private network instead of private cloud?
So a private cloud today for the most part is referred to what happens when you truly enable virtualisation in your own local data centre. In theory, in fairness a private cloud can extend, it's just it implies that you own all of the resources or you directly control all of the resources required to implement that cloud. So you can create a distributed private cloud or what I refer to as an extended private cloud but it still means at the end of the day you either have lockdown service level agreements of all those assets but those assets don't float in and out to go to other people, they pretty much are owned by you, they're managed by you or by somebody who is your direct contractor. That's the difference. In a public cloud you're sharing assets with other people and you're operating in a domain where you have no real control over for instance where your information is stored, where your applications are processed. That's a big issue. There's many applications and many companies that are already in a world where their information has to be stored in their country or they're in violation of their government's privacy controls. So the public cloud in those environments is -- unconstrained public cloud does not work for there. So they need to start locking it down. As soon as you start locking it down you start driving costs up and your flexibility, your real business flexibility goes down.
Hi, it's Peter Hall from Ovum. Can you make a comment about what we as analysts tend to refer to as application-centric networking? To what extent is the network of the future going to understand the applications that are running over it down to individual applications to be able to dynamically adjust itself to the requirements of those applications?
This is one of my favourite topics and I don't want to speak ill of the dead but I think the ultimate company that pursued application-centric networks was Nortel. Nortel did an amazing thing, took the largest networking company in the world and put them out of business. And they did that with application-centric networking. And you can tell me all you want that being the best UC networking company in the world was the perfect thing to have during the deployment of Voice over IP. What better period in networking history could you have been out there pitching a UC Voice over IP-centric network than between 1998 and 2008. Voice over IP was the shiny object in our industry, that was the application everybody was talking about, converting all their networks to being voice capable. It was the biggest buying process that happened during that period of time. Nortel was unquestioningly focused as a leader in UC networking and UC applications. And they failed. They completely failed. Virtually every year they at most held market shares but in most cases unprofitably lost market share.
So the message to people who believe that application-specific networks are really a brilliant idea is customers do not accept that notion. Customers believe that they have one network. They may choose to physically separate or logically separate those networks, they may invest in things that are very specific for applications on their network. But at the end of the day they want to engage with a networking company that understands all applications and recognises that ideally they have one network and ideally all their applications run on it and run exceedingly well on it. So I'm not a big fan of application specific networks.
I didn't say specific at all and I don't mean specific.
You said application aware.
Application-centric is the word I used.
Application-aware would be an equally good word. So what I'm talking about really is customers having a single network but that network saying hey (inaudible) that means we'd better start adapting.
So if you change the term to application-aware, I have no problem with that. Application-centric though I translate into UC networking for example. It's centric, it has a specific application that it thinks it wants to be great at. So aware's better because that means that I'm not -- my last point was optimise applications. The network has to know the applications that are running over it and it has to be great at all of them. And so I completely -- when you say application-aware, absolutely. If I can't handle one of the apps that my customer wants to run over my network I don't get the job. The network fails. And if you miss just one major application you're out of business.
But to the specific point of networks that can dynamically adjust themselves, can in real time give you extra bandwidth, in real time can adjust the parameters of the network so that if you're doing video conferencing it's optimised in a particular way, if you're doing -- if you're increasing your SAP traffic then it's adjusting itself in another way. Will we get to this very dynamic situation where the network knows the applications that are running over it down to the level of individual applications and is dynamically configuring itself to adjust to --?
Absolutely. If you look at what has -- so I talked about innovation versus commoditisation in the networking space and while I believe that the campus edge might be moving toward commoditisation, most of the network is not moving toward commoditisation. In the next five years it will still be a very innovation and new technology-driven. And it's all about the very point you're making. That is exactly what brings customers back to new buy-cycles, to buying networking not from the companies who wrap merchant silicon in boxes and sell it cheaply, but they instead always gravitate toward those that are engineering and innovation driven. Because those folks -- let me give you an example of -- I said the data centres of today were inherently problematic if you're deploying lots of virtualisation. The answer to that is designing networks that are exceedingly good at enabling virtualisation. So moving a VM and operating with scaleable VMs, with having application instances show up on numerous machines and then go down and different applications move around, you need a very special set of services and capabilities within that network to make that so a customer is not dealing with the complexity of thousands of apps running but instead they're just managing a service domain. That's a fundamentally new set of features that don't exist in most network devices today. If you look at Brocade's virtual cluster switching solution, it is a great example of a technology, runs on traditional Ethernet but also provides this overlay of capabilities which are very unique and specific to making that run better than anything else in the world.
So I completely agree with your assertion as it applies in that area.
Okay. John, I'm afraid we've run out of time. Compression hasn't worked. So we're going to have to stop there but if you're around for a little bit I'm sure people will want to (inaudible).
I'm here through Friday night so as you can tell I'm happy to wax on any topic you may have and I appreciate the opportunity. Thank you.
Thank you very much.