If you want to run into Pradeep Sindhu, co-founder of Juniper Networks, your best bet is one of Palo Alto’s many cafes. Chances are, if you spot Sindhu, chairman of a $23 billion (in market capitalization) company, he is likely pondering about the future of the infrastructure of the Internet and the next-generation of networks.
A few days back, I caught up with him, hoping to understand how the Internet is going to evolve especially as we have entered the age of anywhere computing. “The iPhone is doing to the mobile world, what the browser did to the wireline world,” Sindhu told me. The iPhone and Android-based smartphones are changing everything, including how his company thinks about network infrastructure and how it will refine the network architecture in the future.
Why Should You Care?
Many who are developing apps and services for mobile devices don’t pay much attention to the innards of the networks themselves, barring moments when our network behaves like me running up a hill. We should be paying attention to all the underlying networking technologies, mostly because it helps us think about what these front-end services can do.
Sindhu explains that today the “networks” are the enabling technology, instead of super fast, energy hogging microprocessors. As the performance of the network increases, so does the performance of everything connected to that network, and by extension, the apps built on those device platform.
The Apps Make The Platform
“When the platform becomes general enough, it doesn’t matter as much as the apps built on top of that platform.” Sindhu pointed to IBM (s IBM) 360, the PC and the web-browser as platforms that became gigantic because of apps built on those platforms. The PC spawned app companies like Microsoft (s MSFT), Intuit (s INTU) and Adobe (s ADBE). The rise of the browser as a platform helped spawn apps such as Google (s GOOG), Amazon (s AMZN), eBay (s EBAY), Yahoo (s YHOO) and Facebook, all worth many billions of dollars.
The iOS and Android-based Internet connected mobile devices are the new platform, Sindhu argues, except, they are with us all the time, allowing us to use apps built on top of these platforms, anytime, anywhere and as much as we want. This is a brand new type of usage behavior that is going to have a profound impact on the demands put on network infrastructure, and how the traffic flows across the network.
Sindhu points out that the establishment of platforms is becoming faster and faster. Apps, he argues, form the positive feedback loop for a platform – the more applications, the more demand for the platform. What has changed is the network (aka the Internet) which acts as amplification for the feedback loop.
Cloud + Mobility = New Information Infrastructure
Sindhu believes (and many agree) that the rise of mobility, which I like to describe as “anywhere computing,” is going to change the whole notion of information infrastructure. Most of us want “to consume information and information services anytime, anywhere, with no limitations, and preferably in the same way across all devices,” he points out.
In order for this to happen, you need “cloud” based computing. “You need an architecture where storage, especially long-term, persistent storage, needs to be absolutely centralized, logically centralized, in large-scale data centers,” Sindhu argues. And it goes for heavy computing, which needs to be in the cloud as well. Perhaps that is why in this new world of eyeball-oriented computing, mobility and cloud, should be viewed as two sides of the same coin.
For this new environment, you need devices that have just enough local storage but can tap into the cloud for everything else and are graphics intensive. The cloud clients of tomorrow are not thin clients or fat clients, Sindhu says. Instead they are all “fit clients.” And you need fit clients because for the near foreseeable future, the network performance is going to remain quite variable. “You want the device to operate well, even if the network has variable performance,” Sindhu argued.
The Network Under Pressure
It also means that the fundamental shift to anywhere computing will increase the economic pressures on the network. He argues that it is time to stop thinking in terms of silo networks–a network for voice, a network for radio, a network for broadcast television, an enterprise network–and instead think it terms of a single network.
“The silo networks actually destroy value because the value for a network is maximized when it’s a fully connected, any-to-any network. In other words, anyone can reach anyone else,” says Sindhu. It’s like Metcalfe’s Law for the cloud computing and mobile age. Sindhu argues that it is time to stop thinking in terms of boxes, wires and pipes. Instead, we need to start treating the network as a living, breathing platform and build infrastructure for it.
When I asked Sindhu to predict the traffic patterns of the Internet, he declined to be specific because he thinks that we tend to underestimate how creative people find ways to leverage the network. Instead he offered three observations:
- The bandwidth requirements globally are only going to accelerate.
- The traffic is going to get a lot more stochastic in nature.
- The traffic is going to become more dynamic.
Sindhu argues that we shouldn’t distinguish between a wired or a wireless network, for in the future the network traffic is going to be more unpredictable, with demand coming from any client device, from any app at anytime.
where are the billion dollar iphone, android app specific companies.
Not sure if you noticed but Pandora and Netflix are two examples of companies whose future changed because of the iPhone/Android platforms. I think you are going to see many more in a few years.
am talking about revenues, am not sure Pandora is making billion dollars in revenues. I see that netflix is making 2 billion dollars, just wondering how much of netflix viewers time is in ipad/iphone compared to PCs,PS3,Xbox,Nintendo etc
Sure not now but the ecosystem is young and it takes a few years to get to a billion dollars.
I think there are several examples: FaceBook, Twitter, FourSquares, OpenTable etc. Many of these companies may not have reached the popularity or use without an iPhone or Androd Phone.
Geez, what?
1. More bandwidth will get used and
2. It will be kind of random
3. Sorta
Is this what your shareholders pay you for?
Stephen
I am not sure you are following the recent changes in the Juniper product line they are making in order to adapt to the new network dynamics. From their new chip architecture to new software layer, Juniper is obviously doing a lot of work to adapt itself.
That is what is his shareholders pay for.
I have always been a little bit dubious about the need of anything to be “absolutely centralised”. Centralisation makes sense in periods of technological immaturity and resource scarcity as it is a more efficient solution allowing people to get more with less. However, this efficiency comes at the cost of flexibility (big, centralised systems cannot be quickly changed) and, consequently, of fragility in the face of ‘black swan’ events (the events cannot be predicted and the system cannot be changed quickly enough to compensate; moreover, when it fails it takes everything out with it).
Sure, network technologies are progressing at a rapid enough pace to allow for massive amounts of data to be transmitted between centralised datacentres and ‘client’ devices anywhere in the world. But battery, processing, and storage technologies are progressing just as quickly. Indeed, the only thing that remains stagnant is the human capacity to process information. How many terabytes of storage would our phone need to fit before it is able to store all of the information that we regularly access, for instance?
To sum up, cloud services and centralised datacentres will certainly play an important role in the information solution systems of tomorrow. What I have my doubts about, however, is that they will play the crucial role. Why have ‘dumb’ terminals capable only of accessing cloud services, when one will be able to have all of one’s data and processing capacity mirrored and extended across a ‘personal cloud’ composed of the set of one’s own devices, including, but not limited to, the cloud-based storage and processing capacity that one would rent.
Think Dropbox in other words – the data is both hosted on their servers and mirrored across each of your own capable machines. In fact, it would not be stupid to adopt something like ‘user proximity’ as a general design principle – all data should first be processed and held as ‘close’ to the user as possible, propagating across his/her other devices as he/she gets ‘closer’ to them. It should not be a bunch of users orbiting some datacentres; it should be devices and data orbiting users. The user should be at the centre.
Excellent comment. It has more content than Juniper’s CEO.
I think LTE Advanced (not the “4G” or LTE of today) changes everything and results in irrevocable disruption of the telco/wireless market. I was lucky enough to work through the transitions of network technologies from dial-up to T1 to T3 & Ethernet and beyond. I think we’ll see a repeat of the rapid growth of greenfield opportunities that ubiquity of sufficient bandwidth brings. This will be the death of voice as a product (vs. a service), the computer companies win out over the telcos. Now, what happens to TV? And others?
N8nNC while I agree that Andrei’s comment is excellent, I am surprised that your comment about LTE_Advanced. I think many of those changes are happening right now.
As to your point about the Juniper CEO, I think the point of the piece is that the seemingly little toys like iPhone are actually causing people to reconsider the whole network architecture and it is making some if not all executives reshape their businesses for the future and Juniper is one of them.
Thanks again for participating?
Andrei
Thanks for the excellent comment. I am doing a follow up series of posts, including one later today which touches upon some on the issues you bring up. Clearly, it is not possible to fit everything in one piece, but hopefully this new series of posts is going to help.
SUN (remember?) was spot-on some 15 years ago: the network is the computer.
To bad they never were able to capitalize on it. Timing is everything!
I am always impressed at the amount of platitudes these billion-dollars execs can produce. Is there something in this article that wasn’t obvious to anybody who owns a phone?
Om, can u do an article on companies like F5 networks and how they relate to the subject.