The Coming Network Revolution

The Three Horsemen of the Coming Network Revolution

I came back Friday AM fresh from the FIRE conference Meeting of the Minds planning session in Seattle with a head filled with ideas about where the IT industry is headed and what it will mean for the networking industry. Thanks to an animated and robust dinner discussion, which included Mark Anderson, Ty Carlson and Michael Pfeffer, I came to the conclusion that three forces will be combining in the near future to drive a revolution in the network industry.

Before I get to the three forces, let me digress a bit and talk about the simmering preconditions for takeoff. For Infrastructure 2.0 blog and Archimedius readers, these themes aren’t new; yet they need to be rolled up to put the “Horsemen” in proper context.

Preconditions for the Revolution

The Peak IT meme describes a collision between the ongoing expansion of IT infrastructure and rising per-unit management costs, from servers (see slide 4) to IP addresses. Peak IT suggests that rising costs risk crowding out necessary, ongoing investments in productivity. The result is a dynamic of continued falling productivity mixed with continued rising operations expenses, raising the specter of stagflation at the core of enterprise innovation.

The weak global economy could exacerbate the problem by shrinking investments in productivity even further for companies under new cost-cutting pressures. Yet, a slow economy may have little impact on the proliferation of devices joining IT infrastructure.

The Three Horsemen: Virtualization, Netbooks and Cloud Computing

Rising per unit server management costs combined with increasing energy demands are no doubt key drivers behind the adoption of virtualization. Companies like VMware (VMW) and Microsoft (MSFT) have capitalized on the growing expenses of server management and have introduced substantial innovations designed to reduce such costs.

Virtualization technology, in effect, creates an extra abstract arbitration layer between computer hardware and software, allowing a single hypervisor (computer) to run multiple virtual images of servers (as well as PCs), automating the setup and tear down of systems and introducing the ability to run multiple applications and even operating systems on a single computer.

Because hypervisors have automated what used to be manual labor, they make systems more flexible, dynamic and mobile. A hypervisor becomes a network appliance of sorts, because the network terminates inside the hypervisor. That evolution introduces higher velocities of change, including the potential to move server images from one hypervisor to another.

Microsoft’s Ty Carlson told me last week about the sizzling potential of the netbook computer. As Cisco (CSCO) predicts 14 billion IP addresses by 2010, Ty was quick to point out the likely widespread adoption of $300 netbook computers enabled by Intel’s (INTC) new Atom processor. Unlike traditional personal computers, netbooks are very inexpensive and rely more heavily on the network for core functionality. They are game changers from the standpoint of both volume sales and potential demands on wireless and network infrastructure.

As virtualization decouples applications from computer hardware; cloud computing leverages virtualization to go a step further, decoupling IT services from hardware. It introduces new capabilities for cost reduction and efficiencies and makes the network more strategic to the IT industry. Many definitions of cloud are swirling in the air, some overly optimistic and yet other definitions are already in production.

The Opportunity

Out there in today’s world is the next Henry Ford or Bill Gates, with plans of how to leverage cloud computing to transform the delivery of computing resources. At this point, he or she is probably already responsible for delivering some hosting and storage capabilities and will use their existing cloud to leapfrog competitors. They are already dealing with the myriad technical limitations of existing infrastructure and are planning ahead.

There are some significant barriers to the cloud delivery of enterprise applications. Similar to the issues that faced enterprise apps delivered over the WAN (and the emergence of application front ends), most endpoint applications weren’t designed to interact with servers in the clouds; they were designed to interact with hardware existing close by (on the same motherboard) as where they reside.

Changing the dynamics of these processes will require a great deal of changes in how applications, servers and storage solutions interact. Yet that may be easier for Microsoft than Google (GOOG) in their attempt to create and monetize enterprise cloud apps from the ground up.

There are also security issues that have confined virtualization into hypervisor VLANs (virtualization-lite). Those issues will need to be addressed before the full power of virtualization and cloud are unleashed.

That is part of the reason why Cisco, F5 Networks (FFIV) and Microsoft are well positioned to translate their expertise and installed base into new capabilities aimed at unleashing the power of cloud, as a part of the new network vision. They have strategic expertise and intellectual property that should allow them to monetize the map to cloud… versus making the leap with freeware.

Of course, both Google and Amazon (AMZN) have their own advantages, including core business strengths that don’t make them dependent on cloud revenues for innovation. They don’t have to monetize cloud in the early stages. Yet both will need to leapfrog decades of technical experience in order to compete with Microsoft.

While some detractors may criticize cloud computing as hype because there is such a broad range of potentials (from mature to emerging to practically impossible), few would argue cloud computing’s potential to transform the shape of the data center industry. Cisco’s data center blog has some of the boldest and brightest commentary on cloud computing, virtualization and the data center.

These three horsemen (virtualization, netbooks and cloud) promise unprecedented new demands on the network because they introduce new levels of traffic, reliance, complexity and dynamism. They will force a necessary evolution which some are calling Infrastructure 2.0 or dynamic infrastructure.

Today’s mostly static network won’t be able to keep up without massive increases in manual labor and risk (costs could escalate even faster and cannibalize the gains from cloud and virtualization initiatives); hence the case for greater network automation and systems capable of keeping up with the net effects of the three horsemen.

A recent panel (sponsored by Cisco and Infoblox [my employer]) is now available via YouTube. Another panel, which will include Cisco, Infoblox, F5 Networks and VMware, will be held in San Diego on May 20^th at the Future in Review conference.

I’ll also be moderating a cloud panel on May 18 at Interop. If you’re attending either event feel free to stop by and say hello. Or you can simply join the conversation here.

Disclosure: long Cisco

Comments

[Josh B Thom...:]

This article is a day in the conversations of upper management at Microsoft. They've got a lot of thinking to do, and that ball keeps on moving.

Feb 09 06:39 PM

[Richard Glenn:]

Mr Ness,
If you had to pick one company to benefit from this coming trend, which one would it be? Cisco? Who has the most leverage if they get it right? Thanks.

Feb 09 07:20 PM

[GNESS:]

Richard:

That would be tough. I think Cisco may be in a more commanding position in the network at this point in time versus Microsoft in the cloud. Yet potential developments in the next 12-18 months pose multiple wild cards. I do like Cisco's recent moves with VMware, which suggest that they understand the network opportunity better than perhaps their peers. Microsoft has an uincredible footprint in the market and massive software expertise, not to mention server clout. For me... its a toss up or a question of who could monetize better...

Thanks
Greg

Feb 09 07:36 PM

[drone:]

Why not buy both of them?

Feb 10 05:06 PM

[Teutonic Kn...:]

Greg,

Very good article. You sounded like a wonderful tech evangelist. I have one question for you. (I don't ask question that I think I have the answer.)

Is the newly emerging Semantic Web (through the much-touted Ontology Web Language of late) related in any way with the Dynamic Network Infrastructure (Infrastructure 2.0)? I suspect there would be some relationship, since the Semantic Web is supposed to be able to "reason", and the Dynamic Network (mainly web-based?) is supposedly intelligent.

Would you care to relate the two entities?

Thanks in advance.
Teutonic

Feb 10 11:15 PM

[GNESS:]

Teutonic:

I think they are two distinct ideas with very similar perspectives. One is about the emergence of network intelligence based on enhanced application, endpoint and network feedback loops (e.g. this application and VM has moved to a new location and the security policy needs to remain in place, etc); the other seems more about a more sophisticated form of pattern recognition and analysis for web content. I'm not a Semantic Web expert but that is my best guess. Thanks for the comment and question.

Greg

Feb 11 02:26 PM