I purchased my first smartphone back in 2005. It was a Palm Treo, and I remember clearly how I marveled at how it combined a PDA and digital camera with a fully-functional cellphone. The Treo could even load and run software applications. I drooled over its multi-functional capabilities, as I felt my smartphone made me far more productive than my colleagues who were still using their dedicated single-function cellphones.
As we all know smartphones have become far more powerful recently, by adding new processing and software capabilities that allow them to integrate email, instant messaging, web surfing, GPS, and a
host of other interesting capabilities. The cameras included in the newest iPhone and Android devices are so good that many consider a dedicated standalone camera to be completely unnecessary. Smartphones today address a variety of personal productivity needs that only a few years ago required a number of different separate devices.
What is happening with smartphones is also happening in a number of other product categories. Investing in many types of single-function dedicated devices is often considered a waste. Examples of devices that we now consider obsolete include dedicated single-function printers, copiers, typewriters, fax machines, MP3 players, DVD players, and game machines.
Furthermore, technology consolidation trends are not limited to personal appliances and gadgets--we are also seeing consolidation in enterprise IT. Virtualization technologies now allow data centers to remove dedicated single-application servers and host multiple applications and OS instances on the same platform. Indeed, the key initiative behind cloud computing and cloud
storage is to leverage common physical infrastructure for multiple applications and end-customers.
But consolidation has yet to arrive in the networking area. For the past several decades, networking has been the domain of mostly dedicated single-function appliances powered by proprietary software. And when you consider the relative simplicity of the functionality provided by these dedicated appliances--for example layer-2/3 switching and routing--one wonders why it can't be integrated into a more powerful multi-functional device,
especially in less-demanding branch office remote environments. Proprietary hardware and software isn't required to look up ARP tables and forward packets at the WAN speeds that connect branch offices--100Mbps or less. While such technology might have been considered state-of-the-art back in the 1990's, today it is consider trivial and simple technology to implement.
Rather, today's state-of-the-art technology is layer-7 application-level networking. We're talking about more advanced network devices that are aware of applications, can report on them by inspecting packet payloads to identify distinguishing signatures, and can perform operations within the network that address the behavioral
inefficiencies of each application in order to improve end-user performance.
These advanced layer-7 networking devices have considerably more processing and storage resources compared to the older-generation dedicated single-function devices. As a result, by leveraging virtualization technologies they can also host services such as firewalls, secure web gateway, domain controller, DNS/DHCP, and other application services. Of course, the Riverbed Services Platform (RSP) is one example of what I am talking about--RSP has been used by Riverbed
customers to host these and other applications on a shared Steelhead appliance platform. Interestingly, available RSP packages include Vyatta software that allows the Steelhead to provide functionally-equivalent routing and firewall capabilities used in dedicated router appliances.
So with all of these processing and software resources available, it seems there is no reason why layer-7 networking devices can't absorb some of the older simplistic functionality now being served by dedicated single-function layer 2/3 networking devices. For example, it does not take much CPU and memory resources to run OSPF and BGP--layer-7 networking devices that have already been deployed in thousands of enterprise networks are handling far more complicated and demanding responsibilities.
It's clear that even today, layer-2 and 3 switches and routers are already considered commoditized technology. Given past historic trends where we've seen where commodity devices become absorbed into more sophisticated products, will we also see layer-2/3 devices eventually be absorbed into more powerful and intelligent layer-7 products such as those available from Riverbed? Admittedly, such a transition--if it occurred--would take time. What do you think--can or will such a transition take place in the next five years? Why or why not?