What do you expect to accomplish as Google´s chief Internet evangelist?
Cerf: I have several likely responsibilities. One is to serve as a public face for Google, assisting its founders and CEO, Larry Page, Sergey Brin and Eric Schmidt in outreach and interaction with the public and the media. This is a role I have also played in some respects for MCI. I expect to visit most if not all of the Google engineering sites, partly to maintain personal communication among these sites, in addition to all the online linkages and also to serve as a kind of intellectual bumblebee cross pollinating technical ideas and solutions.
Has the Internet become far more than you could have imagined and what do you look forward to in the next-generation Internet?
Cerf: It has had a far larger impact on the public than I had ever imagined in the early stages of its development in the 1970s. It has ignited new ideas from literally millions of sources, with new ideas and applications coming at a massive pace. I am looking forward to increasingly fast access to the Internet, increased high speed mobile operation, more integration with entertainment services, coalescing of voice over IP, increasing amounts of online commerce and business document exchange. It is also reasonable to anticipate improvements in search algorithms as XML tagging becomes increasingly common.
Do you envision a larger Internet based grid-system?
Cerf: GRID computing is sometimes referred to as Web Services – which is a rather confusing reference – or by its commercial versions such as WebSphere and .NET from IBM and Microsoft, respectively. The basic idea is to make a collection of diverse physical resources, such as computing, memory and networking, look fairly uniform and virtual. One wants to pool a large quantity of these resources in the network and then assign them, through a virtual interface, to specific tasks. As the computing tasks are completed, the resources are returned to the pool for use in other tasks.
This notion is quite appealing because it implicitly allows for efficient utilization of the physical facilities by dynamic assignment. It also means that business continuity can be more easily achieved since business processes, built atop these virtual resources can be serviced with any physical resources. One can effectively move the operation from one physical set of equipment to another set without much preparation because the virtual resources are dynamically bound to their physical counterparts.
In practice, this idea has a number of challenges. For one thing, physical separation of the actual computing equipment will expose speed-of-light delays in communication among the computers or capacity of the interconnecting networks. These limitations may be reflected in poorer than desired performance. Clusters of collocated computers may prove more practical for GRID concepts and the utility may still be quite high. The resilience one can achieve by dynamically re-assigning different clusters to tasks may be very beneficial.
Implementations of GRIDs needs to be standardized as much as possible so that there is an organic ability to incorporate computing, storage and communication resources from a wide range of sources into a single system. The development of protocols such as SOAP, XML and registration systems to allow suppliers to “advertise” network-accessible GRID services may have an important impact on the way in which inter-corporate and intra-corporate computing is done in the future.
Back to Google: What do you think of the exploitation of geographically indexed information and its development?
Cerf: As wireless access to Internet increases and as mobile devices become increasingly useful, I expect that material that has been indexed by location will be very valuable. Users may find one another this way. Navigation will obviously be enhanced. Moreover, localizing searches to information about specific locales will almost surely become a common feature for finding all kinds of product and service providers that are within some distance of the user.
One can also use geographical indexing to present demographic, economic and other sociological information based on location. This can be very valuable in planning political campaigns, in studying the demographics of contagious diseases in epidemics, in understanding economic development, and in analyzing the health and environmental condition of cities, counties, countries and regions. It should be obvious that information about weather is geographically significant and may have major value in agriculture, transportation, city planning and so on. Google has found that the use of application programming interfaces (API) in its Google Maps service has stimulated innovative uses of the underlying maps by entrepreneurs who are adding value to the basic Google Maps capabilities.
With regard to your ongoing work with NASA and others on interplanetary communications, could you outline the development of an interplanetary backbone by the end of the decade?
Cerf: This is a very long term project. The first seven years have been spent developing and evaluating a series of protocols suitable for operation in high delay, highly disruptive environments. The various parts of the layered set of protocols are being deployed on current space missions. For example, the standard Link protocol of the Consultative Committee for Space Data Systems is in use on several hundred missions. It is hoped that testing of these protocols terrestrially and in near space missions will lead to more full deployment on missions to the outer planets that are scheduled in the second decade of the 21st Century. The team at the Jet Propulsion Laboratory hopes to be able to use some of these protocols in the Mars Reconnaissance Orbiter that was launched a few weeks ago.
What’s your vision for space communications?
Cerf: In the long run, I hope to see standardized space networking provide a kind of backbone similar in spirit to that which has evolved for the Internet. As each new science mission is launched, it can carry the standard protocols on board and these can interwork with and make use of earlier mission assets that may still be in place and operational.