While the systems vendors and hardware vendors have somewhat dominated the enterprise Grid discussions in the media to date -- there have been a handful of networking players busily using and researching the Globus Toolkit for almost ten years. Today, there are a number of new developments (see Ian Foster's column in this issue of the GCJ; as well as the Q&A with Franco Travostino from Nortel) that suggest the networking and telco players are about to get a little louder about their role in the evolution of Grid.

This rise in Grid interest from the networking arena was also evidenced by the surveys we conducted at GlobusWORLD in February. The attendee survey from GlobusWORLD 2005 showed that 13% of our enterprise attendees were from telecom, eclipsed only by the ever-present computer services vertical market.

In a recent column in "The Economist" a journalist mused that after the recent acquisition of MCI by Verizon, there are essentially two large firms (the other being SBC) running the phone systems in the United States, and that we are again approaching a telecom landscape where one giant runs the show. Such was the case before 1984 when a federal judge ordered the breakup of AT&T. The article points out that while this may be true of traditional wired phone service, the nature of the telecom landscape is quite different than it was in 1984. Instead of wired phones being the "one trick pony" of telecommunications, there are a multitude of technologies open even to the home consumer including wireless services, broadband data services, and digital cable systems with the lines of what types of media they carry (voice, video, and data) completely blurred.

Given this new paradigm in the telecommunications market, where does Grid computing fit in with these traditional players' strategies - and where do these companies fit within the world of Grid computing?

One of the promises of Grid is the ability to make better use of existing resources and cheaper commodity hardware. In a nutshell, the computer services vertical market often speaks of Grid in terms of an aggregator of system resources and data. We believe that Grid will serve a similar function for the telco space. Not only does the new landscape of telecommunication present a variety of technologies, often one traditional service provider will offer a wide spectrum of these services, and perhaps even host various types of media over non-traditional carriers like voice over IP. With open Grid standards, and platform and operating system agnosticism that Grid middleware such as the Globus Toolkit touts, Grid computing may be the great aggregator and data translation platform for these disparate systems.

When looking for an answer to how traditional telcos fit into Grid, the initial reaction is often to look at either the compute Grid, or data Grid taxonomies. While the current physical network is capable of handling the demands of compute Grids, the sheer volume of data and transactions in data Grids could potentially introduce burdens on the physical network. It will be interesting, if indeed these constraints pose a problem, to watch how the service providers approach the new challenges that Grid places on the networking infrastructure. Or perhaps Grid can be its own solution, by dividing up mass data transfers into smaller quanta and utilizing multiple hosts (known as striping) as GridFTP in the latest version of the Globus Toolkit is capable of.

These are the early, formative years of enterprise Grid computing - there are some great opportunities for networking vendors to participate in the evolution of the open source Grid standards, and to establish early Grid networking leadership positions. I encourage the service providers, network hardware vendors and integrators to contact me if you'd like to learn more about how to get involved with the Globus Consortium.

-- Greg Nawrocki

close window