We are a society that is seemingly obsessed with putting labels or signs on everything -- from car bumpers to blimps to items that exist only as ones and zeros in the ether. Case in point, the excellent piece ("Getting Beyond the Compute Grid: The Challenge of 'Grid 2.0'") by William Fellows of The 451 Group on Grid 2.0 that appeared last month in GRIDtoday.

One of the things I liked most about this feature were the elegant and simple definitions given for the terms "virtualization" and "service-oriented architecture," and the differences between Grid 1.0 and Grid 2.0 -- especially as they relate to the five levels of the 451 Grid Adoption model. In a nutshell, Grid 2.0, even though it really is just a marketing term today, adds storage network and data resource management to the services provided in Grid 1.0.

One of the things that bothered me a bit from the report was the following: "When asked, 70 percent of early adopters who responded to a survey said there is a better term than 'Grid' to describe their distributed computing architectures: 23 percent said virtualization, 23 percent said HPC, 19 percent said utility computing, 19 percent said clustering, and 15 percent said SOA."

So, if 70 percent of people surveyed by one of the leading analyst firms in the space do not believe that the term "Grid" suffices, are we really ready for Grid 2.0? Perhaps, but what is it going to take to evolve this term from marketing tool to implemented reality?

I really like the five levels of the 451 Grid Adoption model, and it got me thinking back to the tried and true seven layers of the Open Systems Interconnection (OSI) Reference Model that we all learned about in our computer networking courses.

The majority of grid implementations and even middleware "solutions" take the bottom-up approach, building the network and associated infrastructure and hoping applications are built that will utilize these services. I’ve long preached that we need to concentrate on the top-down approach and actually build, or repurpose, applications that take advantage of potential Grid infrastructures. There are merits to both approaches, and it has caused me to do a bit of rethinking and evolve my belief into a hybrid that I’ll call the "Grid sandwich."

There currently are existing physical network implementations and applications that work well in all five levels of the 451 Grid Adoption model. This is the bread of the sandwich. We need to be concentrating on what we put inside that sandwich and how those ingredients blend with our bread choices. The right combinations will be revealed by experimentation over time and, although not specifically right or wrong, will settle into socially accepted norms. While I’m sure there are those who would argue that rye bread is perfectly acceptable for peanut butter and jelly, I suspect the majority would rather choose corned beef with their rye. Note that I am not excluding a new application or physical network implementation any more than I would a new type of bread.

Fellows' article mentions "existing applications and application servers are also being deployed into Grid-enabled containers without requiring prior knowledge of the infrastructure, and legacy applications are being wrapped for grids." That’s great, but I haven’t seen these featured to the extent that I believe the traction exists. Nor have I seen many that I would consider mainstream applications. If what we seek is mainstream acceptance of Grid, we need to be building our sandwiches with bread we can quickly run to the store and get, not something we have to bake ourselves with eclectic ingredients. Sure, I like a homemade focaccia, but when I have to crank out a bunch of sandwiches for the kids, I’m going to grab a trusty loaf of Wonder Bread. Likewise, we need to be concentrating on mainstream physical networks and applications.

OK, enough of the analogies. The common network choices and vendors, at least from a physical layer, are pretty obvious. But what about the applications? I’ve mentioned this before, and the idea is not originally mine, but the majority of financial calculations are still done using spreadsheets with large overhead, interpreted languages. I think this is a pretty good place to start. I’ve also thought a Grid-enabled version of the Apache Web server would be pretty interesting, both for load balancing and redundant distribution of the data being served.

I really like the way Fellows fit the Grid 1.0/Grid 2.0 definitions into the five levels of the 451 Grid Adoption model. I’ll think about what I said above and try to classify it into taxonomy, perhaps multi-dimensional, that is just as elegant. I guess I’m just as guilty of being obsessed with putting labels or signs on everything.