GCJ: Why did Nortel Join the Globus Consortium, and how would you describe the company's interest in Grid?

Travostino: Nortel is a global networking player and we value our role in delivering mission-critical, end-to-end networking solutions. As such, we continually follow the Grid industry so we can adapt our networking capabilities appropriately. Without question, the Grid paradigm shift prompts for a very interesting new networking direction. By keeping an inside view into Globus, we are forging a tighter integration between what happens at the systems level and what happens in the network. While other members of the Grid community focus on the orchestration of just CPU and data, Nortel is working closely with Globus to understand how the Grid is evolving - and we're building expertise on how to perform CPU and data and network orchestration. Our key focus, as you might expect, isn't to ship Globus software but to provide the best possible network for Globus users.

Nortel is joined by several other networking vendors interested in Grid. I've met representatives from Cisco, Alcatel, Force10, and other networking vendors who are very much interested in the Grid phenomena. Their timing and strategies for Grid are only known to them right now, but I certainly hope the broad networking industry interest will continue, because it's important to the evolution of the infrastructure and the so-called middleware. Between the Grid software and the network, we need to have great representation of various vendors so that we come to agreement around standards in the quickest way that will be conducive to greater adoption and greater scaling of Grids.

At GlobusWORLD in February - in the Nortel booth we showed the first demo of Globus 3.95 interfacing with our network middleware, DRAC, using WSRF interfaces. We are paying close attention to what Globus does, and in fact Nortel was the first vendor to show a live demo of how a commercial product interacts with a pre-GT4 release. That's a testimony to Nortel's strong interest in what's happening in the Globus community.

GCJ: We've been hearing about Nortel's Dynamic Resource Allocation Controller (DRAC) ... can you tell us why this effort is significant?

Travostino: Our DRAC is the embodiment of the concept of network middleware - the middleman between grid infrastructure like Globus and the network. It elevates the network's role in the Grid - adding it to the Grid resources that have traditionally been CPU and data. DRAC is a key capability by which we allow CPU plus data plus network orchestration. It adapts the network to the Grid application, as opposed to the other way around, which most people have done - which is to adapt the Grid to the network.

DRAC has early adopters, test beds and customers. For example, we've been awarded a bid for the SURFnet network in the Netherlands, which is a countrywide network for research and education, with large representation of Grid users. The operational experiences we're deriving from that particular network have taught us a lot of lessons on how we need to design our next-generation platforms and network gear so that Grid requirements are properly addressed. DRAC plays a key role of deciding how to map SURFnet traffic to particular network resources, and it's been very successful so far. The Surfnet contract was awarded to Nortel in March of 2004 and is a four year program.

GCJ: Is the Globus Toolkit present in that SURFnet Grids?

Travostino: Yes, Globus users are also present in that scenario.

GJC: Let's talk about some of the network-specific challenges for Grid.

Travostino: It's not just a matter of creating bigger pipes and expecting that we're going to have greater bandwidth. That will be one key ingredient, but not the only one. In addition to needing to blast as much data as possible, equally important there is the aspect of control of the network, so you can gain virtualized control and access to the network and do, for example, time of the day reservations if you need to - or be able to change the security properties, or the routing properties, or the service discovery properties. It's not about brute force and how many bits per second you can send - but also about how to finesse the network experience. We want to remove all sorts of GUI-driven and operator-driven elements so that the network can be controlled by software, and there are specific feedback loops going on between network and software, without operators involved. We want an addressing schema capable to support hitless joins in a virtual organization without collisions and impact from intervening network address translation boxes or firewalls. So there are very important features above and beyond the basic capability of sending a lot of data through a pipe. Those are the issues that we're most interested in.

GCJ: Given the history of Nortel selling to carriers, how do you see the carriers jumping into the Grid evolution?

Travostino: For Grids to scale and to gain the type of dependability that we're seeing with telephony and other services, we need to have network providers as part of the Grid ecosystems. In my opinion, network providers can play three different, orthogonal roles here. The first is that network providers can be enablers for new network functionalities that Grid users would want -- for example, bandwidth on demand and bigger pipes. The second dimension is where network providers use Grid infrastructure such as Globus internally, for billing, planning, resource modeling and such. And the third, which is probably the most attractive to carriers, is where network providers grow from selling bandwidth and networking into selling network, CPU and data services, thus becoming a provider of Grid managed services. For this role, they would build directly upon their strength of already having a way to do services and billing and dependable customer relationship. This is a very interesting opportunity, in my opinion, because it allows network providers to escape the commoditization of bandwidth and voice services - and to provide new services to the customers by solving a number of different IT problems with a utility type of business.

Several providers have responded to those three different thrusts that I've mentioned - and in fact I'm helping to form a group within the Global Grid Forum called the Telco Research Group. This will be an interesting new voice for the carrier priorities and concerns around Grid. The GGF is an entirely open forum. I invite network providers, network vendors, and software vendors to join this new effort to drive the understanding of the network middleware further and reach consensus on what the interfaces and APIs should look like. We all want solutions that interoperate and fuel greater market adoption for Grids.

close window