Features:

TELECOM FOCUS SESSION AT GGF10 IN BERLIN
by Uwe Harms

At the Global Grid Forum, GGF10, in Berlin, Germany, we found more than 600 participants during March 10 to 12. Around the conference and the several working groups, there could be found a lot of additional events. In the Telco session more than 80 participants discussed the topic of grid and service providing. There have been different approaches concerning the research and scientific networks in contrary to the requests of that community to the commercial telcos, where they should deliver new software and billing systems.

The meeting of this session started early at 8.00 p.m. Thus it was impossible to reach Berlin at that time. Therefore I participated in the second part of that session.

In the early morning, the Grid Market Awareness Committee discussed the awareness of the grid technology in respect to telecommunications. Scientific grids are affecting research and scientific networks. These grids put unscheduled load on the networks, as the researchers and their grid application requires. On the other hand, the scientific networks, like the German DFN (German Research Network), get money from their users and the ministry to improve the bandwidth.

In the second part the group discussed: "How vendors of network equipment and applications that may be provided over networks are preparing for greater grid traffic on networks". Here vendors like Cisco, Juniper, Spanish Network, Alcatel, Foundry, Nortel and T-Systems presented their different view. Foundry sees manageability for the grid enablers as an important issue, especially across multiple geographic and different vendors. Then it is possible to use a technically unlimited grid, but it needs performance. For Nortel the network solutions track a complex ensemble of power laws. Technology enjoys a one size fits all, but their are different IPs, adapters, Ethernet etc. They want to support hybrid packages/Lambda Service as a defining design principle: packets for many to many, Lambdas for few to few. As a reference serves OMNInet as a Metrotestbed, which sets Lambda. Therfore the grid bill has more than one item: the bits + finesse, granularity of control + virtual access to control knots + resource bundling network + standing all alone that makes a lot of information The result is that for all a grid tool chain is necessary plus commercial available products.

Lees de Laat presented the Lambda Grid Software. He expects an available network at an average of 99.995%. On the users side he sees the business management layer, the service management and the element management layer, on the providers side the network element. High topics are to create the best route, to offer manageable elements and a policy based database.

The Poznan Supercomputer and Networking Center described their requirements in the Telco environment. Mew software my be required to support the grid applications, being used over layer 1? For example the new generation network like dark fibre, optical network requires it. Additionally new standards are needed, for the quality of services, APIs concerning the grid and telco networks, bandwidth/Lambdas on demand and a new business model for telcos - a task forces for that.

M. R. Hayley from IBM underpinned the need to support new telco services, offerings in the grid environment. Early grid services succeed, but assumed plentiful network resources. The telcos might be able to responsively allocate bandwidth. But they lack easy monitoring and provisioning capability. IBM sees telcos are shorting attitudes about grid: science demos ->business pilot. IBM plans Grid Telco Billing Pilot, what new software requires.

The telcos are business value driven, when investing in a grid infrastructure. Here the question of ROI and Tools are to be discussed. There is a need of a powerful integration of monitoring and provisioning at network layer. All OGSA and GGF had to be better integrated in standards. Telcos are international masters of billing for services.

From the discussions arise two contrary view points. The science and research grid scene wants the telcos to invest in software and services. But this community has no money for funding and buying the telcos services. If the enterprises use grids and buy the services, this will give an other picture of grid and telcos.