We have built an Ad hoc Protocol Evaluation testbed (APE) in order to perform large-scale, reproducible experiments. APE aims at assessing several different routing protocols in a real world environment instead of by simulation. We present the APE testbed architecture and report on initial experiments with up to 37 physical nodes that show the reproducibility and scalability of our approach. Several scenario scripts have been written that include strict choreographic instructions to the testers who walk around with ORiNOCO equipped laptops. We introduce a metric called Virtual Mobility that we use to compare different testruns. This metric is based on the measured signal quality instead of the geometric distance between nodes, hence it reflects how a routing protocol actually perceives the network's dynamics.

No abstract

Abstract. The NorduGrid project operates a production Grid infrastructure in Scandinavia and Finland using own innovative middleware solutions. The resources range from small test clusters at academic institutions to large farms at several supercomputer centers, and are used for various scientific applications. This talk reviews the architecture and describes the Grid services, implemented via the NorduGrid middleware.

No abstract

No abstract

No abstract

Innovative middleware solutions are key to the NorduGrid testbed, which spans academic institutes and supercomputing centers throughout Scandinavia and Finland and provides continuous grid services to its users. Academic researchers in the Nordic countries participate in many common projects that process large amounts of data. To function effectively, such collaborations need a grid computing infrastructure that works across a wide area and uses distributed resources efficiently. Initial evaluations of existing grid solutions, however, showed that they failed to meet this requirement. The Globus Toolkit (www.globus.org), for example, had no resource brokering capability, and the European Union's DataGrid project (EDG; www.edg.org) did not satisfy stability and reliability demands. Other grid projects had even less functionality.In May 2001, researchers at Scandinavian and Finnish academic institutes launched the NorduGrid project (www.nordugrid.org), with the goal of building a Nordic testbed for wide-area computing and data handling. We first developed a proposal for an original architecture and implementation.1,2 We then developed middleware based on the Globus libraries and API, adding a set of completely new services such as resource brokering and monitoring. The NorduGrid middleware thus preserves Globus compatibility and permits interoperability with other Globus-based solutions. It also meets our goal of providing a lightweight, yet robust solution that is noninvasive, portable, and requires minimal intervention from system administrators. We launched our testbed in May 2002, and it has been continuously operating since August of that year, providing reliable, round-the-clock services for academic users. NorduGrid spans several countries and incorporates several national computing centers, making it one of the largest operational grids in the world. NorduGrid OverviewWe built the NorduGrid testbed using the resources of national supercomputer centers and academic institutes, and based it on the Nordic countries' academic networks. Currently, the testbed uses more than 900 CPUs in 20 clusters that range in size from 4 to 400 processors. Except for a few test clusters, the resources are not grid-dedicated. Among the active NorduGrid users are high-energy physics researchers who use the Grid daily and physics theorists who use it to perform complex computational tasks. 3,4Guiding Philosophy We planned and designed the NorduGrid architecture to satisfy the needs of both users and system administrators. These needs constitute a general philosophy:• Start with something simple that works.• Avoid single points of failure.• Give resource owners full control over their resources.• Ensure that NorduGrid software can use the existing system and, eventually, other preinstalled grid middleware, such as different Globus versions.• Leave installation details (method, operating system, configuration, and so on) up to system administrators.• Pose as few restrictions on site configuration as possible -for example, permit c...

No abstract