Alberto Gianoli

Virtual Organization Management Across Middleware Boundaries

One of the most important challenges in production grids is to achieve interoperation across several heterogeneous grid middleware platforms: escience applications need a coordinated resource sharing among dynamic collections of individuals/institutions, independently from whatever middleware the resources are running. For this reason, there is a great effort going on to define standard interfaces, in order to implement common services that can be used to achieve cross-middlewares interoperability. In this paper, we present our modifications to the virtual organization management service (VOMS), a widely-known and used tool that acts as an attribute authority. We enhanced VOMS to expose the standardized interface of the Security Assertion Markup Language (SAML), and therefore to release SAML assertions. This way we want VOMS to be available on the larger possible number of grid middleware platforms.

Computing at SuperB

Domenico Del Prete*, Fabrizio Bianchi, Vania Boccia, Vincenzo Ciaschini, Marco Corvo, Guglielmo De Nardo, Andrea Di Simone, Giacinto Donvito, Armando Fella, Paolo Franchini, Francesco Giacomini, Alberto Gianoli, Giuliano Laccetti, Stefano Longo, Steffen Luitz, Eleonora Luppi, Matteo Manzali, Leonardo Merola, Silvio Pardi, Alejandro Perez, Matteo Rama, Guido Russo, Bruno Santeramo, Roberto Stroili, Luca Tommasetti

A parallel framework for the SuperB super flavor factory

The SuperB asymmetric energy e+e- collider and detector [1] to be built at the newly founded Nicola Cabibbo Lab [2] will provide a uniquely sensitive probe of New Physics in the flavor sector of the Standard Model. Studying minute effects in the heavy quark and heavy lepton sectors requires a data sample of 75 ab-1 and a peak luminosity of 1036 cm-2s-1. These parameters require a substantial growth in computing requirements and performances: we roughly estimate that in few years of operations we will have to cope with near half EB of raw data and that the CPU required for processing will be close to 6000 KHep-Spec06 per year. The SuperB collaboration is thus investigating the advantages of new CPU architectures (multi and many cores, now largely available on the market), with the aim to be able to treat this amount of data both efficiently and within reasonable amounts of time. At the same time the collaboration is analyzing the current software, in large part inherited from previous experiments (mainly BaBar), to understand the underlying parallelism level, how to exploit it and to find the better mapping to emergent hardware architectures. In this work we first present the measurements done on the analysis and simulation software, then the Framework architecture we are designing. We complete the presentation with a description of our Framework prototype and some preliminary performance measurements.

A prototype suite for data-analysis management of the SuperB experiment

The SuperB asymmetric e+e- collider and detector to be built at the newly founded Nicola Cabibbo Lab will provide a uniquely sensitive probe of New Physics in the flavor sector of the Standard Model. Studying minute effects in the heavy quark and heavy lepton sectors requires a data sample of 75ab-1 and a peak luminosity of 1036cm-2s-1. Providing a user-friendly solution to workload management in a distributed resource system is one of the key goal of a HEP community as SuperB experiment. Physicists involved in Monte Carlo simulation productions and in data analysis should be able to perform job management and basic data transfer operations limiting as possible training costs and maximizing flexibility in resource exploitation. The SuperB computing group adopted Ganga as the interface layer of such distributed analysis infrastructure. A SuperB-specific Ganga-plugin has been developed to accomplish experiment requirements as information system interface, use cases implementation, dataset management and job wrapper interactions. This work will present the Ganga plugin design and implementation, and its integration with the wider distributed system adopted by the collaboration.

SuperB production system for simulated events

The SuperB asymmetric e+e- collider and detector to be built at the newly founded Nicola Cabibbo Lab will provide a uniquely sensitive probe of New Physics in the flavor sector of the Standard Model. Studying minute effects in the heavy quark and heavy lepton sectors requires a data sample of 75 ab-1 and a peak luminosity of 1036 cm-2 s-1. The SuperB Computing group is working on developing a production system for distributed event simulation capable to support the detector design definition and its performance evaluation studies. During last year the system has evolved from the point of view of job workflow, Grid services, and technologies adoption. A complete code refactoring and sub-component language porting now allows to sustain large distributed production cycles involving resources from three continents and Grid Flavors. In this paper we present a complete description of the production system, its evolution and its integration with Grid services; in particular, we will focus on the utilization of new Grid component features as in LB and WMS version 3.

Exploiting new CPU architectures in the SuperB software framework

The SuperB asymmetric-energy e+e? collider and detector to be built at the newly founded Nicola Cabibbo Lab will provide a uniquely sensitive probe of New Physics in the flavour sector of the Standard Model. Studying minute effects in the heavy quark and heavy lepton sectors requires a data sample of 75ab?1 and a luminosity target of 1036cm?2s?1. These parameters require a substantial growth in computing requirements and performances. The SuperB collaboration is thus investigating the advantages of new CPU architectures (multi and many cores) and how to exploit their capability of task parallelization in the framework for simulation and analysis software. In this work we present the underlying architecture which we intend to use and some preliminary performance results of the first framework prototype.

SuperB evaluation of DIRAC Distributed Infrastructure

The SuperB asymmetric energy e+e− collider and detector to be built at the newly founded Nicola Cabibbo Lab will provide a uniquely sensitive probe of New Physics in the flavour sector of the Standard Model. SuperB distributed computing group performed a detailed evaluation of DIRAC Distributed Infrastructure for use in the SuperB experiment based on the two use cases: End User Analysis and Monte Carlo Production. Test aims to evaluate DIRAC capabilities to manage both gLite and OSG sites, File Catalog management, job and data management features in SuperB realistic use cases.

SuperB Simulation Production System

The SuperB asymmetric e+e− collider and detector to be built at the newly founded Nicola Cabibbo Lab will provide a uniquely sensitive probe of New Physics in the flavor sector of the Standard Model. Studying minute effects in the heavy quark and heavy lepton sectors requires a data sample of 75 ab−1 and a peak luminosity of 1036 cm−2 s−1. The SuperB Computing group is working on developing a simulation production framework capable to satisfy the experiment needs. It provides access to distributed resources in order to support both the detector design definition and its performance evaluation studies. During last year the framework has evolved from the point of view of job workflow, Grid services interfaces and technologies adoption. A complete code refactoring and sub-component language porting now permits the framework to sustain distributed production involving resources from two continents and Grid Flavors. In this paper we will report a complete description of the production system status of the art, its evolution and its integration with Grid services; in particular, we will focus on the utilization of new Grid component features as in LB and WMS version 3. Results from the last official SuperB production cycle will be reported.

Distributed policy framework across multiple grid domains

A key feature of grid environment is the sharing of computing and storage: users operate on resources not directly owned by them. Often users working on the same research project are grouped in a virtual organization (VO) to use a common authorization policy on this resources. Many international experiments, however, use different Grid middleware platforms with their own authorization framework. This leads to interoperability problems for scientists of the same experiment, using their national Grid infrastructure. Usually VOs and resource providers share contracts to regulate resource usage. The enforcement of such arrangements needs an agreed interoperable authorization mechanism based on policies that can be written by VOs and resources providers. This process can be applied using a flexible and distributed policy framework, where complex relationships can be enforced being able to manage both policies created by VOs and policies created by Grid sites. G-PBox policy framework, in conjunction with VOMS Attribute Authority, is our proposal to represent, manage and distribute such policies in a transparent way. G-PBox approach is based on a set of XACML policies databases belonging separately to VOs and resource providers, each containing at least policies regarding it own organization. In this paper we describe how VO oriented tools like VOMS and G-PBox can be deployed across different VOs and resource providers. It will show how VO managers and sites administrators can set up agreed policies for resource sharing optimization and experiment computing prioritization, making best use of their time and resources. It will underline also that adoption of assertion and policy Grid standard, as SAML and XACML, provides an effective advantage in order to allow an accepted authentication and authorization interoperability among services of different Grid domains based on different mechanisms.