Miroslav Ruda

Practical approaches to Grid workload and resource management in the EGEE project

Resource management and scheduling of distributed, data-driven applications in a Grid environment are challenging problems. Although significant results were achieved in the past few years, the development and the proper deployment of generic, reliable, standard components present issues that still need to be completely solved. Interested domains include workload management, resource discovery, resource matchmaking and brokering, accounting, authorization policies, resource access, reliability and dependability. The evolution towards a service-oriented architecture, supported by emerging standards, is another activity that will demand attention. All these issues are being tackled within the EU-funded EGEE project (Enabling Grids for E-science in Europe), whose primary goals are the provision of robust middleware components and the creation of a reliable and dependable Grid infrastructure to support e-Science applications. In this paper we present the plans and the preliminary activities aiming at providing adequate workload and resource management components, suitable to be deployed in a production-quality Grid.

The DataGrid Workload Management System: Challenges and Results

The workload management task of the DataGrid project was mandated to define and implement a suitable architecture for distributed scheduling and resource management in a Grid environment. The result was the design and implementation of a Grid Workload Management System, a super-scheduler with the distinguishing property of being able to take data access requirements into account when scheduling jobs to the available Grid resources. Many novel issues in various fields were faced such as resource management, resource reservation and co-allocation, Grid accounting. In this paper, the architecture and the functionality provided by the DataGrid Workload Management System are presented.

gLite job provenance

The Job Provenance (JP) service is designed to automate keeping track of computations on large scale Grids, giving thus users a tool to correctly archive information about their jobs and to re-submit any job in a reconstructed environment. JP provides a permanent minimal record of job (and its environment) related information, to which free-form user annotations can be added. JP also offers the capability of configuring any number of indexed logical views on the large collections of raw data, allowing efficient processing of even complex user queries selecting on both system data and the annotations. The scalable architecture, capable to handle millions of jobs in a single JP installation, and integrated into the EGEE gLite middleware environment is presented.

User Management for Virtual Organizations

Scalable and fine-grained Grid authorization requires moving away from a gridmapfile based access control and 1-to-l mappings to individual OS user accounts. This is recognized and addressed to by virtual organization (VO) authorization services, e. g. VOMS/LCAS and CAS. They, however, do not address user OS account management and isolation/sandboxing requirements, such as flexible pooling of accounts while maintaining auditing records. This paper describes some existing systems for user management for VOs and provides a list of requirements for a new user management system on which our current research is focused on.

Distributed Tracking, Storage, and Re-use of Job State Information on the Grid

The Logging and Bookkeeping service tracks jobs passing through the Grid. It collects important events generated by both the grid middleware components and applications, and processes them at a chosen LB server to provide the job state. The events are transported through secure and reliable channels. Job tracking is fully distributed and does not depend on a single information source, the robustness is achieved through speculative job state computation in case of reordered, delayed or lost events. The state computation is easily adaptable to modified job control flow.

Grid Infrastructure Monitoring as Reliable Information Service

A short overview of Grid infrastructure status monitoring is given followed by a discussion of key concepts for advanced status monitoring systems: passive information gathering based on direct application instrumentation, indirect one based on service and middleware instrumentation, multidimensional matrix testing, and on-demand active testing using non-dedicated user identities. We also propose an idea of augmenting information provided traditionally using Grid information services by information from the infrastructure status monitoring which gives verified and thus valid information only. The approach is demonstrated using a Testbed Status Monitoring Tool prototype developed for a GridLab project.

A uniform job monitoring service in multiple job universes

We describe an ongoing work of extending the gLite Logging and Bookkeeping (L&B) service to be able to track additional types of jobs, with the vision of being able to uniformly follow jobs on the Grid, even when they pass between different middleware domains. Details are given on the simpler case of PBS jobs, which prove the cababilityof L&B to deal with additional job types,as well as started more complex and challenging work on Condor jobs, where theimpact of eventual success is larger.

Experimental evaluation of job provenance in ATLAS environment

Grid middleware stacks, including gLite, matured into the state of being able to process up to millions of jobs per day. Logging and Bookkeeping, the gLite job-tracking service, keeps pace with this rate; however, it is not designed to provide a long-term archive of information on executed jobs. ATLAS — representative of a large user community — addresses this issue with its own job catalogue (ProdDB). Development of such a customized service, not easily reusable, took considerable effort which is not affordable by smaller communities. On the contrary, Job Provenance (JP), a generic gLite service designed for long-term archiving of information on executed jobs focusing on scalability, extensibility, uniform data view, and configurability, allows more specialized catalogues to be easily built. We present the first results of an experimental JP deployment for the ATLAS production infrastructure where a JP installation was fed with a part of ATLAS jobs, and also stress tested with real production data. The main outcome of this work is a demonstration that JP can complement large-scale application-specific job catalogue services, while serving a similar purpose where there are none available.

PRACTICAL EXPERIENCES WITH TORQUE META-SCHEDULING IN THE CZECH NATIONAL GRID

The Czech National Grid Infrastructure went through a complex transition in the last year. The production environment has been switched from a commercial batch system PBSPro, which was replaced by an open source alternative Torque batch system. This paper concentrates on two aspects of this transition. First, we will present our practical experience with Torque being used as a production ready batch system. Our modified version of Torque, with all the necessary PBSPro ex- clusive features re-implemented and further extended with new features like cloud-like behaviour, was deployed across the entire production environment, covering the entire Czech Republic for almost a full year. In the second part, we will present our work on meta-scheduling. This in- volves our work on distributed architecture and cloud-grid convergence. The distributed architecture was designed to overcome the limitations of a central server setup, which was originally used and presented stability and performance issues. While this paper does not discuss the inclusion of cloud interfaces into grids, it does present the dynamic infrastructure, which is a requirement for sharing the grid infrastructure between a batch system and a cloud gateway. We are also inviting everyone to try out our fork of the Torque batch system, which is now publicly available.

Job Provenance --- Insight into Very Large Provenance Datasets

Following the job-centric monitoring concept, Job Provenance (JP) service organizes provenance records on the per-job basis. It is designed to manage very large number of records, as was required in the EGEE project where it was developed originally. The quantitative aspect is also a focus of the presented demonstration. We show JP capability to retrieve data items of interest from a large dataset of full records of more than 1 million of jobs, to perform non-trivial transformation on those data, and organize the results in such a way that repeated interactive queries are possible. The application area of the demo is derived from that of previous Provenance Challenges. Though the topic of the demo -- a computational experiment -- is arranged rather artificially, the demonstration still delivers its main message that JP supports non-trivial transformations and interactive queries on large data sets.