Josva Kleist

A distributed storage system with dCache

The LCG collaboration is encompassed by a number of Tier 1 centers. The Nordic LCG Tier 1, operated by NDGF, is in contrast to many other Tier 1 centers distributed over the Nordic countries. A distributed setup was chosen for both political and technical reasons, but also provides a number of unique challenges. dCache is well known and respected as a powerful distributed storage resource manager, and was chosen for implementing the storage aspects of the Nordic Tier 1. In contrast to classic dCache deployments, we deploy dCache over a WAN with limited bandwidth, high latency, frequent network failures, and spanning many administrative domains. These properties provide unique challenges, covering topics such as security, administration, maintenance, upgradability, reliability, and performance. Our initial focus has been on implementing the GFD.47 OGF recommendation (which introduced the GridFTP 2 protocol) in dCache and the Globus Toolkit. Compared to GridFTP 1, GridFTP 2 allows for more intelligent data flow between clients and storage pools, thus enabling more efficient use of our limited bandwidth.

Mobile objects as mobile processes

Obliq is a lexically scoped, distributed, object-based programming language. In Obliq, the migration of an object is proposed as creating a clone of the object at the target site, whereafter the original object is turned into an alias for the clone. Obliq has only an informal semantics, so there is no proof that this style of migration is safe, i.e., transparent to object clients. In previous work, we introduced ojeblik, an abstraction of Obliq, where, by lexical scoping, sites have been abstracted away. We used ojeblik in order to exhibit how the semantics behind Obliqs implementation renders migration unsafe. We also suggested a modified semantics that we conjectured instead to be safe. In this paper, we rewrite our modified semantics of ojeblik in terms of the π-calculus, and we use it to formally prove the correctness of object surrogation, the abstraction of object migration in Ojeblik.

Aliasing Models for Object Migration

In Obliq, a lexically scoped, distributed, object-oriented programming language, object migration was suggested as the creation of a copy of an objects state at the target site, followed by turning the object itself into an alias, also called surrogate, for the remote copy. We consider the creation of object surrogates as an abstraction of the abovementioned style of migration. We introduce Ojeblik, a distribution-free subset of Obliq, and provide three different conguratlaion-style semantics, which only differ in the respective aliasing model. We show that two of the semantics, one of which matches Obliqs implementation, render migration unsafe, while our new proposal for a third semantics is provably safe. Our work suggests a straightforward repair of Obliqs aliasing model such that it allows programs to safely migrate objects.

Interoperability between ARC and gLite - Understanding the Grid-Job Life Cycle

ARC and gLite are two of the leading production-ready Grid middleware solutions being used by thousands of researchers every day. Even though the middlewares leverage the same technologies, there are substantial architectural and implementation divergences. Today, users face difficulties trying to cross the boundaries of the two systems. The gLite clients have so far not been capable of accessing ARC resources and vice versa. This paper is a follow up on an earlier proposal on how to enable interoperability between these two middlewares. Further, the paper presents a thorough walkthrough of the protocols and steps involved in the submission of a job in grids built up of the two different middlewares.

Roadmap for the ARC grid middleware

The Advanced Resource Connector (ARC) or the NorduGrid middleware is an open source software solution enabling production quality computational and data Grids, with special emphasis on scalability, stability, reliability and performance. Since its first release in May 2002, the middleware is deployed and being used in production environments. This paper aims to present the future development directions and plans of the ARC middleware in terms of outlining the software development roadmap.

Imperative objects as mobile processes

An interpretation of Abadi and Cardellis first-order Imperative c-calculus into a typed π-calculus is presented. The interpretation validates the subtyping relation and the typing judgments of the c-calculus, and is computationally adequate. The proof of computational adequacy makes use of (a π-calculus version) of ready simulation, and of a factorization of the interpretation into a functional part and a very simple imperative part. The interpretation can be extended to accommodate various type features. The interpretation can be used to compare and contrast the Imperative and the Functional c-calculus, and to prove properties about them, within a unified framework.

Local pi-Calculus at Work: Mobile Objects as Mobile Processes

Obliq is a lexically-scoped, distributed, object-based programming language. In Obliq, the migration of an object is proposed as creating a clone of the object at the target site, whereafter the original object is turned into an alias for the clone. Obliq has only an informal semantics, so there is no proof that this style of migration is safe, i.e., transparent to object clients. In previous work, we introduced Obliq, an abstraction of Obliq, where, by lexical scoping, sites have been abstracted away. We used Obliq in order to exhibit how the semantics behind Obliqs implementation renders migration unsafe. We also suggested a modified semantics that we conjectured instead to be safe. In this paper, we rewrite our modified semantics of Obliq in terms of pi-calculus, and we use it to formally prove the correctness of object surrogation, the abstraction of object migration in Obliq.

Grid Interoperation with ARC Middleware for the CMS Experiment

The Compact Muon Solenoid (CMS) is one of the general purpose experiments at the CERN Large Hadron Collider (LHC). CMS computing relies on different grid infrastructures to provide computational and storage resources. The major grid middleware stacks used for CMS computing are gLite, Open Science Grid (OSG) and ARC (Advanced Resource Connector). Helsinki Institute of Physics (HIP) hosts one of the Tier-2 centers for CMS computing. CMS Tier-2 centers operate software systems for data transfers (PhEDEx), Monte Carlo production (ProdAgent) and data analysis (CRAB). In order to provide the Tier-2 services for CMS, HIP uses tools and components from both ARC and gLite grid middleware stacks. Interoperation between grid systems is a challenging problem and HIP uses two different solutions to provide the needed services. The first solution is based on gLite-ARC grid level interoperability. This allows to use ARC resources in CMS without modifying the CMS application software. The second solution is based on developing specific ARC plugins in CMS software.

Energy savings in dynamic and resilient optical networks based on traffic-aware strategies

An analysis of the energy savings is presented when taking into account a complete traffic model for a one-year time period. Daily and weekly traffic fluctuations as well as yearly traffic growth are considered when analyzing the power consumption. Low power mode in optoelectronic devices (sleep-mode and rate adaptive capabilities) is implemented within a traffic-aware networking approach. The impact of dedicated 1:1 protection in a dynamic network scenario is considered and a comparison is made with the unprotected case. A GMPLS control plane is implemented and used to re-configure the power-adaptive devices and connections. Results show that symbol-rate adaptation provides high savings for unprotected scenarios (37% energy savings w.r.t. unprotected Baseline), while for the protected scenarios better results are obtained for modulation format adaptation which includes sleep-mode (57.1% energy savings w.r.t. protected Baseline). Moreover, compared to the Baseline scenarios the Mixed adaptation, combining both symbol-rate and modulation format, is the most power-efficient strategy providing 39% energy savings for unprotected scenario and 70% energy savings for dedicated protection scenario.

Recent ARC developments: through modularity to interoperability

The Advanced Resource Connector (ARC) middleware introduced by NorduGrid is one of the basic Grid solutions used by scientists worldwide. While being well-proven in daily use by a wide variety of scientific applications at large-scale infrastructures like the Nordic DataGrid Facility (NDGF) and smaller scale projects, production ARC of today is still largely based on conventional Grid technologies and custom interfaces introduced a decade ago. In order to guarantee sustainability, true cross-system portability and standards-compliance based interoperability, the ARC community undertakes a massive effort of implementing modular Web Service (WS) approach into the middleware. With support from the EU KnowARC project, new components were introduced and the existing key ARC services got extended with WS technology based standard-compliant interfaces following a service-oriented architecture. Such components include the hosting environment framework, the resource-coupled execution service, the re-engineered client library, the self-healing storage solution and the peer-to-peer information system, to name a few. Gradual introduction of these new services and client tools into the production middleware releases is carried out together with NDGF and thus ensures a smooth transition to the next generation Grid middleware. Standard interfaces and modularity of the new component design are essential for ARC contributions to the planned Universal Middleware Distribution of the European Grid Initiative.