Barry Linnert

The virtual resource manager: an architecture for SLA-aware resource management

The next generation Grid will demand the Grid middleware to provide flexibility, transparency, and reliability. This implies the appliance of service level agreements to guarantee a negotiated level of quality of service. These requirements also affect the local resource management systems providing resources for the Grid. At this a gap between these demands and the features of todays resource management systems becomes apparent. In this paper we present an approach which closes this gap. Introducing the architecture of the virtual resource manager we highlight its main features of runtime responsibility, resource virtualization, information hiding, autonomy provision, and smooth integration of existing resource management system installations.

Distributed dynamic processor allocation for multicomputers

Current processor allocation techniques for highly parallel systems use centralized front-end based algorithms which restrict applied strategies to static allocation, low parallelism, and weak fault tolerance. To lift these restrictions, we are investigating a distributed approach to processor allocation in multicomputers where currently no centralized data structure with information about the state of all processors exists. This approach will allow the implementation of more complex allocation schemes and possibly the consideration of dynamic allocation, where parallel applications would be able to adapt the allocated processor partition to its actual demand at running time, resulting in a more efficient utilization of system resources. Noncontiguous versions of a distributed dynamic processor allocation scheme are proposed and studied in this paper as an alternative for parallel programming models to allow dynamic creation and task deletion. Simulations compare the performance of the proposed dynamic strategies with static counterparts and also with well-known centralized algorithms in an environment with growing and shrinking processor demands. To demonstrate dynamic allocation is feasible with current technologies, results of the experiments are presented for a 96 nodes SCI hpcLine Primergy Server cluster.

Rerouting strategies for networks with advance reservations

Network transmissions in high performance networking scenarios, e.g., used for e-science or grid applications, require quality-of-service guarantees concerning bandwidth availability, but also timing constraints, e.g., deadlines, must be met. Current research efforts concentrate on supporting such environments with SLA-aware advance reservation management systems. Hence, the robustness of the management system against network failures is an important issue, especially since failures frequently occur in networks. Since accurate knowledge about the failure duration is unlikely available and estimations lead to considerably degraded performance, in this paper we present a novel load-based approach for dealing with link failures in advance reservation environments. The approach does not rely on prediction of the downtime, but instead reroutes flows only based on available information about the network

Distributed workflow management for large-scale grid environments

Workflow management in large-scale grid environments is a very challenging task centralized management systems are not able to cover sufficiently. Therefore, we present our workflow on-line resource management (WORM) architecture built on top of active network technology. The approach integrates a peer-to-peer like organized workflow management system with existing or newly built management systems for the resources building the grid. In our approach, each workflow is represented by a mobile autonomous entity which uses the active network infrastructure to move through the grid, which is represented by an active overlay network on top of existing network infrastructure. Thus, control of the workflow execution is handed over to the autonomous code without requiring a central system to be in charge of the computation and cope with reservation, failures, etc. The WORM architecture is presented together with a classification into the taxonomy of workflow management systems.

VRM: a failure-aware Grid resource management system

For resource management in grid environments, advance reservations turned out to be very useful and hence are supported by a variety of grid toolkits. However, failure recovery for such systems has not yet received the attention it deserves. In this paper, we address the problem of remapping reservations to other resources, when the originally selected resource fails. Instead of dealing with jobs already running, which usually means checkpointing and migration, our focus is on jobs that are scheduled on the failed resource for a specific future period of time but not started yet. The most critical factor when solving this problem is the estimation of the downtime. We avoid the drawbacks of under- or overestimating the downtime by a dynamic load-based approach that is evaluated by extensive simulations in a grid environment and shows superior performance compared to estimation-based approaches.

VRM: a failure-aware grid resource management system

For resource management in grid environments, advance reservations turned out to be very useful and hence are supported by a variety of grid toolkits. However, failure recovery for such systems has not yet received the attention it deserves. In this paper, we address the problem of remapping reservations to other resources, when the originally selected resource fails. Instead of dealing with jobs already running, which usually means checkpointing and migration, our focus is on jobs that are scheduled on the failed resource for a specific future period of time but not started yet. The most critical factor when solving this problem is the estimation of the downtime. We avoid the drawbacks of under- or overestimating the downtime by a dynamic load-based approach that is evaluated by extensive simulations in a grid environment and shows superior performance compared to estimation-based approaches.

A distributed load-based failure recovery mechanism for advance reservation environments

Resource reservations in advance are a mature concept for the allocation of various resources, particularly in grid environments. Common grid tool kits support advance reservations and assign jobs to resources at admission time. In such a distributed environment, it is necessary to develop carefully tailored failure recovery mechanisms that provide seamless transparent migration of jobs from one resource to another. As the migration of running jobs is difficult, an important issue in advance reservation, i.e., planning based, management infrastructures is to determine the duration of a failure in order to remap jobs that are already allocated to a currently failed resource but not yet active. As shown in previous work, underestimations of the failure duration and as a consequence the remapping of too few jobs results in an increased amount of job terminations. In order to overcome this drawback, we propose a load-based computation of the jobs to be remapped. A centralized and a distributed version of the strategy are presented, showing it is not necessary to have knowledge beyond the local allocation on the failed resource. These load-based strategies achieve effective remapping of jobs while avoiding - inevitably inaccurate - estimations of the failure duration.

Failure Recovery in Distributed Environments with Advance Reservation Management Systems

Resource reservations in advance are a mature concept for the allocation of various resources, particularly in grid environments. Common grid toolkits such as Globus support advance reservations and assign jobs to resources at admission time. While the allocation mechanisms for advance reservations are available in current grid management systems, in case of failures the advance reservation perspective demands for strategies that support more than recovery of jobs or applications that are active at the time the resource failure occurs. Instead, also already admitted, but not yet started applications are affected by the failure and hence, need to be dealt with in an appropriate manner. In this paper, we discuss the properties of advance reservations with respect to failure recovery and outline a number of strategies applicable in such cases in order to reduce the impact of resource failures and outages. It can be shown that it pays to remap also affected but not yet started jobs to alternative resources if available. Alike reserving in advance, this can be considered as remapping in advance. In particular, a remapping strategy that prefers requests that were allocated a long time ago, provides a high fairness for clients as it implements similar functionality as advance reservations, while achieving the same performance as the other strategies.

The Virtual Resource Manager: Local Autonomy Versus QoS Guarantees for Grid Applications

In this paper, we describe the architecture of the virtual resource manager VRM, a management system designed to reside on top of local resource management systems for cluster computers and other kinds of resources. The most important feature of the VRM is its capability to handle quality-of-service (QoS) guarantees and service-level agreements (SLAs). The particular emphasis of the paper is on the various opportunities to deal with local autonomy for resource management systems not supporting SLAs. As local administrators may not want to hand over complete control to the Grid management, it is necessary to define strategies that deal with this issue. Local autonomy should be retained as much as possible while providing reliability and QoS guarantees for Grid applications, e.g., specified as SLAs.

A quality-of-service architecture for future grid computing applications

The next generation grid applications demand grid middleware for a flexible negotiation mechanism supporting various ways of quality-of-service (QoS) guarantees. In this context, a QoS guarantee covers simultaneous allocations of various kinds of different resources, such as processor runtime, storage capacity, or network bandwidth, which are specified in the form of service level agreements (SLA). Currently, a gap exists between the capabilities of grid middleware and the underlying resource management systems concerning their support for QoS and SLA negotiation. In this paper we present an approach which closes this gap. Introducing the architecture of the virtual resource manager, we highlight its main QoS management features like run-time responsibility, co-allocation, and fault tolerance.