A novel approach for distributed application scheduling based on prediction of communication events

The evolution of commodity computing lead to the possibility of efficient usage of interconnected machines to solve computationally-intensive tasks, which were previously solvable only by using expensive supercomputers. This, however, required new methods for process scheduling and distribution, considering the network latency, communication cost, heterogeneous environments and distributed computing constraints. An efficient distribution of processes over such environments requires an adequate scheduling strategy, as the cost of inefficient process allocation is unacceptably high. Therefore, a knowledge and prediction of application behavior is essential to perform effective scheduling. In this paper, we overview the evolution of scheduling approaches, focusing on distributed environments. We also evaluate the current approaches for process behavior extraction and prediction, aiming at selecting an adequate technique for online prediction of application execution. Based on this evaluation, we propose a novel model for application behavior prediction, considering chaotic properties of such behavior and the automatic detection of critical execution points. The proposed model is applied and evaluated for process scheduling in cluster and grid computing environments. The obtained results demonstrate that prediction of the process behavior is essential for efficient scheduling in large-scale and heterogeneous distributed environments, outperforming conventional scheduling policies by a factor of 10, and even more in some cases. Furthermore, the proposed approach proves to be efficient for online predictions due to its low computational cost and good precision. (C) 2009 Elsevier B.V. All rights reserved.

EFFECTIVE NETWORKS FOR REAL-TIME DISTRIBUTED PROCESSING

This paper applies the concepts and methods of complex networks to the development of models and simulations of master-slave distributed real-time systems by introducing an upper bound in the allowable delivery time of the packets with computation results. Two representative interconnection models are taken into account: Uniformly random and scale free (Barabasi-Albert), including the presence of background traffic of packets. The obtained results include the identification of the uniformly random interconnectivity scheme as being largely more efficient than the scale-free counterpart. Also, increased latency tolerance of the application provides no help under congestion.

Elementary Petri net inside RFID distributed database (PNRD)

Usually, a Petri net is applied as an RFID model tool. This paper, otherwise, presents another approach to the Petri net concerning RFID systems. This approach, called elementary Petri net inside an RFID distributed database, or PNRD, is the first step to improve RFID and control systems integration, based on a formal data structure to identify and update the product state in real-time process execution, allowing automatic discovery of unexpected events during tag data capture. There are two main features in this approach: to use RFID tags as the object process expected database and last product state identification; and to apply Petri net analysis to automatically update the last product state registry during reader data capture. RFID reader data capture can be viewed, in Petri nets, as a direct analysis of locality for a specific transition that holds in a specific workflow. Following this direction, RFID readers storage Petri net control vector list related to each tag id is expected to be perceived. This paper presents PNRD cornerstones and a PNRD implementation example in software called DEMIS Distributed Environment in Manufacturing Information Systems.

A Unified Load Generator for Geographically Distributed Generation ofNetwork Traffic

During the last decade, the Internet usage has been growing at an enormous rate which has beenaccompanied by the developments of network applications (e.g., video conference, audio/videostreaming, E-learning, E-Commerce and real-time applications) and allows several types ofinformation including data, voice, picture and media streaming. While end-users are demandingvery high quality of service (QoS) from their service providers, network undergoes a complex trafficwhich leads the transmission bottlenecks. Considerable effort has been made to study thecharacteristics and the behavior of the Internet. Simulation modeling of computer networkcongestion is a profitable and effective technique which fulfills the requirements to evaluate theperformance and QoS of networks. To simulate a single congested link, simulation is run with asingle load generator while for a larger simulation with complex traffic, where the nodes are spreadacross different geographical locations generating distributed artificial loads is indispensable. Onesolution is to elaborate a load generation system based on master/slave architecture.

The effects of Kanban in software development teams : a study of the implementation at Sandvik

In software development organizations there is sometimes a need for change. In order to meet continuously increasing demands from their customers, Sandvik IT Services- SITS, at Sandvik in Sweden, required improving the way they worked with software development. Due to issues like a lot of work in progress and lot of simultaneous tasks for individuals in the teams that caused stress, it was almost impossible to address the question of working with improvements. In order to enable the improvement process Kanban was introduced in the software development teams. Kanban for software development is a change method created by David J. Anderson. The purpose of this thesis is twofold. One part is to assess what effects Kanban has had on the software development teams. The other part is to make a documentation of the Kanban implementation process at SITS. The documentation has been made on the basis of both company internal resources and observations of the Kanban implementation process. The effects of Kanban have been researched with an interview survey to the teams that have gone through the Kick start of the Kanban process. The result of the thesis is also twofold. One part of the result is an extensive documentation of the implementation process of Kanban at SITS. The other part is an assessment of the effects that Kanban has had at SITS. The major effects have been that the teams are experiencing less stress, more focus on quality and better customer collaboration. It is also evident is that it takes time for some effects to evolve when implementing Kanban

Semantic Distributed Messaging Middleware

Messaging middleware provides asynchronous communication between services in distributed environments. However, security, reliability and performance issues compel such middleware to be distributed, and distribution throws up its own problems such as identifying messaging channels which could then be subscribed to. In particular, interested parties need to identify channels defined in remote locations while not knowing details of how they are defined. A common vocabulary using semantic descriptions offers a solution to this problem. In this paper, we describe the design and implementation of federated messaging middleware using semantic description of channels.

Chained Negotiation for Distributed Notification Services

Distributed notification services allow consumers and publishers of notifications to interact with different notification services. However, such a distributed infrastructure makes it difficult to share notifications between consumers when consumers are allowed to specify Quality of Service levels. In this paper, we present a chained negotiation engine, enabling distributed notification services to support negotiation and to reuse existing subscriptions. We demonstrate the benefit to the system as a whole by reducing load on service providers and enabling content to be shared.