Support for User-Defined Metrics in the Online Performance Analysis Tool G-PM

Procedia Computer Science

Janczykowski

Jopek

et al. 2013

In this paper we present an evolved and extended approach to the monitoring of data flow and resources' usage in multiscale applications built with the MUSCLE environment. Most multi-scale platforms provide the ability of running complex computations, but do not sufficiently support monitoring features if any. Combining the monitored multi-scale application with components that support gathering low-level monitoring data into high abstraction level performance information and visualizing it at runtime is the main objective of our approach. Performance monitoring data should be presented to allow the user to easily observe and interpret computation progress. Data access and processing based on ontologies enables easy reconfiguration of monitored resources and application parameters. Moreover, ontologies facilitate automatic reasoning on performance flaws.

Section: Related Workmentioning

confidence: 99%

An Ontology-based Approach to Performance Monitoring of MUSCLE-bound Multi-scale Applications

Procedia Computer Science

Janczykowski

Jopek

et al. 2013

“…In the general case, the common framework is executing a bag-of-tasks application in the master-slave topology [32], while the scalability of hierarchical platforms is still a subject of study [25,26,18,35]. BoTs scheduling can successfully be performed using agents running in Cloud resources [17,21,22].…”

Section: Research Backgroundmentioning

confidence: 99%

Agent-based hierarchical approach for executing bag-of-tasks in clouds

Mazurek

Kruczkowski

2014

csci

Unlike message-passing applications, "bag-of-tasks" applications (BoTs), whose tasks are unrelated and independent (no inter-task communication), can be highly parallelized and executed in any acceptable order. A common practice when executing bag-of-tasks applications (BoT) is to exploit the master-slave topology. Cloud environments offer some features that facilitate the execution of BoT applications. One of the approaches to control-cloud resources is to use agents that are flexible actors in a dynamic environment. Given these assumptions, we have designed a combination of approaches which can be classified as distributed, hierarchical solutions to the issue of scalable execution of bagof-tasks. The concept of our system relates to a project that is focused on processing huge quantities of incoming data from a network of sensors through the Internet. Our aim is to create a mechanism for processing such data as a system that executes jobs while exploiting load balancing for cloud resources which use applications such as Eucalyptus. The idea is to create a hybrid architecture which takes advantage of some centralized parts of the system and full distributedness in other parts. On the other hand, we balance dependencies between system components using a hierarchical master-slave structure.

“…In addition to the data oriented monitoring tools [17] we analyzed which of AI techniques like fuzzy logic, neural network or rules may be applicable to the data management issues. At first fuzzy logic [8] as lower-level knowledge was considered.…”

Section: Related Workmentioning

confidence: 99%

“…Due to that it is important to develop a system which will allow us to deal with this task. In order to enable effective decision-making it is often necessary to get custom metrics which characterise the behavior of the objects under monitoring, like those described in [7,17].…”

Section: Concept Of the Presented Solutionmentioning

confidence: 99%

Data Storage Management Using Ai Methods

Szura

2013

csci

Data management and monitoring is an important issue in scientific computation. Scientists want to access their data as quickly as possible. Some experiments need to store a lot of data which have to be secure. By saying this we mean that this data can not disappear or be damaged also the data storage should be as cheap as possible. In this paper we present an approach to the automation of monitoring and management of data storage. We introduce a knowledge based system which is able to manage data, i.e., make decisions on migrating data, replicating or removing it. We discuss some of the existing solutions which are popular on the market. In this paper we aim to present our system which uses such AI techniques like fuzzy logic or a rule-based expert system to deal with data storage management. We exploit in this system a cost model to analyze the proposed solutions. The operations performed by our system are aimed to optimize the usage of the monitored infrastructure.