19th IEEE International Parallel and Distributed Processing Symposium
We propose a cluster-based web server where a few computing nodes are separately reserved for hig... more We propose a cluster-based web server where a few computing nodes are separately reserved for high-performance computing applications, such as multimedia, SSL, and CGI. As an example application, we consider a multimedia server that dynamically generates video units to satisfy the bit rate and bandwidth requirements of a variety of clients. To perform QoS aware scheduling of multiple multimedia jobs on the computing servers, a two-step algorithm is proposed. The first step is to fairly schedule multimedia streams to satisfy each stream's QoS requirement; and the second step is to balance the workload among heterogeneous computing nodes in the cluster. We propose a new Quota-based Adaptive CoScheduling (QACS) algorithm that greatly reduces delay jitter by eliminating the out-of-order departure for outgoing streams, as well as achieves high throughput in a heterogeneous cluster. Experimental results show that the proposed scheduling technique gives adequate QoS guarantees to multiple streams.
The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003
Grid computing allows flexible resource sharing among geographically distributed computing resour... more Grid computing allows flexible resource sharing among geographically distributed computing resources in multiple administrative domains. Virtualization of resources allows jobs to be run on remote resources participating in a Grid. While this computing paradigm has been used primarily for batch jobs, we study interactive Grid applications rich in graphics and multimedia such as scientific visualization and digital content creation. A host of security issues need to be addressed for such Interactive Grids to gain acceptance, particularly in industry. The purpose of this paper is to scope these security issues. The Grid Security Infrastructure (GSI), a component of the Globus Toolkit [1], creates Grid credentials for every user and resource. We describe how this may be extended to securely set up an interactive session on a remote host, and the additional security issues associated with interactive session management. We propose controlled shell and controlled desktop mechanisms that restrict the user to execute only authorized commands and applications, and controlled user and super-user accounts that customize the shell and desktop using policy files. We also propose a new approach to scoping the security needs of Grid systems by defining three generic scenarios: mutual trust, partial trust and mutual distrust. New security issues arise when the user may not be trusted, or the user and the host computer's owner are mutually suspicious.
Proceedings International Parallel and Distributed Processing Symposium
We are developing mmGrid (Multimedia Grid) as an extensible middleware architecture supporting mu... more We are developing mmGrid (Multimedia Grid) as an extensible middleware architecture supporting multimedia applications in a grid computing environment. Our vision is to provide support for interactive applications from the following domains: graphics, visualization, streaming media and tele-immersion. The initial deployment will be within an enterprise as a mechanism for provisioning computing resources. However the scheduling system of mmGrid will be flexible, and will allow interactive and batch jobs to use the grid-computing paradigm. This paper presents our argument for using remote display technology in this environment. We also report on the use cases we support at this point, the system architecture for mmGrid and our research directions.
Proceedings International Parallel and Distributed Processing Symposium
Active routers allow computation to be performed within the network by processing packets when th... more Active routers allow computation to be performed within the network by processing packets when they pass through the routers. We design and implement a cluster-based active router system that provides multimedia stream transcoding service. The performance of the system is evaluated with three different load balancing schemes. We evaluate the out-of-order phenomenon and analyze the tradeoff between this phenomenon and the processing speed. We present a stream-based round robin algorithm for the transcoding service offered in the router and demonstrate its superiority over the conventional round-robin scheme. The main design criteria are to minimize the total transcoding time and maintain the order of media units for each outgoing stream.
Proceedings of the 2nd workshop on Middleware for grid computing -, 2004
Emerging large scale utility computing systems like Grids promise computing and storage to be pro... more Emerging large scale utility computing systems like Grids promise computing and storage to be provided to end users as a utility. System management services deployed in the middleware are a key to enabling this vision. Utility Grids provide a challenge in terms of scale, dynamism, and heterogeneity of resources and workloads. In this paper, we present a model based architecture for resource allocation services for Utility Grids. The proposed service is built in the context of interactive remote desktop session workloads and takes application performance QoS models into consideration. The key design guidelines are hierarchical request structure, application performance models, remote desktop session performance models, site admission control, multi-variable resource assignment system, and runtime session admission control. We have also built a simulation toolkit that can handle mixed batch and remote desktop session requests, and have implemented our proposed resource allocation service into the toolkit. We present some results from experiments done using the toolkit. Our proposed architecture for resource allocation services addresses the needs of emerging utility computing systems and captures the key concepts and guidelines for building such services in these environments.
Traditional use of grid computing allows a user to sub- mit batch jobs in a grid environment. We ... more Traditional use of grid computing allows a user to sub- mit batch jobs in a grid environment. We believe, next gen- eration grids will extend the application domain to include interactive graphical sessions. We term such grids interac- tive grids. In this paper, we describe some of the challenges involved in building interactive grids. These include fine grain access control, QoS guarantees, and dynamic account management. In order to architect interactive grids, we pro- pose and describe I-GENV, an environment for enabling in- teractive grids. I-GENV consists of GISH-'Grid Interactive Shell', Controlled Desktop, SAC-'Session Admission Con- trol' module, GMMA-'Grid Monitoring and Management Agents', System Policies, and Dynamic Account Manager. We also present our testbed implementation of I-GENV us- ing and extending Globus Toolkit 2.0 for the Grid middle- ware infrastructure, and VNC as the remote display tech- nology.
The letter has been motivated by the need to validate the sea surface salinity (SSS) derived by A... more The letter has been motivated by the need to validate the sea surface salinity (SSS) derived by Aquarius mission in the Indian Ocean. Validation has been performed on daily level-3 SSS data obtained by Aquarius in the Indian Ocean region during 2011-12. The in situ data sets available from Argo floats have been utilized for this purpose. The coefficient of determination between SSS and reference measurements is 0.84 and root mean square difference (RMSD) is about 0.45 psu. Spatial distributions of average salinity and variability of salinity have been found to agree reasonably well with the known features of SSS in this ocean basin. Under the assumption that the autocorrelation of SSS anomaly has anisotropic Gaussian characteristics, we also evaluated the space-time decorrelation scales of this anomaly, with the usual definition that decorrelation scales are the e-folding scales of SSS anomalies.
Concurrency and Computation: Practice and Experience, 2006
Emerging large scale utility computing systems such as Grids promise computing and storage to be ... more Emerging large scale utility computing systems such as Grids promise computing and storage to be provided to end users as a utility. System management services deployed in the middleware are a key to enabling this vision. Utility Grids provide a challenge in terms of scale, dynamism, and heterogeneity of resources and workloads. In this paper, we present a model based architecture for resource allocation services for Utility Grids. The proposed service is built in the context of interactive remote desktop session workloads and takes application performance QoS models into consideration. The key design guidelines are hierarchical request structure, application performance models, remote desktop session performance models, site admission control, multi-variable resource assignment system, and runtime session admission control. We have also built a simulation framework that can handle mixed batch and remote desktop session requests, and have implemented our proposed resource allocation service into the framework. We present some results from experiments done using the framework. Our proposed architecture for resource allocation services addresses the needs of emerging utility computing systems and captures the key concepts and guidelines for building such services in these environments.
Concurrency and Computation: Practice and Experience, 2004
Traditional use of Grid Computing Systems has been for batch jobs in the scientific and academic ... more Traditional use of Grid Computing Systems has been for batch jobs in the scientific and academic computing. We envision the next generation Grid computing systems to support graphical interactive sessions. In this paper, we propose a resource management framework for supporting graphical interactive sessions in a Grid computing system. We describe the high level architectural resource management framework distributed among the submission node, central scheduler node, and the execution node. We then describe in detail the resource management framework on the execution node. The description of the resource management framework on the scheduler node is kept at a high level in this paper. The framework on execution nodes consists of resource management agents, an admission control system and application predictor system. The agents on the execution node are startup agents, sensor agents, monitoring agents, aggregator agents, enforcement agents and registration agents. The session admission control system is responsible for determining if a new application session can be admitted to the execution node. An application predictor system is responsible for predicting the resource utilization behavior of applications based on data obtained from the resource management agents. The proposed framework allows for implementation of a scalable and extensible middleware for interactive grid resource management. It supports fine grained performance guarantees specified in service level agreements and brings forth some important and novel contributions to enable graphical interactive sessions on Grids.
Uploads
Papers by raj kumar