H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. In Proceedings of the 2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....filtering with the datasets forming the view. Data Replication and Caching Service: In collaborative environments, data replication across storage systems and caching of data hot spots close to groups of clients can significantly increase throughput and decrease response times seen by clients [12 16]. Data replication and caching would alleviate the bottleneck that will be created by a single data source. Challenging issues include management of distributed memory and disk space for in core and persistent caching of data from multiple applications, and management of multiple replicas across ....

....efforts in this context. We have already developed implementations of some of the services presented in this paper using our frameworks. We have implemented prototypes of the proxy, query scheduling, and data replication and caching services using our multiple query optimization (MQO) framework [12 16]. MQO implements an active semantic cache, and enables multi threaded execution on a cluster of SMPs. The semantic caching allows user defined data structures for intermediate and final results from a query to be cached and reused by other queries. A prototype implementation of the distributed ....

Andrade, H., Kurc, T., Sussman, A., and Saltz, J. Scheduling Multiple Data Visualization Query Workloads on a Shared Memory Machine. 2002.

No context found.

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. In Proceedings of the 2002.

No context found.

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. In Proceedings of the 2002.

....to the server. Multiple query optimization techniques mainly rely on caching common subexpressions. Carefully scheduling queries also plays an important role, because the execution of queries can be ordered in a way to better exploit expressions that have been already cached. In an earlier work [2, 3], we investigated optimizations for multiple query workloads in an application for visualizing digitized microscopy images. Our results show that significant performance improvements can be achieved by data caching and carefully scheduling queries. In this section, we describe an approach for ....

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. Technical Report CS-TR-4290 and UMIACS-TR-2001.

....Therefore, we take a bottom up approach in which the high level operators responsible for the query processing chain are described in terms of low level primitives implemented by the application developer. The query processing system uses this information to infer points of reuse. In earlier work [5, 6, 7, 8, 9], we investigated frameworks, query scheduling, and cache replacement issues for data analysis applications. This paper differs from our previous work in that we look at the effect on data reuse of functional decomposition. There are a number of research projects that focus on component based ....

....is made as to which query from the waiting queue is selected for execution. The second step determines the data reuse for the query and the set of subqueries that should be executed to generate the portions of the query result that cannot be generated from cached aggregates. In earlier work [9], we developed methods for scheduling queries in the waiting queue for execution. We describe here the second step to determine reuse for functionally decomposed queries. For a given query, the query graph, G i (V; E) is traversed in a breadth first, top down fashion, starting from the sink ....

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. In Proceedings of the 2002.

....integrated into a much larger infrastructure, by being compliant to the models that are going to become protocols, best practices, and, eventually, standards. Finally, for a discussion of the bulk of our work on the multiple query optimization problem, we refer the reader to our previous papers [6, 7, 9], in which we extensively discuss related research and compare it to the approach we have employed. 3 Query Processing and Data Reuse in Data Analysis Applications Although many data analysis applications differ greatly in terms of their input datasets and resulting data products, processing of ....

....a 2 dimensional region in a slide, and the output is a potentially lower resolution image generated by applying a user defined aggregation operation on high resolution image chunks. We have implemented two functions to process high resolution input chunks to produce lower resolution images in VM [9]. Each function results in a different version of VM with very different computational requirements, but similar I O patterns. The first function employs a simple subsampling operation, and the second implements an averaging operation over a window. For a magnification level of N given in a query, ....

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. In Proceedings of the 2002.

....have to evolve in order to be integrated to a much larger infrastructure, by being compliant to the models that are eventually become standards, protocols and best practices. Finally, for the whole bulk of work in the multiple query optimization problem, we refer the reader to our previous works [5, 6, 7], in which we extensively discuss the related research and compare it to our own approach. 3 Query Processing and Data Reuse in Data Analysis Applications Although many data analysis applications seemingly differ greatly in terms of their input datasets and resulting data products, processing of ....

....a 2 dimensional region in a slide, and the output is a potentially lower resolution image generated by applying a user defined aggregation operation on high resolution image chunks. We have implemented two functions to process high resolution input chunks to produce lower resolution images in VM [7]. Each function results in a different version of VM with very different computational requirements, but similar I O patterns. The first function employs a simple subsampling operation, and the second implements an averaging operation over a window. For a magnification level of 3 given in a ....

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. In Proceedings of the 2002.

....We have previously developed an object oriented framework to support efficient utilization of common subexpressions and partial results [2] The underlying runtime system implements an in memory semantic cache to maintain user defined data structures for intermediate results. In earlier work [4] we addressed the query scheduling problem, and in this paper we evaluate cache replacement policies. We describe the implementation of two applications using the object oriented framework. These applications come from different domains and exhibit different data access and processing ....

....data reuse opportunities. Query Server: The query server interacts with clients for receiving queries and returning query results, and is implemented as a fixed size thread pool (typically the number of threads is set to the number of processors available on 4 an SMP node) A query scheduler [4] is employed to dynamically order client requests for assignment to available threads. A client request contains a query type id and userdefined parameters to a query object that the application developer implemented. The user defined parameters include a dataset id for the input dataset, query ....

[Article contains additional citation context not shown here]

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. In Proceedings of the

....and similar processing requirements (i.e. the same operations on data) Hence, several optimizations can be applied to improve system response time. These optimizations include reuse of intermediate and final results, data prefetching and caching, and scheduling to improve inter query locality [2, 3]. This paper investigates the use of SMP clusters to improve response times and overall system performance. In particular, we look at the effective use of aggregate processing power and I O bandwidth for executing single and multiple queries efficiently. Unlike previous work on query execution in ....

....visualization application, exploring multiple system configurations using a fixed amount of caching memory. 2 Runtime System Architecture We have deployed the runtime system within a middleware framework we have developed for evaluating multiple, simultaneous queries on a shared memory system [2, 3]. In the current implementation, each SMP node essentially runs a copy of the middleware with extensions to handle data exchange among SMP nodes. We briefly describe the middleware in this section. The extensions we have implemented for execution on a cluster of SMPs are presented in Section 3. ....

[Article contains additional citation context not shown here]

H. Andrade, T. Kurc, A. Sussman, and J. Saltz. Scheduling multiple data visualization query workloads on a shared memory machine. Technical Report CS-TR-4290 and UMIACS-TR-

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC