A dynamic hybrid scheduling algorithm for heterogeneous asymmetric environments†

The essence of efficient scheduling and data transmission techniques lies in providing the Web-applications with advanced data processing capabilities. In this paper we have efficiently combined the push and the pull scheduling to develop a new, practical, dynamic, hybrid scheduling strategy for heterogenous, asymmetric environments. The proposed algorithm dynamically computes the probabilities and the optimal cutoff-point to separate the push and the pull data sets. The data items are also assumed to be of variable lengths. While the push strategy uses the flat, round-robin scheduling, the pull items are determined by stretch-optimal (max-request min-service time) scheduling policy. In order to make the scheduling more practical, we have considered the impact of the impatience of the clients waiting to get the service of a particular data item. The effects of this impatience can lead to departure of specific client(s) from the system. Our proposed hybrid scheduling strategy takes care of these effects to capture a real portrayal of the system dynamics. These scenarios are modelled by suitable birth and death process to analyze the overall expected delay of the system. Subsequently, simulation results corroborate the average system performance and points out significant improvement over existing hybrid systems in terms of average waiting time spent by a client.

The ever-increasing popularity of Web services, growing demand for wireless multimedia and introduction of new, feature-enhanced, hand-held devices has already given birth to a new set of data-centric applications. Providing such applications with enhanced data processing capability calls for an efficient scheduling and transmission technique. The goal of most scheduling strategy lies in reducing the average waiting time. However, in most practical systems the variation of waiting time often results in client's impatience, thus provoking the clients to send repeated requests for the particular data item(s). In this paper we have developed a new hybrid scheduling framework for heterogeneous, asymmetric environments, by exploring the advantages of broadcasting very popular (push) data and dissemination of less popular (pull) data. The data access probabilities and the cut-off point used to segregate push and pull sets are dynamically computed. Packet fair scheduling (PFS) and stretch-optimal scheduling principle is deployed to obtain the push and pull schedule respectively. The framework explicitly takes care of the repeated requests originating from the impatient clients and minimizes the overall expected access time by obtaining an optimal cut-off point. Extensive performance analysis and simulation experiments are performed to show the efficiency of the system in reducing the overall expected access time (delay).

A strategy of distributed load balancing based on hybrid scheduling (LBHS) is proposed in this paper. This method is a hybrid scheduling strategy, which is based on the static allocation, and used dynamic scheduling system with the amply consideration of all the factors in the process of load balancing. This method can improve the utilization of the processor. The method is tested in the land surface disaster and ecology model. The result indicates that the strategy could improve the performance of parallel computing effectively, especially in the case that the size and data of the task is partially mastered beforehand.

We consider a communication problem, in which requests for data items arrive from time to time via a terrestrial network and at each broadcast tick the single server chooses a data item to broadcast via a satellite downlink to satisfy all the requests for the item. The goal of the server to minimize the total wait time of the requests. In this paper, we examine two well-known on-line algorithms used by the server to decide which data item to broadcast at each broadcast tick. In particular, we present complete competitive analysis of the algorithms, a research aspect of the problem which had not been studied before. Our results are consistent with observations obtained by simulation experiments reported in past literature. In addition, we prove a general lower bound to the competitive ratios of all online algorithms. The competitive ratios obtained for the two on-line algorithms in fact match our general lower bound, indicating the optimality of the algorithms.

In a cloud environment, a scheduler assigns arriving tasks to one of many servers, with the goal of minimizing response times. There are two conventional approaches to cloud scheduling. The first is called the Join-the-Shortest-Queue (JSQ) algorithm, which directs an arriving task to the least loaded server. Despite its excellent delay performance, JSQ is throughput-limited, and thus doesn't scale well with the number of servers. The second is called the Power-of-d-choices (Pod) algorithm, which selects d servers at random and routes a task to the least loaded server of the d servers. Despite its scalability, Pod suffers from long tail response times. In this paper, a hybrid scheduling strategy is proposed, and it consists of a Pod scheduler and a throughput-limited helper. Hybrid scheduling takes the best of both worlds, enjoying scalability and low tail response times. In particular, hybrid scheduling has bounded maximum queue size in the large-system regime, which is in sharp contrast to the Pod scheduling whose maximum queue size is unbounded.

In heterogeneous environments, dynamic scheduling algorithms are a powerful tool towards performance improvement of scientific applications via load balancing. However, these scheduling techniques employ heuristics that require prior knowledge about workload via profiling resulting in higher overhead as problem sizes and number of processors increase. In addition, load imbalance may appear only at run-time, making profiling work tedious and sometimes even obsolete. Recently, the integration of dynamic loop scheduling algorithms into a number of scientific applications has been proven effective. This paper reports on performance improvements obtained by integrating the Adaptive Weighted Factoring, a recently proposed dynamic loop scheduling technique that addresses these concerns, into two scientific applications: computational field simulation on unstructured grids, and N-Body simulations. Reported experimental results confirm the benefits of using this methodology, and emphasize its high potential for future integration into other scientific applications that exhibit substantial performance degradation due to load imbalance.

It is believed that broadcast is an efficient way to transmit data in an asymmetric communication system. Most of the previous work focused on either pull-based or push-based scheduling. However, for systems with a very large number of data items, none of these schemes is efficient individually. We propose a novel hybrid scheduling algorithm which uses both pull- and push-based schemes. In our approach, data items are divided, by a suitable cut-off point, into two disjoint sets: one consisting of more-popular items and the other of less-popular items. The items in the former set are broadcast by a push-based schedule, while those in the latter set by a pull-based schedule. A cut-off point is said to be optimal when it minimizes the expected access time in the system. The optimal cut-off point is derived analytically and validated by simulation experiments. Results show that the proposed hybrid push- and pull-based scheduling performs always better than a pure push-based scheduling and improves on the expected access time of an existing hybrid scheduling scheme.

On-demand broadcast is an effective wireless data dissemination technique to enhance system scalability and capability to handle dynamic user access patterns. With the rapid development of mobile applications, there is an increasing need for systems to support efficient processing of requests for multiple data items in multiple channels broadcast environments. Few studies, however, have considered the on-demand scheduling mechanisms for multi-item requests in multi-channel broadcast environments. In this paper, we investigate the scheduling problems arising in this new environment and observe that existing single-item requests based algorithms are unable to perform efficiently. Two potential problems are identified and examined. First, these algorithms take an excessively long time to serve the last few data items in a request because they disregard the relationship between data items and their parent requests. We claim that these algorithms suffer the request starvation problem in scheduling multi-item requests. Second, these algorithms cannot achieve the expected performance gain with multiple channels. We observed a broadcast mismatch problem in multi-channel broadcast environments. Thus, we propose an innovative algorithm to solve these two potential problems. The simulation results show that the performance of the proposed algorithm is superior to other classical algorithms under a variety of circumstances.

In mobile ad hoc networks, when a mobile host receives a data request from another host, the host usually transmits the requested data item by unicast. However, if mobile hosts hold data items that are frequently requested by others, they have to transmit the data items many times and consume a large amount of power. In this paper, we discuss effective data transmission for not only maintaining data availability but also prolonging the lifetime of mobile hosts. In our proposed method, each mobile host sends a data request attached with the deadline to receive the requested data item by the determined time. Moreover, each mobile host collects multiple requests for data items and transmits the requested data items by multicast. Therefore, our proposed method reduces data traffic.

Enhanced wireless communication improves the connectivity of vehicular networks in which vehicles are utilized as infrastructures for communication and computation. Thus, a new concept “Vehicular Edge Computing (VEC)” is formed. As VEC utilizes a collaborative multitude of near-user edge resources (i.e. vehicles) in the Internet of Vehicles, the capability of these joint resources becomes heterogeneous especially in their movements. Therefore, one critical problem is how to efficiently schedule each task under such mobile environments. For the reason, we propose a hybrid dynamic scheduling scheme (HDSS) that has the ability to optimize the task scheduling dynamically based on the changeable system environments. HDSS provides a decision function (DF) to select a better-performed scheduling algorithm from two provided candidates: the queue-based dynamic scheduling (QDS) algorithm and the time-based dynamic scheduling (TDS). QDS coincides with the Join-the-Shortest Queue scheme, which decides the scheduling by sorting out a server with the shortest queue-length; nevertheless, TDS is novel scheme that is designed to implement task allocation by estimating the waiting time of each server in order to select a server with the fastest response. Finally, this research generates formal models of each scheduling algorithm and the hybrid scheduling scheme to conduct performance evaluation with a fluid flow approximation technique. The analysis results in a superior performance of HDSS in the unstable VEC environments.

Caching systems using the Least Recently Used (LRU) principle have now become ubiquitous. A fundamental question for these systems is whether the cache space should be pooled together or divided to serve multiple flows of data item requests in order to minimize the miss probabilities. In this paper, we show that there is no straight yes or no answer to this question, depending on complex combinations of critical factors, including, e.g., request rates, overlapped data items across different request flows, data item popularities and their sizes. To this end, we characterize the performance of multiple flows of data item requests under resource pooling and separation for LRU caching when the cache size is large. Analytically, we show that it is asymptotically optimal to jointly serve multiple flows if their data item sizes and popularity distributions are similar and their arrival rates do not differ significantly; the self-organizing property of LRU caching automatically optimizes the resource allocation among them asymptotically. Otherwise, separating these flows could be better, e.g., when data sizes vary significantly. We also quantify critical points beyond which resource pooling is better than separation for each of the flows when the overlapped data items exceed certain levels. Technically, for a broad class of heavy-tailed distributions we derive the asymptotic miss probabilities of multiple flows of requests with varying data item sizes in a shared LRU cache space. It also validates the characteristic time approximation under certain conditions. These results provide new insights on improving the performance of caching systems.

It is useful to share location-dependent data items generated by vehicles such as a picture image of a traffic accident and the traffic condition around an intersection in vehicular ad hoc networks (VANETs) for improving the driver safety and comfort. In VANETs, due to the mobility of vehicles, it is difficult to reliably access data on other nodes. To improve the accessibility of data from a vehicle to a vehicle, it will be effective to distribute replicas of location-dependent data in a certain areas such as the area around the birthplace of the data item so that a request message geocasted to a location of interest can reache a vehicle that has the replica of the requested data item. In this paper, we propose an information-sharing scheme for location-dependent data items generated by vehicles using the push and pull approaches to balance the reliable message delivery and the traffic for the data dissemination called “assigning populated area as message storage area (APAM) scheme”. In the APAM scheme, request messages for location-dependent data are sent from each vehicle to locations along routes where there are many vehicles. Additionally, replicas of the data items are distributed around the birthplace and on the main streets where the vehicle density is high. These operations enable the request messages to encounter the vehicles that have the replica of the corresponding data item on the main streets. Simulation results showed that the proposed scheme improved the access success rate compared with the existing strategies that distribute replicas to vehicles near the birthplace of the original data items.

Today, more and more distributed computer applications are being modeled and constructed using real‐time principles and concepts. In 1989, the Object Management Group (OMG) formed a Real‐Time Special Interest Group (RT SIG) with the goal of extending the Common Object Request Broker Architecture (CORBA) standard to include real‐time specifications. This group's most recent efforts have focused on the requirements of dynamic distributed real‐time systems.One open problem in this area is resource access synchronization for tasks employing dynamic priority scheduling. This paper presents two resource synchronization protocols that the authors have developed which meet the requirements of dynamic distributed real‐time systems as specified by Dynamic Scheduling Real‐Time CORBA (DSRT CORBA). The proposed protocols can be applied to both Earliest Deadline First (EDF) and Least Laxity First (LLF) dynamic scheduling algorithms, allow distributed nested critical sections, and avoid unnecessary runtime overhead. In order to evaluate the performance of the proposed protocols, we analyzed each protocol's schedulability. Since the schedulability of the system is affected by numerous system configuration parameters, we have designed simulation experiments to isolate and illustrate the impact of each individual system parameter. Simulation experiments show the proposed protocols have better performance than one would realize by applying a schema that utilizes dynamic priority ceiling update. Copyright © 2004 John Wiley & Sons, Ltd.

Caching systems using the Least Recently Used (LRU) principle have now become ubiquitous. A fundamental question for these systems is whether the cache space should be pooled together or divided to serve multiple flows of data item requests in order to minimize the miss probabilities. In this paper, we show that there is no straight yes or no answer to this question, and depends on complex combinations of critical factors, including, e.g., request rates, overlapped data items across different request flows, data item popularities and their sizes. To this end, we characterize the performance of multiple flows of data item requests under resource pooling and separation when the cache size is large. Analytically we show that it is asymptotically optimal to jointly serve multiple flows if their data item sizes and popularity distributions are similar, and their arrival rates do not differ significantly; the self-organizing property of LRU caching automatically optimizes the resource allocation among them asymptotically. Otherwise, separating these flows could be better, e.g., when data sizes vary significantly. We also quantify critical points beyond which resource pooling is better than separation for each of the flows when the overlapped data items exceed certain levels. These results provide new insights on the performance of caching systems.

Caching systems using the Least Recently Used (LRU) principle have now become ubiquitous. A fundamental question for these systems is whether the cache space should be pooled together or divided to serve multiple flows of data item requests in order to minimize the miss probabilities. In this paper, we show that there is no straight yes or no answer to this question, and depends on complex combinations of critical factors, including, e.g., request rates, overlapped data items across different request flows, data item popularities and their sizes. To this end, we characterize the performance of multiple flows of data item requests under resource pooling and separation when the cache size is large. Analytically we show that it is asymptotically optimal to jointly serve multiple flows if their data item sizes and popularity distributions are similar, and their arrival rates do not differ significantly; the self-organizing property of LRU caching automatically optimizes the resource allocation among them asymptotically. Otherwise, separating these flows could be better, e.g., when data sizes vary significantly. We also quantify critical points beyond which resource pooling is better than separation for each of the flows when the overlapped data items exceed certain levels. These results provide new insights on the performance of caching systems.