Search

Striking a balance between improved cluster utilization and guaranteed application QoS is a long-standing research problem in cluster resource management. The majority of current solutions require a large number of sandboxed experimentation for different workload combinations and leverage them to predict possible interference for incoming workloads. This results in non-negligible time complexity that severely restricts its applicability to complex workload co-locations. The nature of pure offline profiling may also lead to model aging problem that drastically degrades the model precision. In this paper, we present Perph, a runtime agent on a per node basis, which decouples ML-based performance prediction and resource inference from centralized scheduler. We exploit the sensitivity of long-running applications to multi-resources for establishing a relationship between resource allocation and consequential performance. We use Online Gradient Boost Regression Tree (OGBRT) to enable the continuous model evolution. Once performance degradation is detected, resource inference is conducted to work out a proper slice of resources that will be reallocated to recover the target performance. The integration with Node Manager (NM) of Apache YARN shows that the throughput of Kafka data-streaming application is 2.0x and 1.82x times that of isolation execution schemes in native YARN and pure cgroup cpu subsystem. In TPC-C benchmarking, the throughput can also be improved by 35% and 23% respectively against YARN native and cgroup cpu subsystem.

Recently Cyber-Physical-Human Systems (CPHS) have been attracted much attention. Unlike existing computing paradigms such as Cloud Computing and Edge Computing, CPHS usually comprise many types of hybrid resources such as software and hardware in the cloud layer, network devices and edge servers in the network layer, as well as the physical devices and human participation. Such hybrid resource makes it challenging to efficient resource scheduling for CPHS. Previous studies on CPHS often focus on improving scheduling efficiency for specific application scenarios and thus are limited for other CPHS. In this work, we present a unified hybrid resource scheduling framework for CPHS, including a hybrid resource model for multiple types of resource providers and participants in CPHS, a hybrid resource orchestration tool that can describe our hybrid resource model as the resource requirements, and a scheduling mechanism to satisfy the resource requirements especially by considering the resource constraints and crosslayer scheduling in CPHS. We perform the case study of typical CPHS scenarios and show that our hybrid resource model and scheduling framework is effective, and provide a new view for improving the resource provision efficiency in CPHS.

Balancing resource utilization and application QoS is a long-standing research topic in cluster resource management. Big data YARN clusters need to co-schedule diverse workloads on shared resources including batch processing jobs, streaming jobs, and other long-running applications such as web services, database services, etc. Current resource managers are only responsible for resource allocation among applications/jobs but completely unaware of runtime QoS requirements of interactive and latency-sensitive applications. Prior works to maximize the QoS of monolithic applications ignore inherent dependencies and temporal-spatio performance variability of components, characteristics of distributed applications primarily driven by microservices. In this paper, we present Toposch, a new resource management system to adaptively co-locate batch tasks and microservices by harvesting runtime latency. In particular, Toposch tracks full footprints of every request across microservices over time. A latency graph is periodically generated for identifying victim microservices through an end-to-end latency critical path analysis. We then exploit per-microservice and per-node risk assessment to gauge the visible resources to the capacity scheduler in YARN. Execution of batch tasks are adaptively throttled or delayed, thereby avoiding latency increase due to node over-saturation. TOPOSCH is integrated with YARN and experiments show that the latency of DLRAs can be reduced by up to 39.8% against the default capacity scheduling in YARN.

Cluster administrators are facing great pressures to improve cluster utilization through workload co-location. Guaranteeing performance of long-running applications (LRAs), however, is far from settled as unpredictable interference across applications is catastrophic to QoS [2]. Current solutions such as [1] usually employ sandboxed and offline profiling for different workload combinations and leverage them to predict incoming interference. However, the time complexity restricts the applicability to complex co-locations. Hence, this issue entails a new framework to harness runtime performance and mitigate the time cost with machine intelligence: i) It is desirable to explore a quantitative relationship between allocated resource and consequent workload performance, not relying on analyzing interference derived from different workload combinations. The majority of works, however, depend on offline profiling and training which may lead to model aging problem. Moreover, multi-resource dimensions (e.g., LLC contention) that are not completely included by existing works but have impact on performance interference need to be considered [3]. ii) Workload co-location also necessitates fine-grained isolation and access control mechanism. Once performance degradation is detected, dynamic resource adjustment will be enforced and application will be assigned an access to specific slices of each resources. Inferring a "just enough" amount of resource adjustment ensures the application performance can be secured whilst improving cluster utilization. We present Perphon, a runtime agent on a per node basis, that decouples ML-based performance prediction and resource inference from centralized scheduler. Figure 1 outlines the proposed architecture. We initially exploit sensitivity of applications to multi-resources to establish performance prediction. To achieve this, Metric Monitor aggregates application fingerprint and system-level performance metrics including CPU, memory, Last Level Cache (LLC), memory bandwidth (MBW) and number of running threads, etc. They are enabled by Intel-RDT and precisely obtained from resource group manager. Perphon employs an Online Gradient Boost Regression Tree (OGBRT) approach to resolve model aging problem. Res-Perf Model warms up via offline learning that merely relies on a small volume of profiling in the early stage, but evolves with arrival of workloads. Consequently, parameters will be automatically updated and synchronized among agents. Anomaly Detector can timely pinpoint a performance degradation via LSTM time-series analysis and determine when and which application need to be re-allocated resources. Once abnormal performance counter or load is detected, Resource Inferer conducts a gradient ascend based inference to work out a proper slice of resources, towards dynamically recovering targeted performance. Upon receiving an updated re-allocation, Access Controller re-assigns a specific portion of the node resources to the affected application. Eventually, Isolation Executor enforces resource manipulation and ensures performance isolation across applications. Specifically, we use cgroup cpuset and memory subsystem to control usage of CPU and memory while leveraging Intel-RDT technology to underpin the manipulation of LLC and MBW. For fine-granularity management, we create different groups for LRA and batch jobs when the agent starts. Our prototype integration with Node Manager of Apache YARN shows that throughput of Kafka data-streaming application in Perphon is 2.0x and 1.82x times that of isolation execution schemes in native YARN and pure cgroup cpu subsystem.

Over a decade, cloud and subsequent joint cloud computing has been evolving into one of biggest disruptive technologies in modern digital age. The rapidly maturing cloud service and system management still heavily relies on virtualization which underpins Infrastructure as a Service (IaaS) to offer on-demand and low-cost computing services. Nevertheless datacenters still suffer from low utilization and resource imbalance. IaaS systems and their workloads, as legacy estates, are intricate to be migrated or re-planned, thereby increasing the complexity of utilization improvement. Arguably workload co-location of long-running applications encapsulated in virtual machines and latency-insensitive batch jobs is an alternative to improve overall resource utilization. However, guaranteeing the quality of long-running services is still challenging. In this context, we proposed an isolation-based cluster resource sharing system Shaready to enable workload co-residences. By means of global resource quota configuration and multi-resource isolation, long-running services in virtual machines can be prioritized with maximized resource provisioning. We implemented and validated it based on Openstack and Yarn clusters, and experiments demonstrate that system CPU and memory utilization can be improved by roughly 50% and 16.67% respectively on average with at most 7% performance degradation.

Container technology has been prevalent and widely-adopted in production environment considering the huge benefits to application packing, deploying and management. However, the deployment process is relatively slow by using conventional approaches. In large-scale concurrent deployments, resource contentions on the central image repository would aggravate such situation. In fact, it is observable that the image pulling operation is mainly responsible for the degraded performance. To this end, we propose Cider — a novel deployment system to enable rapid container deployment in a high concurrent and scalable manner at scale. Firstly, on-demand image data loading is proposed by altering the local Docker storage of worker nodes into all-nodes-sharing network storage. Also, the local copy-on-write layer for containers can ensure Cider to achieve the scalability whilst improving the cost-effectiveness during the holistic deployment. Experimental results reveal that Cider can shorten the overall deployment time by 85% and 62% on average when deploying one container and 100 concurrent containers respectively.

The Resource Description Framework (RDF) andSPARQL query language are gaining increasing popularity andacceptance. The ever-increasing RDF data has reached a billionscale of triples, resulting in the proliferation of distributed RDFstore systems within the Semantic Web community. However, theelasticity and performance issues are still far from settled inface of data volume explosion and workload spike. In addition, providers face great pressures to provision uninterrupted reliablestorage service whilst reducing the operational costs due to avariety of system failures. Therefore, how to efficiently realizesystem fault tolerance remains an intractable problem. In this paper, we introduce ScalaRDF, a distributed and elastic in-memoryRDF triple store to provision a fault-tolerant and scalable RDFstore and query mechanism. Specifically, we describe a consistenthashing protocol that optimizes the RDF data placement, dataoperations (especially for online RDF triple update operations)and achieves an autonomously elastic data re-distribution in theevent of cluster node joining or departing, avoiding the holisticoscillation of data storage. In addition, the data store is ableto realize a rapid and transparent failover through replicationmechanism which stores in-memory data replica in the next hashhop. The experiments demonstrate that query time and updatetime are reduced by 87% and 90% respectively compared to otherapproaches. For an 18G source dataset, the data redistributiontakes at most 60 seconds when system scales out and at most 100seconds for recovery when nodes undergo crash-stop failures.

Virtual machine live migration technology allows a running VM migrates from one physical host to another with no impact on users. As the scale of distributed computing gets larger and larger, hybrid cloud, which is integrated cloud service utilizing both private and public clouds to perform distinct functions within the same organization becomes a hotspot for both academia and industry. Thus, it will be vital to migrate virtual machines in hybrid clouds. The biggest issue which VM migrationinhybridcloudsfacestodayishowtotransferthelarge amount of storage data. Storage migration of virtual machine with existing mechanisms of memory migration has been studied for a long time. Unfortunately, no one can perform perfectly without modification. It is compelling to consider a newmigration model and optimization methods. In this paper, we analyze the advantage and disadvantage of classical methods, and put forward a method which combines the strengths of pre-copy and post-copy migration model and we also use disk working set to optimize virtual machines storage migration. We have developed and implemented our approach to the QEMU/KVM hypervisor and run a series of experiments. The presented technique shows a reduction of up 20.5% on average of the total transfer time for most of the selected scenario. Our approach can also reduce the migration downtime in most cases.

Live virtual machine migration is a technique often used to migrate an entire OS with running applications in a non-disruptive fashion. Prior works concerned with one-to-one live migration with many techniques have been proposed such as pre-copy, post-copy and log/replay. In contrast, we propose VMScatter, a one-to-many migration method to migrate virtual machines from one to many other hosts simultaneously. First, by merging the identical pages within or across virtual machines, VMScatter multicasts only a single copy of these pages to associated target hosts for avoiding redundant transmission. This is impactful practically when the same OS and similar applications running in the virtual machines where there are plenty of identical pages. Second, we introduce a novel grouping algorithm to decide the placement of virtual machines, distinguished from the previous schedule algorithms which focus on the workload for load balance or power saving, we also focus on network traffic, which is a critical metric in data-intensive data centers. Third, we schedule the multicast sequence of packets to reduce the network overhead introduced by joining or quitting the multicast groups of target hosts. Compared to traditional live migration technique in QEMU/KVM, VMScatter reduces 74.2% of the total transferred data, 69.1% of the total migration time and achieves the network traffic reduction from 50.1% to 70.3%.

Software as a Service (SaaS) is a software delivery and business model widely used by Cloud computing. Instead of purchasing and maintaining a software suite permanently, customers only need to lease the software on-demand. The domain of high assurance distributed systems has focused greatly on the areas of fault tolerance and dependability. In a multi-tenant context, it is particularly important to store, manage and provision data services to customers in a highly efficient and dependable manner due to a large number of file operations involved in running such services. It is also desirable to allow a user group to share and cooperate (e.g., co-edit) on some specific data. In this paper we present a dependable data provisioning service in a multitenant Cloud environment. We describe a metadata management approach and leverage multiple replicated metadata caching to shorten the file access time, with the improved efficiency of data sharing. In order to reduce frequent data transmission and data access latency, we introduce a distributed cooperative disk cache mechanism that supports effective cache placement and pull-push cache synchronization. In addition, we use efficient component failover to enhance the service dependability whilst avoiding negative impact from system failures. Our experimental results show that our system can significantly reduce both unused data transmission and response latency. Specifically, over 50% network transmission and operational latency can be saved for random reads while 28.24% network traffic and 25% response latency can be reduced for random write operations. We believe that these findings are demonstrating positive results along the right direction of resolving storage-related challenges in a multitenant Cloud environment.

As the sizes of IT infrastructure continue to grow, cloud computing is a natural extension of virtualisation technologies that enable scalable management of virtual machines over a plethora of physically connected systems. The so-called virtualisation-based cloud computing paradigm offers a practical approach to green IT/clouds, which emphasise the construction and deployment of scalable, energy-efficient network software applications (NetApp) by virtue of improved utilisation of the underlying resources. The latter is typically achieved through increased sharing of hardware and data in a multi-tenant cloud architecture/environment and, as such, accentuates the critical requirement for enhanced security services as an integrated component of the virtual infrastructure management strategy. This paper analyses the key security challenges faced by contemporary green cloud computing environments, and proposes a virtualisation security assurance architecture, CyberGuarder, which is designed to address several key security problems within the 'green' cloud computing context. In particular, CyberGuarder provides three different kinds of services; namely, a virtual machine security service, a virtual network security service and a policy based trust management service. Specifically, the proposed virtual machine security service incorporates a number of new techniques which include (1) a VMM-based integrity measurement approach for NetApp trusted loading, (2) a multi-granularity NetApp isolation mechanism to enable OS user isolation, and (3) a dynamic approach to virtual machine and network isolation for multiple NetApp's based on energy-efficiency and security requirements. Secondly, a virtual network security service has been developed successfully to provide an adaptive virtual security appliance deployment in a NetApp execution environment, whereby traditional security services such as IDS and firewalls can be encapsulated as VM images and deployed over a virtual security network in accordance with the practical configuration of the virtualised infrastructure. Thirdly, a security service providing policy based trust management is proposed to facilitate access control to the resources pool and a trust federation mechanism to support/optimise task privacy and cost requirements across multiple resource pools. Preliminary studies of these services have been carried out on our iVIC platform, with promising results. As part of our ongoing research in large-scale, energy-efficient/green cloud computing, we are currently developing a virtual laboratory for our campus courses using the virtualisation infrastructure of iVIC, which incorporates the important results and experience of CyberGuarder in a practical context.

The main goal of our key project, the CROWN Grid, is to empower in-depth integration of resources and the cooperation of researchers nationwide and worldwide. CROWN exploits a service-oriented architecture based on OGSA. In CROWN, remote service deployment is highly desirable. To the best of our knowledge, however, there is no successful solution to ensure the enabling remote and hot service deployment in grid systems. Traditionally, remote deployment is supported in a cold fashion, which results in many disadvantages, such as low efficiency. Moreover, since the deployer and the target container may be in different domains, great security challenges arise when a service is deployed to a remote container. In this paper, we present ROST, an original scheme of Remote and hOt Service deployment with Trustworthiness. By dynamically updating runtime environment configurations, ROST avoids restarting the runtime system during deployment. In addition, we include trust negotiation in ROST, which greatly increases the flexibility and security of the CROWN Grid. ROST has been successfully implemented. We conduct comprehensive experiments with real applications, and the results show that ROST is viable and significantly improves the service efficiency and quality of CROWN. We believe that the wide deployment of ROST would also benefit other grid systems.

Information service plays a key role in grid system, handles resource discovery and management process. Employing existing information service architectures suffers from poor scalability, long search response time, and large traffic overhead. In this paper, we propose a service club mechanism, called S-Club, for efficient service discovery. In S-Club, an overlay based on existing Grid Information Service (GIS) mesh network of CROWN is built, so that GISs are organized as service clubs. Each club serves for a certain type of service while each GIS may join one or more clubs. S-Club is adopted in our CROWN Grid and the performance of S-Club is evaluated by comprehensive simulations. The results show that S-Club scheme significantly improves search performance and outperforms existing approaches.

In this paper, for the first time, we tackle the scalability problem of Influence Maximization (IM) via distributed computing. First, we propose a distributed IM algorithm based on IRIE, one of the most state-of-the-art IM algorithms. Then an incremental updating method is proposed to reduce the overhead of repeated computation. Furthermore, based on some new insights, we redesign our algorithm with a strategy, which we call reservoir, to accumulate increments and delay exchange between machines. Experiments on real-world and synthetic networks show our redesigned algorithm, i.e. dIRIEr (distributed IRIE with Reservoir), reduces communication traffic dramatically and speeds up continuously as more machines are added in. dIRIEr can handle giant networks with hundreds of millions of nodes where centralized algorithms become infeasible.

Virtualization is one of the most fascinating techniques because it can facilitate the infrastructure management and provide isolated execution for running workloads. Despite the benefits gained from virtualization and resource sharing, improved resource utilization is still far from settled due to the dynamic resource requirements and the widely-used over-provision strategy for guaranteed QoS. Additionally, with the emerging demands for big data analytic, how to effectively manage hybrid workloads such as traditional batch task and long-running virtual machine (VM) service needs to be dealt with. In this paper, we propose a system to combine long-running VM service with typical batch workload like MapReduce. The objectives are to improve the holistic cluster utilization through dynamic resource adjustment mechanism for VM without violating other batch workload executions. Furthermore, VM migration is utilized to ensure high availability and avoid potential performance degradation. The experimental results reveal that the dynamically allocated memory is close to the real usage with only 10% estimation margin, and the performance impact on VM and MapReduce jobs are both within 1%. Additionally, at most 50% increment of resource utilization could be achieved. We believe that these findings are in the right direction to solving workload consolidation issues in hybrid computing environments.

MapReduce is a widely used programming model for large scale data processing. In order to estimate the performance of MapReduce job and analyze the bottleneck of MapReduce job, a practical performance model for MapReduce is needed. Many works have been done on modeling the performance of MapReduce jobs. However, existing performance models ignore some important factors, such as I/O congestions and task failures over cluster, which may significantly change the execution costs of MapReduce job. This paper, aiming at predicting the execution time of a MapReduce job, presents an enhanced performance model that takes the resource contention and task failures into consideration. In addition, the experimental results show that the model is more accurate than those without considering the contention and failure factors.

Mobile computing is increasingly growing in terms of massive computation as well as user demand and use of mobile devices. Remote execution techniques enrich the service experience of mobile devices by leveraging the resource pools of computation and storage capabilities of the cloud data center. However, the user experience and quality of service will be severely affected due to the inherent high latency and low bandwidth of a WAN environment. In this paper, we introduce a cloud base station "CloudAP" which is a small-scale computing infrastructure close to mobile users with local network access. In addition, we present a two-tier architecture consisting of CloudAP and Cloud center and show how to synthesize them to form a general computing environment. Furthermore, we propose a prompt execution environment migration scheme implemented by an efficient whole-system VM migration. It makes the execution environment move following the location of mobile device. Our experimental results demonstrate that the proposed architecture is effective and the execution environment migration scheme is efficient, consisting of up to 10ms and 30s for service downtime and execution environment switch time respectively. These improvements make vital contributions to user experience and QoS in mobile pervasive environment.

The combination of cloud computing and virtualization technology introduces a new pattern on resource allocation and utilization. Memory scanning deduplication techniques based on eliminating duplicated pages among virtual machines can promote the resource utilization, and decrease the total cost of ownership. However, the existing memory deduplication technologies lack the supporting of isolation and trustworthiness mechanism. This paper proposes a memory sharing mechanism based on user groups. This mechanism guarantees isolation between the different users on the same host. In addition, we designed a sampling hash algorithm to make the memory scanning process more efficient. We have implemented our approach in Linux by modifying the KSM scanning mechanism and splitting the global ksmd thread into per-group ksmds. The experiment results show the work can optimize the memory-intensive VMs, and efficiently accelerate the memory scanning process.

Newly emerging requirements of network application instant service bring in the challenges of a flexible, scalable, and reliable resource environment. Virtualization based resource isolation is helpful. However the internal structure of network software is often neglected in current virtualization based solutions. To improve resource sharing and utilization, the granularity of isolation should evolve from a single host to a network. NeTrOS, a virtual resource operating platform is proposed. Novel core concepts of cyber let and cyber-interrupt are introduced to express the first class runtime managed object and the evolution of virtual computing environment. Key technologies of virtual machine networks (VMNs), the implementation binding of cyber let in NeTrOS, are discussed. A programmable VMN architecture is proposed. Efficient maintenance mechanisms are involved to support the connectivity and dynamic nature of VMNs. Migrations of VMN in both LAN and WAN environment are supported. Detail implementation and deployment of NeTrOS are presented. A series of system evaluation is conducted to show the effectiveness and efficiency of NeTrOS. Experience of a mobile Internet application case supported by NeTrOS is also discussed.

Checkpoint-recovery based on system virtualization is an attractive approach for providing the transparent and economic fault tolerance service in virtualized environments. The previous approaches introduce either great performance degradation or complex implementation issues. In this work, we propose a whole system hot-mirroring platform, namely Radiata, to provide fault-tolerance for any type of service by encapsulating the service instance into a virtual machine, and hot-mirroring the state changes of the virtual machine via the continual state replication. Our approach exploits three key optimizations for further reduction of the performance overhead: the asynchronous state replication, the COW-based memory checkpoint and the dirty page prediction. Based on the KVM platform, we have implemented the prototype system. The comprehensive evaluations under a variety of workloads demonstrate that Radiata is able to effectively support rapid and transparent fail-over in case of unexpected hardware failure, and outperforms the existing mechanisms in terms of the performance degradation in failure-free condition.

Compared to the conventional physical cluster, the virtual cluster technology has features such as flexible configuration, easy management and high system security, which provides users with an easily customized computing environment. Virtual machine image files as the most important stored objects in the virtual computing environment include all the information about the whole life cycle of virtual machine. The storage of the virtual machine images is an important part of virtual computing. There are mainly two ways to store the images, which include local storage and networking storage. This paper proposes a mechanism to manage the distributed image files to support the virtual computing environment. It uses the computing nodes` local storage as a chunk server in a distributed file system, and regards the distribution of the image files as a primary factor to choose the computing node to deploy virtual machine and to implement a backup mechanism of distributed images based on historical records of image files. Functional test and performance test results show that it can effectively support the virtual computing environment and has great values of applications and research.

Data-Intensive Computing emerges as the fourth paradigm for modern scientific discoveries. MapReduce, a programming paradigm for large-scale data-parallel applications, is widely applied to web indexing, machine learning, and scientific simulations in industries as well as in academia. Recently, the virtualized "utility computing" environments, such as campus cloud, are becoming an important scenario to run MapReduce jobs. For a MapReduce job, the straggler tasks may dominate the response time and delay whole job. Various speculation schemes have been proposed to alleviate such problem, however, most of them implicitly assume that the time cost for data movement on launching speculative map tasks is trivial, which does not always hold for the virtualized Hadoop clusters in a campus cloud. In this paper, we propose a novel approach, CREST(Combination Re-Execution Scheduling Technology), which can achieve the optimal running time for speculative map tasks and decrease the response time of MapReduce jobs. The main idea is that re-executing a combination of tasks on a group of computing nodes may progress faster than directly speculating the straggler task on target node, due to data locality. The evaluation validates our approach and demonstrates that CREST can reduce the running time of a speculative map task by 70% with best cases and 50% on average, comparing with LATE.

Network Virtualization has recently emerged to provide scalable, customized and on-demand virtual network services over a shared substrate network. How to provide VN services with resiliency guarantees against network failures has become a critical issue, meanwhile the service resource usages should be minimized under the strict constraints such as link bandwidth capability and service resiliency guarantees etc. In this paper, we present a resource allocation algorithm to balance the tradeoff between service resource consumptions and service resiliency. By exploiting a heuristic VN mapping scheme and a restoration path selection scheme based on intelligent bandwidth sharing, the algorithm simultaneously makes cost-effective usage of network resources and protects VN services against network failures. We perform evaluations and find that the algorithm is near optimal in terms of network resource usage, especially the additional restoration bandwidth cost for resiliency protection.

Grid computing has emerged as a new paradigm of distributed computing technology on large-scale resource sharing and coordinated problem solving. Based on a proposed Web service-based grid architecture, we have designed a service grid middleware system called CROWN which aims to promote the utilization of valuable resources and cooperation of researchers nationwide and world-wide. To address the issues of CROWN resource management, we proposed some key technologies including trustworthy remote and hot service deployment, overlay-based distributed resource organization, resource scheduling and load balance, and federation-based virtual organization management. A status of the wide-area CROWN testbed is also introduced in this paper. Three typical applications including AREM, MDP and gViz are deployed on the CROWN testbed. Experience of CROWN testbed deployment and application development shows that the middleware can support the typical scenarios such as computing-intensive applications and data-intensive applications etc.

Although many efforts have been put on the load balance in network and job scheduling systems, most of them, however, can not be applied in the service grid environment directly since they are often designed for a homogeneous system with limited scalability. It is still a challenge problem to balance the load among service grid nodes which are often highly dynamic, heterogeneous and linked by wide-area network. In this paper, we present a load balance strategy using several bulletin-boards as load intermediates among grid nodes. A modified thresholds based load transfer algorithm has been applied with a non-preemptive selection policy. Based on the strategy above, a load balance system is realized in CROWN, a service oriented grid middleware, and deployed in the CROWN testbed. The performance evaluations have shown that our strategy can effectively balance the load of service invocation, and improve the system throughput.

Currently some complex grid applications developing often need orchestrate multiple diverse grid services into a workflow of tasks that can submit for executing on the grid environment. In this paper, we present CROWN FlowEngine--a GPEL-based grid workflow engine for executing grid workflow instances. Besides basic functions of a conventional BPEL4WS- based workflow engine, CROWN FlowEngine has many features including hierarchical processing mechanism, multiple types of task scheduling, transaction processing, etc, which are of paramount importance to supporting workflow instances using GPEL language. CROWN FlowEngine will be adopted and widely deployed in CROWN Grid environment to support a wide range of service grid applications integration. We conduct several experiments to evaluate the performance of CROWN FlowEngine, and the results of comparing our work with GWES are presented as well.

The purpose of virtual computing environment is to improve resource utilization by providing a unified integrated operating platform for users and applications based on aggregation of heterogeneous and autonomous resources. With the rapid development in recent years, hypervisor technologies have become mature and comprehensive with four features, including transparency, isolation, encapsulation and manageability. In this paper, a hypervisor based virtual computing infrastructure, named CIVIC, is proposed. Compared with existing approaches, CIVIC may benefit in several ways. It offers separated and isolated computing environment for end users, and realizes hardware and software consolidation and centralized management. Beside this, CIVIC provides a transparent view to upper layer applications, by hiding the dynamicity, distribution and heterogeneity of underlying resources. Performance of the infrastructure is evaluated by an initial deployment and experiment. The result shows that CIVIC can facilitate installation, configuration and deployment of network-oriented applications.

Providing guaranteed QoS for grid services using resource reservation and allocation is an important feature for today's service grid. Reservation requests with existing mechanisms are often rejected during the resource utilization peak and lead to resource capacity fragment problem. In this paper, we propose a flexible capacity reservation mechanism, called FIRST, which employs the slack time-enabled request admission control with differentiated selection strategies. We implement the prototype of FIRST in our CROWN node server, the service container of CROWN service grid middleware. The performance of the admission control algorithm for FIRST is evaluated by comprehensive simulations and implementations. Experimental results show that a better resource utilization ratio can be achieved by introducing the slack time into fixed reservation scenario, and a min-min based request selection strategy obtains a better performance compared with existing strategies.

Grid workflow is an important technique to implement cooperation of services in a grid. QoS allows for the selection and execution of Grid workflow to better fulfill customer expectations. Before the execution of a workflow, users need to provide the QoS requirements, and then the workflow service evaluates the requirements and returns the evaluation results. This paper proposes a QoS negotiation protocol to support such negotiation process and also introduces the architecture of supporting software system.

ChinaGrid Support Platform (CGSP) is proposed to provide grid toolkit for ChinaGrid application developers and specific grid constructors, in order to reduce their development cost as greatly as possible. CGSP extensible and reconfigurable framework, which satisfies the expansion and autonomy requirement of ChinaGrid, is mainly discussed in the paper. In the framework, domain is presented to denote one unit which could provide grid service for end users by itself. Layered structure of domains and corresponding interactive relationship are paid much more attention. CGSP design motivation and simple execution management mechanism are also described in this paper.

CROWN Grid aims to empower in-depth integration of resources and cooperation of researchers nationwide and worldwide. In such a distributed environment, to facilitate adoption of services, remote and hot service deployment is highly desirable. Furthermore, when the deployer and the target container are from different domains, great security challenges arise when a service is deployed to the remote container. In this paper, we present ROST, an original scheme of Remote & hOt Service deployment with Trustworthiness. By dynamically updating runtime environment configurations, ROST avoids restarting the runtime system during deployment. Moreover, we adopt trust negotiation in ROST to assure the security of service deployment. We conduct experiments in a real grid environment, and evaluate ROST comprehensively.

The grid information service (GIS), which is mainly used for resource discovery and monitoring, plays an important role in grid systems. Although many GIS systems have been proposed, e.g. MDS2, no appropriate solutions exist for the OGSA-based grid systems today. The architecture for OGSA-based GIS, i.e. FOSIS (forest structured information service), is proposed in this paper. FOSIS is featured by its forest structure, which is composed of multiple trees of GIS.

WS core is an open-source implementation of grid service container in Globus Toolkit 4 (GT 4). However, GT 4 WS core is far from a full-fledged grid service container. Basing on the Globus work, we develop CROWN Node Server, an enhanced grid service container. On the one hand, besides basic functions of a WSRF grid service container, CROWN Node Server encompasses features including remote and hot service deployment with trustworthiness, monitoring, logging and management, etc, which are important to build grid applications. On the other hand, we identify several security challenges faced by service grids, and propose CROWN-ST, which provides thorough security support for running grid services across domains. CROWN Node Server has been adopted and widely deployed in CROWN Grid to support a wide range of applications. We evaluate the performance of Node Server, and the comparing results with GT 4 WS core are presented as well.