Soodeh Farokhi, Pooyan Jamshidi, Dražen Lučanin, I. Brandić
Cloud computing offers the elasticity features by dynamically resizing the infrastructure in response to changes in workload demands to meet performance guarantees and minimize costs. In the last decade, a large body of work has been done in the area of horizontal elasticity, while only few research efforts addressed vertical elasticity. This paper develops a vertical elasticity controller for cloud-based applications using control theory principles to guarantee performance requirements by adjusting the memory allocation as a control knob. The novelty of our work lies on applying a controller synthesis technique by guaranteeing robustness and stability of the controlled system, using the application response time as a decision making criterion. The experimental results reveal that the controller is able to efficiently save at least 47% memory usage while keeping an acceptable user experience.
{"title":"Performance-Based Vertical Memory Elasticity","authors":"Soodeh Farokhi, Pooyan Jamshidi, Dražen Lučanin, I. Brandić","doi":"10.1109/ICAC.2015.51","DOIUrl":"https://doi.org/10.1109/ICAC.2015.51","url":null,"abstract":"Cloud computing offers the elasticity features by dynamically resizing the infrastructure in response to changes in workload demands to meet performance guarantees and minimize costs. In the last decade, a large body of work has been done in the area of horizontal elasticity, while only few research efforts addressed vertical elasticity. This paper develops a vertical elasticity controller for cloud-based applications using control theory principles to guarantee performance requirements by adjusting the memory allocation as a control knob. The novelty of our work lies on applying a controller synthesis technique by guaranteeing robustness and stability of the controlled system, using the application response time as a decision making criterion. The experimental results reveal that the controller is able to efficiently save at least 47% memory usage while keeping an acceptable user experience.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"31 1","pages":"151-152"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83096329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Virtualized systems such as public and private clouds are emerging as important new computing platforms with great potential to conveniently deliver computing across the Internet and efficiently utilize resources consolidated via virtualization. Resource management in virtualized systems remains a key challenge because of their intrinsically dynamic and complex nature, where the applications have dynamically changing workloads and virtual machines (VMs) compete for the shared resources in a convolved manner. To address this challenge, this paper proposes a new resource management approach that can effectively capture the nonlinear behaviors in VM resource usages through fuzzy modeling and quickly adapt to the changes in the system through predictive control. The resulting fuzzy-model-predictive-control (FMPC) approach is capable of optimizing the VM resource allocations to applications according to their QoS targets. This approach is incorporated in a two-level cloud resource management framework where at the VM host level the node controllers employ FMPC to optimize dynamic VM resource allocations within individual hosts, and at the cloud zone level the global scheduler coordinates the node controllers to optimize resource utilization across hosts through dynamic VM migrations. The proposed approaches were implemented for Xen-based virtualized systems and evaluated using typical benchmarks (RUBiS, Free Bench) on a test bed with over 100 concurrent VMs. The results demonstrate that FMPC can accurately model the resource demands for dynamic applications and optimize the resource allocations to VMs with complex contentions. It substantially outperforms the traditional linear modeling based predictive control approach. The two-level resource management can make effective use of VM migrations to further improve performance across hosts as the host-level loads vary over time.
{"title":"QoS-Driven Cloud Resource Management through Fuzzy Model Predictive Control","authors":"Lixi Wang, Jing Xu, H. Duran-Limon, Ming Zhao","doi":"10.1109/ICAC.2015.41","DOIUrl":"https://doi.org/10.1109/ICAC.2015.41","url":null,"abstract":"Virtualized systems such as public and private clouds are emerging as important new computing platforms with great potential to conveniently deliver computing across the Internet and efficiently utilize resources consolidated via virtualization. Resource management in virtualized systems remains a key challenge because of their intrinsically dynamic and complex nature, where the applications have dynamically changing workloads and virtual machines (VMs) compete for the shared resources in a convolved manner. To address this challenge, this paper proposes a new resource management approach that can effectively capture the nonlinear behaviors in VM resource usages through fuzzy modeling and quickly adapt to the changes in the system through predictive control. The resulting fuzzy-model-predictive-control (FMPC) approach is capable of optimizing the VM resource allocations to applications according to their QoS targets. This approach is incorporated in a two-level cloud resource management framework where at the VM host level the node controllers employ FMPC to optimize dynamic VM resource allocations within individual hosts, and at the cloud zone level the global scheduler coordinates the node controllers to optimize resource utilization across hosts through dynamic VM migrations. The proposed approaches were implemented for Xen-based virtualized systems and evaluated using typical benchmarks (RUBiS, Free Bench) on a test bed with over 100 concurrent VMs. The results demonstrate that FMPC can accurately model the resource demands for dynamic applications and optimize the resource allocations to VMs with complex contentions. It substantially outperforms the traditional linear modeling based predictive control approach. The two-level resource management can make effective use of VM migrations to further improve performance across hosts as the host-level loads vary over time.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"13 1","pages":"81-90"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88334951","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
There are several research projects ongoing to apply cloud computing to industrial systems. The main focus of them is real-time performance of virtual machines (VMs) since it is important to guarantee a time-critical feature of industrial systems. However, there is another important issue that how much computing resource (CPU, memory, etc.) should be allocated to each VM which runs processes of an industrial system. In this paper, we propose a resource management method which manages VM resources with millisecond precision. In the proposed method, resource usage is measured and predicted, considering several microseconds allocation delay of Xen. Our experimental results show that the proposed method can guarantee 99% operation timing with higher CPU utilization in comparison with conventional resource management methods.
{"title":"A Virtual Machine Resource Management Method with Millisecond Precision","authors":"Y. Kaneko, Toshio Ito, Tomonori Maegawa","doi":"10.1109/ICAC.2015.24","DOIUrl":"https://doi.org/10.1109/ICAC.2015.24","url":null,"abstract":"There are several research projects ongoing to apply cloud computing to industrial systems. The main focus of them is real-time performance of virtual machines (VMs) since it is important to guarantee a time-critical feature of industrial systems. However, there is another important issue that how much computing resource (CPU, memory, etc.) should be allocated to each VM which runs processes of an industrial system. In this paper, we propose a resource management method which manages VM resources with millisecond precision. In the proposed method, resource usage is measured and predicted, considering several microseconds allocation delay of Xen. Our experimental results show that the proposed method can guarantee 99% operation timing with higher CPU utilization in comparison with conventional resource management methods.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"71 1","pages":"223-226"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89764605","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
It has been argued that the value of communities is that they can solve certain types of collective action problems which are resistant to purely market-based or policy-based solutions. Such problems increasingly arise in a data-driven information economy, the (so-called) sharing economy, and in economies of scarcity, where the added-value of information, reciprocity or other pro-social behaviour is indeterminate, and/or the qualitative nature of traded services is subjective and cannot simply be measured by kilowatts, tons, etc. It is therefore predicted that self-organised community systems will be of increasing importance as a mechanism for solving collective action problems in the digital society. Using community energy systems as an exemplar, this paper investigates the inter-weaving of (holonic) structure and (algorithmic) governance which are required to deliver air, sustainable and successful community-based solutions.
{"title":"Structure and Governance of Communities for the Digital Society","authors":"J. Pitt, A. Diaconescu","doi":"10.1109/ICAC.2015.62","DOIUrl":"https://doi.org/10.1109/ICAC.2015.62","url":null,"abstract":"It has been argued that the value of communities is that they can solve certain types of collective action problems which are resistant to purely market-based or policy-based solutions. Such problems increasingly arise in a data-driven information economy, the (so-called) sharing economy, and in economies of scarcity, where the added-value of information, reciprocity or other pro-social behaviour is indeterminate, and/or the qualitative nature of traded services is subjective and cannot simply be measured by kilowatts, tons, etc. It is therefore predicted that self-organised community systems will be of increasing importance as a mechanism for solving collective action problems in the digital society. Using community energy systems as an exemplar, this paper investigates the inter-weaving of (holonic) structure and (algorithmic) governance which are required to deliver air, sustainable and successful community-based solutions.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"29 1","pages":"279-284"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79219918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We discuss the opportunities for autonomous systems to perform reflection on their planners by adapting the models used to build plans. We first describe model-based planning systems, a form of automated planning system driven by declarative models of the planning domain. These models include descriptions of the conditions and effects of actions on the state of the world. When planning the activities of cyber-physical systems, the command and data representation of the system must be formally abstracted to the actions and states described in the planning system model. When the execution of a plan either fails or produces unexpected outcomes, the execution trace can be abstracted and compared to the predicted state according to the planning model, producing a list of discrepancies, these discrepancies can then be used to fix the model. This provides part of a reflection capability, namely, a set of well-formed problems with the domain model, the abstractions, or both. The challenge lies in the rest of the reflection capability, namely, a set of techniques for changing the models or the abstractions. We discuss these challenges and describe some of the options for addressing them.
{"title":"Reflecting on Planning Models: A Challenge for Self-Modeling Systems","authors":"J. Frank","doi":"10.1109/ICAC.2015.72","DOIUrl":"https://doi.org/10.1109/ICAC.2015.72","url":null,"abstract":"We discuss the opportunities for autonomous systems to perform reflection on their planners by adapting the models used to build plans. We first describe model-based planning systems, a form of automated planning system driven by declarative models of the planning domain. These models include descriptions of the conditions and effects of actions on the state of the world. When planning the activities of cyber-physical systems, the command and data representation of the system must be formally abstracted to the actions and states described in the planning system model. When the execution of a plan either fails or produces unexpected outcomes, the execution trace can be abstracted and compared to the predicted state according to the planning model, producing a list of discrepancies, these discrepancies can then be used to fix the model. This provides part of a reflection capability, namely, a set of well-formed problems with the domain model, the abstractions, or both. The challenge lies in the rest of the reflection capability, namely, a set of techniques for changing the models or the abstractions. We discuss these challenges and describe some of the options for addressing them.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"13 1","pages":"255-260"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86764256","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Phyllis R. Nelson, C. Landauer, K. Bellman, Shotaro Goto, Jesse Taylor
In previous work we showed that reflection and Wrappings are useful tools for system integration. But System of Systems integration also needs to accommodate specific hardware challenges. We discuss simple examples from the operation of a single cyber-physical agent that accepts top-down commands, but uses the Wrappings architecture to self-organize its context specific implementation of them. Our experimental context is CARS (Computational Architectures for Reflective Systems) a test bed for exploring the behavior of distributed autonomous cyber-physical agents in a complex environment.
{"title":"System of Systems Integration Also Includes Hardware Integration: A Small Demonstration of Providing Some Reflection Processes for HW","authors":"Phyllis R. Nelson, C. Landauer, K. Bellman, Shotaro Goto, Jesse Taylor","doi":"10.1109/ICAC.2015.74","DOIUrl":"https://doi.org/10.1109/ICAC.2015.74","url":null,"abstract":"In previous work we showed that reflection and Wrappings are useful tools for system integration. But System of Systems integration also needs to accommodate specific hardware challenges. We discuss simple examples from the operation of a single cyber-physical agent that accepts top-down commands, but uses the Wrappings architecture to self-organize its context specific implementation of them. Our experimental context is CARS (Computational Architectures for Reflective Systems) a test bed for exploring the behavior of distributed autonomous cyber-physical agents in a complex environment.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"280 1","pages":"285-288"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90789166","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Sarah Edenhofer, Christopher Stifter, Uwe Jänen, Jan Kantert, Sven Tomforde, J. Hähner, C. Müller-Schloer
Self-integration in open, distributed technical systems needs a mechanism for establishing and evaluating trust relationships to work in a stable and efficient manner. Based on a case study concerned with a Trusted Desktop Grid, this paper investigates techniques to isolate malicious agents. Therefore, we introduce a novel distributed strategy to identify and accuse nonbenevolentagents. Since intentionally bad behaviour is comparatively easy to detect, we further present novel agent types that either exploit the system or behave inconsistently. Afterwards, we demonstrate the potential benefit of the strategy in terms of simulations of the Trusted Desktop Grid and show that the overall system performance can be improved significantly.
{"title":"An Accusation-Based Strategy to Handle Undesirable Behaviour in Multi-agent Systems","authors":"Sarah Edenhofer, Christopher Stifter, Uwe Jänen, Jan Kantert, Sven Tomforde, J. Hähner, C. Müller-Schloer","doi":"10.1109/ICAC.2015.69","DOIUrl":"https://doi.org/10.1109/ICAC.2015.69","url":null,"abstract":"Self-integration in open, distributed technical systems needs a mechanism for establishing and evaluating trust relationships to work in a stable and efficient manner. Based on a case study concerned with a Trusted Desktop Grid, this paper investigates techniques to isolate malicious agents. Therefore, we introduce a novel distributed strategy to identify and accuse nonbenevolentagents. Since intentionally bad behaviour is comparatively easy to detect, we further present novel agent types that either exploit the system or behave inconsistently. Afterwards, we demonstrate the potential benefit of the strategy in terms of simulations of the Trusted Desktop Grid and show that the overall system performance can be improved significantly.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"30 1","pages":"243-248"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83387327","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Bassem Debbabi, T. Calmant, Olivier Gattaz, Sandra Massonnat, Patrick Emin
While the functions of Business Process Management (BPM) tools are already studied and standardized, new challenges regarding the architecture of such type of tools are emerging including the need for more scalability to support increasing demands, and more resilience of the overall solution to detect and avoid third-party code problems, that can causes failure of all the system. In this paper we present the new architecture of Agilium BPM tool that benefits of self-managed distributed architecture provided by Cohorte framework. We discuss the issues of the old architecture and the benefits and results of the introduced autonomic architecture.
{"title":"Self-Managed Component-Based Software Architecture for Business Process Management","authors":"Bassem Debbabi, T. Calmant, Olivier Gattaz, Sandra Massonnat, Patrick Emin","doi":"10.1109/ICAC.2015.25","DOIUrl":"https://doi.org/10.1109/ICAC.2015.25","url":null,"abstract":"While the functions of Business Process Management (BPM) tools are already studied and standardized, new challenges regarding the architecture of such type of tools are emerging including the need for more scalability to support increasing demands, and more resilience of the overall solution to detect and avoid third-party code problems, that can causes failure of all the system. In this paper we present the new architecture of Agilium BPM tool that benefits of self-managed distributed architecture provided by Cohorte framework. We discuss the issues of the old architecture and the benefits and results of the introduced autonomic architecture.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"8 1","pages":"145-146"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78419021","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Zichen Xu, Nan Deng, Christopher Stewart, Xiaorui Wang
Internet services replicate data to geo-diverse sites around the world, often via consistent hashing. Collectively, these sites span multiple power authorities that independently control carbon emissions at each site. Serving data from a carbon-heavy site increases the service's carbon footprint, but it is hard to place data at sites that will have low emission rates without replicating to too many sites. We present CADRE, a carbon-aware data replication approach. CADRE forecasts emission rates at each site and replicates data to sites that combine together to yield low carbon footprints. It makes replication decisions online, i.e., When data is created, and thus avoids emissions caused by moving data frequently in response to changing emission rates. CADRE uses the multiple-choice secretary algorithm to replicate objects with large footprints to low emission sites. It models carbon footprints for each object using the footprint-replication curve, a graph that maps replication factors to expected carbon footprints. CADRE also achieves availability goals, respects storage capacity limits and balances data across sites. Compared to consistent hashing, our approach reduces carbon footprints by 70%. It also supports and enhances the state-of-the-art green load balancing, reducing the carbon footprint by an additional 21%.
{"title":"CADRE: Carbon-Aware Data Replication for Geo-Diverse Services","authors":"Zichen Xu, Nan Deng, Christopher Stewart, Xiaorui Wang","doi":"10.1109/ICAC.2015.15","DOIUrl":"https://doi.org/10.1109/ICAC.2015.15","url":null,"abstract":"Internet services replicate data to geo-diverse sites around the world, often via consistent hashing. Collectively, these sites span multiple power authorities that independently control carbon emissions at each site. Serving data from a carbon-heavy site increases the service's carbon footprint, but it is hard to place data at sites that will have low emission rates without replicating to too many sites. We present CADRE, a carbon-aware data replication approach. CADRE forecasts emission rates at each site and replicates data to sites that combine together to yield low carbon footprints. It makes replication decisions online, i.e., When data is created, and thus avoids emissions caused by moving data frequently in response to changing emission rates. CADRE uses the multiple-choice secretary algorithm to replicate objects with large footprints to low emission sites. It models carbon footprints for each object using the footprint-replication curve, a graph that maps replication factors to expected carbon footprints. CADRE also achieves availability goals, respects storage capacity limits and balances data across sites. Compared to consistent hashing, our approach reduces carbon footprints by 70%. It also supports and enhances the state-of-the-art green load balancing, reducing the carbon footprint by an additional 21%.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"13 1","pages":"177-186"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84964842","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper proposes a theoretical foundation for specifying and reasoning about adaptations based on our middleware system that introduces the relocation of software components to define functions between computers as a basic mechanism for adaptation on distributed systems. It provides a language for specifying adaptations policies. The language is useful to reason about adaptations and can be executed in the middleware system.
{"title":"Specifying Distributed Adaptation through Software Component Relocation","authors":"Jingtao Sun, I. Satoh","doi":"10.1109/ICAC.2015.65","DOIUrl":"https://doi.org/10.1109/ICAC.2015.65","url":null,"abstract":"This paper proposes a theoretical foundation for specifying and reasoning about adaptations based on our middleware system that introduces the relocation of software components to define functions between computers as a basic mechanism for adaptation on distributed systems. It provides a language for specifying adaptations policies. The language is useful to reason about adaptations and can be executed in the middleware system.","PeriodicalId":6643,"journal":{"name":"2015 IEEE International Conference on Autonomic Computing","volume":"88 1","pages":"337-342"},"PeriodicalIF":0.0,"publicationDate":"2015-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86851531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：