Security is an important issue in wireless sensor networks (WSNs), which are often deployed in hostile environments. The q-composite key predistribution scheme has been recognized as a suitable approach to secure WSNs. Although the q-composite scheme has received much attention in the literature, there is still a lack of rigorous analysis for secure WSNs operating under the q-composite scheme in consideration of the unreliability of links. One main difficulty lies in analyzing the network topology whose links are not independent. Wireless links can be unreliable in practice due to the presence of physical barriers between sensors or because of harsh environmental conditions severely impairing communications. In this paper, we resolve the difficult challenge and investigate k-connectivity in secure WSNs operating under the q-composite scheme with unreliable communication links modeled as independent on/off channels, where k-connectivity ensures connectivity despite the failure of any (k - 1) sensors or links, and connectivity means that any two sensors can find a path in between for secure communication. Specifically, we derive the asymptotically exact probability and a zero-one law for k-connectivity. We further use the theoretical results to provide design guidelines for secure WSNs. Experimental results also confirm the validity of our analytical findings.
{"title":"Secure Connectivity of Wireless Sensor Networks Under Key Predistribution with on/off Channels","authors":"Jun Zhao","doi":"10.1109/ICDCS.2017.186","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.186","url":null,"abstract":"Security is an important issue in wireless sensor networks (WSNs), which are often deployed in hostile environments. The q-composite key predistribution scheme has been recognized as a suitable approach to secure WSNs. Although the q-composite scheme has received much attention in the literature, there is still a lack of rigorous analysis for secure WSNs operating under the q-composite scheme in consideration of the unreliability of links. One main difficulty lies in analyzing the network topology whose links are not independent. Wireless links can be unreliable in practice due to the presence of physical barriers between sensors or because of harsh environmental conditions severely impairing communications. In this paper, we resolve the difficult challenge and investigate k-connectivity in secure WSNs operating under the q-composite scheme with unreliable communication links modeled as independent on/off channels, where k-connectivity ensures connectivity despite the failure of any (k - 1) sensors or links, and connectivity means that any two sensors can find a path in between for secure communication. Specifically, we derive the asymptotically exact probability and a zero-one law for k-connectivity. We further use the theoretical results to provide design guidelines for secure WSNs. Experimental results also confirm the validity of our analytical findings.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129483533","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}
In cognitive radio networks (CRNs), the established communication sessions between secondary users (SUs) may be affected or even get interrupted because the SUs need to relinquish the spectrum when the licensed users (PUs) appear and reclaim the spectrum/channel. On detecting PU activities, the SUs on the affected links either switch to another available idle spectrum using the same link or the SUs seek for an alternative path/link. In either approach, the ongoing session is destined to experience delay or even gets interrupted, which is intolerable to quality of service-sensitive applications such as multimedia streaming or audio/video conferencing. In this paper, we study the problem of establishing k-protected routes in CRNs. A k-protected route consists of a set of main links with preassigned backup spectrum and backup paths and is guaranteed to sustain from k PU appearances without being interrupted. For a CRN, we find a k-protected route for each session request and maximize the number of sessions that can be supported. We propose both centralized and distributed k-protected routing algorithms for this problem. Simulation results show that our k-protected routing protocol outperforms existing opportunistic spectrum switching approaches in terms of delay and interruption rate.
在认知无线电网络(cognitive radio network, crn)中，当授权用户(licensed user)出现时，辅助用户(secondary user)需要放弃频谱，收回频谱/信道，从而可能影响到辅助用户(secondary user)之间已经建立的通信会话，甚至中断会话。在检测到PU活动时，受影响链路上的单元或者使用相同的链路切换到另一个可用的空闲频谱，或者寻找替代路径/链路。在任何一种方法中，正在进行的会话都注定要经历延迟甚至中断，这对于服务质量敏感的应用程序(如多媒体流或音频/视频会议)来说是无法忍受的。本文研究了在crn中建立k保护路由的问题。k保护路由由一组具有预先分配的备份频谱和备份路径的主链路组成，并保证从k个PU出现时持续不中断。对于CRN，我们为每个会话请求找到一条k保护的路由，并最大限度地支持会话数。针对这一问题，我们提出了集中式和分布式k保护路由算法。仿真结果表明，我们的k保护路由协议在延迟和中断率方面优于现有的机会频谱交换方法。
{"title":"k-Protected Routing Protocol in Multi-hop Cognitive Radio Networks","authors":"Chin-Jung Liu, Li Xiao","doi":"10.1109/ICDCS.2017.266","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.266","url":null,"abstract":"In cognitive radio networks (CRNs), the established communication sessions between secondary users (SUs) may be affected or even get interrupted because the SUs need to relinquish the spectrum when the licensed users (PUs) appear and reclaim the spectrum/channel. On detecting PU activities, the SUs on the affected links either switch to another available idle spectrum using the same link or the SUs seek for an alternative path/link. In either approach, the ongoing session is destined to experience delay or even gets interrupted, which is intolerable to quality of service-sensitive applications such as multimedia streaming or audio/video conferencing. In this paper, we study the problem of establishing k-protected routes in CRNs. A k-protected route consists of a set of main links with preassigned backup spectrum and backup paths and is guaranteed to sustain from k PU appearances without being interrupted. For a CRN, we find a k-protected route for each session request and maximize the number of sessions that can be supported. We propose both centralized and distributed k-protected routing algorithms for this problem. Simulation results show that our k-protected routing protocol outperforms existing opportunistic spectrum switching approaches in terms of delay and interruption rate.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129581107","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}
Paul C. Castro, Vatche Isahagian, Vinod Muthusamy, Aleksander Slominski
.
．
{"title":"Serverless Programming (Function as a Service)","authors":"Paul C. Castro, Vatche Isahagian, Vinod Muthusamy, Aleksander Slominski","doi":"10.1109/ICDCS.2017.305","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.305","url":null,"abstract":".","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130122773","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}
Ehsan Aryafar, A. Keshavarz-Haddad, Carlee Joe-Wong, M. Chiang
We study the resource allocation problem in RAN-level integrated HetNets. This emerging HetNets paradigm allows for dynamic traffic splitting across radio access technologies for each client, and then for aggregating the traffic inside the network to improve the overall resource utilization. We focus on the max-min fair service rate allocation across the clients, and study the properties of the optimal solution. Based on the analysis, we design a low complexity distributed algorithm that tries to achieve max-min fairness. We also design a hybrid network architecture that leverages opportunistic centralized network supervision to augment the distributed solution. We analyze the performance of our proposed algorithms and prove their convergence. We also derive conditions under which the outcome is optimal. When the conditions are not satisfied, we provide constant upper and lower bounds on the optimality gap. Finally, we study the convergence time of our distributed solution and show that leveraging appropriate policies in its design significantly reduces the convergence time.
{"title":"Max-Min Fair Resource Allocation in HetNets: Distributed Algorithms and Hybrid Architecture","authors":"Ehsan Aryafar, A. Keshavarz-Haddad, Carlee Joe-Wong, M. Chiang","doi":"10.1109/ICDCS.2017.205","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.205","url":null,"abstract":"We study the resource allocation problem in RAN-level integrated HetNets. This emerging HetNets paradigm allows for dynamic traffic splitting across radio access technologies for each client, and then for aggregating the traffic inside the network to improve the overall resource utilization. We focus on the max-min fair service rate allocation across the clients, and study the properties of the optimal solution. Based on the analysis, we design a low complexity distributed algorithm that tries to achieve max-min fairness. We also design a hybrid network architecture that leverages opportunistic centralized network supervision to augment the distributed solution. We analyze the performance of our proposed algorithms and prove their convergence. We also derive conditions under which the outcome is optimal. When the conditions are not satisfied, we provide constant upper and lower bounds on the optimality gap. Finally, we study the convergence time of our distributed solution and show that leveraging appropriate policies in its design significantly reduces the convergence time.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130360135","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}
Traditional privacy-preserving data disclosure solutions have focused on protecting the privacy of individual's information with the assumption that all aggregate (statistical) information about individuals is safe for disclosure. Such schemes fail to support group privacy where aggregate information about a group of individuals may also be sensitive and users of the published data may have different levels of access privileges entitled to them. We propose the notion of Eg-Group Differential Privacy that protects sensitive information of groups of individuals at various defined privacy levels, enabling data users to obtain the level of access entitled to them. We present a preliminary evaluation of the proposed notion of group privacy through experiments on real association graph data that demonstrate the guarantees on group privacy on the disclosed data.
传统的保护隐私的数据披露解决方案侧重于保护个人信息的隐私，并假设有关个人的所有汇总(统计)信息都可以安全披露。此类方案无法支持群体隐私，因为关于一组个人的汇总信息也可能是敏感的，并且发布数据的用户可能具有不同级别的访问权限。我们提出了egg - group Differential Privacy的概念，该概念在不同定义的隐私级别上保护个人群体的敏感信息，使数据用户能够获得有权访问的级别。通过对真实关联图数据的实验，我们对所提出的群体隐私概念进行了初步评估，证明了对公开数据的群体隐私保证。
{"title":"Group Differential Privacy-Preserving Disclosure of Multi-level Association Graphs","authors":"Balaji Palanisamy, C. Li, P. Krishnamurthy","doi":"10.1109/ICDCS.2017.223","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.223","url":null,"abstract":"Traditional privacy-preserving data disclosure solutions have focused on protecting the privacy of individual's information with the assumption that all aggregate (statistical) information about individuals is safe for disclosure. Such schemes fail to support group privacy where aggregate information about a group of individuals may also be sensitive and users of the published data may have different levels of access privileges entitled to them. We propose the notion of Eg-Group Differential Privacy that protects sensitive information of groups of individuals at various defined privacy levels, enabling data users to obtain the level of access entitled to them. We present a preliminary evaluation of the proposed notion of group privacy through experiments on real association graph data that demonstrate the guarantees on group privacy on the disclosed data.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123397403","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}
Smartphone lock screens are implemented to reduce the risk of data loss or compromise given the fact that increasing amount of person data are accessible on smartphones nowadays. Unfortunately, many smartphone users abandon lock screens due to the inconvenience of unlocking their phones many times a day. With the wide adoption of wearables, token-based approaches have gained popularity in simplifying unlocking and retaining security at the same time. To this end, we propose to take advantage of the smartwatch for easy smartphone unlocking. In this paper, we have designed WearLock, a system that uses acoustic tones as tokens to automate the unlocking securely. We build a sub-channel selection and an adaptive modulation in the acoustic modem to maximize unlocking success rate against ambient noise only when those two devices are nearby. We leverage the motion sensor on the smartwatch to reduce the unlock frequency. We offload smartwatch tasks to the smartphone to speed up computation and save energy. We have implemented the WearLock prototype and conducted extensive evaluations. Results achieved a low average bit error rate (BER) as 8% in various experiments. Compared to traditional manual personal identification numbers (PINs) entry, WearLock achieves at least 18% unlock speedup without any manual effort.
{"title":"WearLock: Unlocking Your Phone via Acoustics Using Smartwatch","authors":"Shanhe Yi, Zhengrui Qin, Nancy Carter, Qun A. Li","doi":"10.1109/ICDCS.2017.183","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.183","url":null,"abstract":"Smartphone lock screens are implemented to reduce the risk of data loss or compromise given the fact that increasing amount of person data are accessible on smartphones nowadays. Unfortunately, many smartphone users abandon lock screens due to the inconvenience of unlocking their phones many times a day. With the wide adoption of wearables, token-based approaches have gained popularity in simplifying unlocking and retaining security at the same time. To this end, we propose to take advantage of the smartwatch for easy smartphone unlocking. In this paper, we have designed WearLock, a system that uses acoustic tones as tokens to automate the unlocking securely. We build a sub-channel selection and an adaptive modulation in the acoustic modem to maximize unlocking success rate against ambient noise only when those two devices are nearby. We leverage the motion sensor on the smartwatch to reduce the unlock frequency. We offload smartwatch tasks to the smartphone to speed up computation and save energy. We have implemented the WearLock prototype and conducted extensive evaluations. Results achieved a low average bit error rate (BER) as 8% in various experiments. Compared to traditional manual personal identification numbers (PINs) entry, WearLock achieves at least 18% unlock speedup without any manual effort.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"13 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120907972","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}
The loose coupling and the inherent scalability make publish/subscribe systems an ideal candidate for event-driven services for wireless networks using low power protocols such as IEEE 802.15.4. This work introduces a distributed algorithm to build and maintain a routing structure for such networks. The algorithm dynamically maintains a multicast tree for each node. While previous work focused on minimizing these trees we aim to keep the effort to maintain them in case of fluctuations of subscribers low. The multicast trees are implicitly defined by a novel structure called augmented virtual ring. The main contribution is a distributed algorithm to build and maintain this augmented virtual ring. Maintenance operations after sub-and unsubscriptions require message exchange in a limited region only. We compare the average lengths of the constructedforwarding paths with an almost ideal approach. As a resultof independent interest we present a distributed algorithm using messages of size O(logn) for constructing virtual rings of graphs that are on average shorter than rings based on depth first search.
{"title":"Scalable Routing for Topic-Based Publish/Subscribe Systems Under Fluctuations","authors":"V. Turau, Gerry Siegemund","doi":"10.1109/ICDCS.2017.27","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.27","url":null,"abstract":"The loose coupling and the inherent scalability make publish/subscribe systems an ideal candidate for event-driven services for wireless networks using low power protocols such as IEEE 802.15.4. This work introduces a distributed algorithm to build and maintain a routing structure for such networks. The algorithm dynamically maintains a multicast tree for each node. While previous work focused on minimizing these trees we aim to keep the effort to maintain them in case of fluctuations of subscribers low. The multicast trees are implicitly defined by a novel structure called augmented virtual ring. The main contribution is a distributed algorithm to build and maintain this augmented virtual ring. Maintenance operations after sub-and unsubscriptions require message exchange in a limited region only. We compare the average lengths of the constructedforwarding paths with an almost ideal approach. As a resultof independent interest we present a distributed algorithm using messages of size O(logn) for constructing virtual rings of graphs that are on average shorter than rings based on depth first search.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"210 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121533900","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}
In this paper, we consider the problem of multiparty deep learning (MDL), wherein autonomous data owners jointly train accurate deep neural network models without sharing their private data. We design, implement, and evaluate ∝MDL, a new MDL paradigm built upon three primitives: asynchronous optimization, lightweight homomorphic encryption, and threshold secret sharing. Compared with prior work, ∝MDL departs in significant ways: a) besides providing explicit privacy guarantee, it retains desirable model utility, which is paramount for accuracy-critical domains; b) it provides an intuitive handle for the operator to gracefully balance model utility and training efficiency; c) moreover, it supports delicate control over communication and computational costs by offering two variants, operating under loose and tight coordination respectively, thus optimizable for given system settings (e.g., limited versus sufficient network bandwidth). Through extensive empirical evaluation using benchmark datasets and deep learning architectures, we demonstrate the efficacy of ∝MDL.
{"title":"Private, Yet Practical, Multiparty Deep Learning","authors":"Xinyang Zhang, S. Ji, Hui Wang, Ting Wang","doi":"10.1109/ICDCS.2017.215","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.215","url":null,"abstract":"In this paper, we consider the problem of multiparty deep learning (MDL), wherein autonomous data owners jointly train accurate deep neural network models without sharing their private data. We design, implement, and evaluate ∝MDL, a new MDL paradigm built upon three primitives: asynchronous optimization, lightweight homomorphic encryption, and threshold secret sharing. Compared with prior work, ∝MDL departs in significant ways: a) besides providing explicit privacy guarantee, it retains desirable model utility, which is paramount for accuracy-critical domains; b) it provides an intuitive handle for the operator to gracefully balance model utility and training efficiency; c) moreover, it supports delicate control over communication and computational costs by offering two variants, operating under loose and tight coordination respectively, thus optimizable for given system settings (e.g., limited versus sufficient network bandwidth). Through extensive empirical evaluation using benchmark datasets and deep learning architectures, we demonstrate the efficacy of ∝MDL.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121554735","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}
Di Wu, Dmitri I. Arkhipov, Thomas Przepiorka, Qiang Liu, J. Mccann, A. Regan
Accessing online social media content on underground metro systems is a challenge due to the fact that passengers often lose connectivity for large parts of their commute. As the oldest metro system in the world, the London underground represents a typical transportation network with intermittent Internet connectivity. To deal with disruption in connectivity along the sub-surface and deep-level underground lines on the London underground, we have designed a context-aware mobile system called DeepOpp that enables efficient offline access to online social media by prefetching and caching content opportunistically when signal availability is detected. DeepOpp can measure, crowdsource and predict signal characteristics such as strength, bandwidth and latency; it can use these predictions of mobile network signal to activate prefetching, and then employ an optimization routine to determine which social content should be cached in the system given real-time network conditions and device capacities. DeepOpp has been implemented as an Android application and tested on the London Underground; it shows significant improvement over existing approaches, e.g. reducing the amount of power needed to prefetch social media items by 2.5 times. While we use the London Underground to test our system, it is equally applicable in New York, Paris, Madrid, Shanghai, or any other urban underground metro system, or indeed in any situation in which users experience long breaks in connectivity.
{"title":"DeepOpp: Context-Aware Mobile Access to Social Media Content on Underground Metro Systems","authors":"Di Wu, Dmitri I. Arkhipov, Thomas Przepiorka, Qiang Liu, J. Mccann, A. Regan","doi":"10.1109/ICDCS.2017.269","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.269","url":null,"abstract":"Accessing online social media content on underground metro systems is a challenge due to the fact that passengers often lose connectivity for large parts of their commute. As the oldest metro system in the world, the London underground represents a typical transportation network with intermittent Internet connectivity. To deal with disruption in connectivity along the sub-surface and deep-level underground lines on the London underground, we have designed a context-aware mobile system called DeepOpp that enables efficient offline access to online social media by prefetching and caching content opportunistically when signal availability is detected. DeepOpp can measure, crowdsource and predict signal characteristics such as strength, bandwidth and latency; it can use these predictions of mobile network signal to activate prefetching, and then employ an optimization routine to determine which social content should be cached in the system given real-time network conditions and device capacities. DeepOpp has been implemented as an Android application and tested on the London Underground; it shows significant improvement over existing approaches, e.g. reducing the amount of power needed to prefetch social media items by 2.5 times. While we use the London Underground to test our system, it is equally applicable in New York, Paris, Madrid, Shanghai, or any other urban underground metro system, or indeed in any situation in which users experience long breaks in connectivity.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129400712","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}
Scaling web applications such as e-commerce in cloud by adding or removing servers in the system is an important practice to handle workload variations, with the goal of achieving both high quality of service (QoS) and high resource efficiency. Through extensive scaling experiments of an n-tier application benchmark (RUBBoS), we have observed that scaling only hardware resources without appropriate adaptation of soft resource allocations (e.g., thread or connection pool size) of each server would cause significant performance degradation of the overall system by either under- or over-utilizing the bottleneck resource in the system. We develop a dynamic concurrency management (DCM) framework which integrates soft resource allocations into the system scaling management. DCM introduces a model which determines a near-optimal concurrency setting to each tier of the system based on a combination of operational queuing laws and online analysis of fine-grained measurement data. We implement DCM as a two-level actuator which scales both hardware and soft resources in an n-tier system on the fly without interrupting the runtime system performance. Our experimental results demonstrate that DCM can achieve significantly more stable performance and higher resource efficiency compared to the state-of-the-art hardware-only scaling solutions (e.g., Amazon EC2-AutoScale) under realistic bursty workload traces.
{"title":"DCM: Dynamic Concurrency Management for Scaling n-Tier Applications in Cloud","authors":"Hui Chen, Qingyang Wang, Balaji Palanisamy, Pengcheng Xiong","doi":"10.1109/ICDCS.2017.22","DOIUrl":"https://doi.org/10.1109/ICDCS.2017.22","url":null,"abstract":"Scaling web applications such as e-commerce in cloud by adding or removing servers in the system is an important practice to handle workload variations, with the goal of achieving both high quality of service (QoS) and high resource efficiency. Through extensive scaling experiments of an n-tier application benchmark (RUBBoS), we have observed that scaling only hardware resources without appropriate adaptation of soft resource allocations (e.g., thread or connection pool size) of each server would cause significant performance degradation of the overall system by either under- or over-utilizing the bottleneck resource in the system. We develop a dynamic concurrency management (DCM) framework which integrates soft resource allocations into the system scaling management. DCM introduces a model which determines a near-optimal concurrency setting to each tier of the system based on a combination of operational queuing laws and online analysis of fine-grained measurement data. We implement DCM as a two-level actuator which scales both hardware and soft resources in an n-tier system on the fly without interrupting the runtime system performance. Our experimental results demonstrate that DCM can achieve significantly more stable performance and higher resource efficiency compared to the state-of-the-art hardware-only scaling solutions (e.g., Amazon EC2-AutoScale) under realistic bursty workload traces.","PeriodicalId":127689,"journal":{"name":"2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129407271","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：