Mobile edge computing(MEC)provides services to devices and reduces latency in cellular internet of things(IoT)networks.However,the challenging problem is how to deploy MEC servers economically and efficiently.This pap...Mobile edge computing(MEC)provides services to devices and reduces latency in cellular internet of things(IoT)networks.However,the challenging problem is how to deploy MEC servers economically and efficiently.This paper investigates the deployment problem of MEC servers of the real-world road network by employing an improved genetic algorithm(GA)scheme.We first use the threshold-based K-means algorithm to form vehicle clusters according to their locations.We then select base stations(BSs)based on clustering center coordinates as the deployment locations set for potential MEC servers.We further select BSs using a combined simulated annealing(SA)algorithm and GA to minimize the deployment cost.The simulation results show that the improved GA deploys MEC servers effectively.In addition,the proposed algorithm outperforms GA and SA algorithms in terms of convergence speed and solution quality.展开更多
To determine CDN cache servers'placement reasonably,an idea that using graph partitioning to solve the problem was put forward through theoretical analysis and the specific algorithm of partitioning was researched...To determine CDN cache servers'placement reasonably,an idea that using graph partitioning to solve the problem was put forward through theoretical analysis and the specific algorithm of partitioning was researched. The concept of graph partitioning for CDN was defined. The conditions of graph partitioning for CDN were demonstrated: the sum of the weights of the nodes in each subarea is as close as possible; edge cut between the subareas is as large as possible; internal nodes in each subarea are connected as far as possible. By reference to light vertex matching algorithm of graph partitioning for network simulation,a multilevel k-way algorithm of graph partitioning for CDN was proposed. The maximized edge cut k-way KL refinement algorithm was discussed. Graph partitioning is a feasible way to solve the problem of CDN servers'placement. Multilevel k-way algorithm is a feasible algorithm for CDN graph partitioning.展开更多
The Web cluster has been a popular solution of network server system because of its scalability and cost effective ness. The cache configured in servers can result in increasing significantly performance, In this pape...The Web cluster has been a popular solution of network server system because of its scalability and cost effective ness. The cache configured in servers can result in increasing significantly performance, In this paper, we discuss the suitable configuration strategies for caching dynamic content by our experimental results. Considering the system itself can provide support for caching static Web page, such as computer memory cache and disk＇s own cache, we adopt a special pattern that only caches dynamic Web page in some experiments to enlarge cache space. The paper is introduced three different replacement algorithms in our cache proxy module to test the practical effects of caching dynamic pages under different conditions. The paper is chiefly analyzed the influences of generated time and accessed frequency on caching dynamic Web pages. The paper is also provided the detailed experiment results and main conclusions in the paper.展开更多
The work deals with the development of analytical model of multichannel technical queuing system with unreliable servers and input memory where server failure flows and incoming request flows comply with Poissonian la...The work deals with the development of analytical model of multichannel technical queuing system with unreliable servers and input memory where server failure flows and incoming request flows comply with Poissonian laws, while the flows of failed facilities repairs and flows of incoming requests comply with exponential laws of probability distribution. Random process of system change-over is a Markovian process with continuous time and discrete states. Relations binding basic parameters and output characteristics of the system indicated are obtained as probabilities of system staying in the given moment in one of the possible states. The proposed model is the most generalized compared to some models known in literature which could be considered as special cases of the considered model.展开更多
With results on the infinite servers queue systems with Poisson arrivals - M｜G｜∞ queues - busy period, it is displayed an application of those queue systems in the unemployment periods time length parameters and di...With results on the infinite servers queue systems with Poisson arrivals - M｜G｜∞ queues - busy period, it is displayed an application of those queue systems in the unemployment periods time length parameters and distribution function study. These queue systems are adequate to the study of many population processes, and this quality is brought in here. The results presented are mainly on unemployment periods length and their number in a certain time interval. Also, some questions regarding the practical applications of the outlined formulas are briefly discussed.展开更多
In this paper, we conduct research on the high robustness JavaEE enterprise development mode based on Hadoop and cloud servers. The current virtual machine real-time migration can only achieve manual migration, and ca...In this paper, we conduct research on the high robustness JavaEE enterprise development mode based on Hadoop and cloud servers. The current virtual machine real-time migration can only achieve manual migration, and cannot achieve full-automatic migration. In other words, when the server overload requires the administrator to artificially select a low-load host, and then hit migration command to implement the migration. In recent years, the Hadoop is becoming popular, and the read performance of the data is measured in terms of the time overhead for reading the required data. The key to reducing read time is to optimize that Hadoop cloud data read time and the RDBMS data query time. This paper integrates the mentioned techniques to construct the novel JavaEE enterprise development pattern that will promote the further development of the related techniques.展开更多
A prison can be something different as ordinary people imagine. It can be a therapeutic and rehabilitative environment, a place where human beings can learn to cope with the stressful situations they could not deal wi...A prison can be something different as ordinary people imagine. It can be a therapeutic and rehabilitative environment, a place where human beings can learn to cope with the stressful situations they could not deal with in their past. That is what happens in the Rencheng prison under the administration of Lineng Prison Sub-Bureau of East China’s Shandong Province.展开更多
In distributed machine learning(DML)based on the parameter server(PS)architecture,unbalanced communication load distribution of PSs will lead to a significant slowdown of model synchronization in heterogeneous network...In distributed machine learning(DML)based on the parameter server(PS)architecture,unbalanced communication load distribution of PSs will lead to a significant slowdown of model synchronization in heterogeneous networks due to low utilization of bandwidth.To address this problem,a network-aware adaptive PS load distribution scheme is proposed,which accelerates model synchronization by proactively adjusting the communication load on PSs according to network states.We evaluate the proposed scheme on MXNet,known as a realworld distributed training platform,and results show that our scheme achieves up to 2.68 times speed-up of model training in the dynamic and heterogeneous network environment.展开更多
The Internet of Vehicles(IoV)has evolved as an advancement over the conventional Vehicular Ad-hoc Networks(VANETs)in pursuing a more optimal intelligent transportation system that can provide various intelligent solut...The Internet of Vehicles(IoV)has evolved as an advancement over the conventional Vehicular Ad-hoc Networks(VANETs)in pursuing a more optimal intelligent transportation system that can provide various intelligent solutions and enable a variety of applications for vehicular traffic.Massive volumes of data are produced and communicated wirelessly among the different relayed entities in these vehicular networks,which might entice adversaries and endanger the system with a wide range of security attacks.To ensure the security of such a sensitive network,we proposed a distributed authentication mechanism for IoV based on blockchain technology as a distributed ledger with an ouroboros algorithm.Using timestamp and challenge-responsemechanisms,the proposed authentication model can withstand several security attacks such asMan-in-Middle(MiM)attacks,Distributed Denial of Service(DDoS)attacks,server spoofing attacks and more.The proposed method also provides a solution for single-point failure,forward secrecy,revocability,etc.We exhibit the security of our proposed model by using formal(mathematical)analysis and informal analysis.We used Random Oracle Model to perform themathematical analysis.In addition,we compared the communication cost,computation cost,and security of the proposed model with the related existing studies.We have verified the security of the model by using AVISPA tool simulation.The security analysis and computation analysis show that the proposed protocol is viable.展开更多
In cloud data centers,the consolidation of workload is one of the phases during which the available hosts are allocated tasks.This phenomenon ensures that the least possible number of hosts is used without compromise ...In cloud data centers,the consolidation of workload is one of the phases during which the available hosts are allocated tasks.This phenomenon ensures that the least possible number of hosts is used without compromise in meeting the Service Level Agreement(SLA).To consolidate the workloads,the hosts are segregated into three categories:normal hosts,under-loaded hosts,and over-loaded hosts based on their utilization.It is to be noted that the identification of an extensively used host or underloaded host is challenging to accomplish.Thresh-old values were proposed in the literature to detect this scenario.The current study aims to improve the existing methods that choose the underloaded hosts,get rid of Virtual Machines(VMs)from them,andfinally place them in some other hosts.The researcher proposes a Host Resource Utilization Aware(HRUAA)Algorithm to detect those underloaded and place its virtual machines on different hosts in a vibrant Cloud environment.The mechanism presented in this study is contrasted with existing mechanisms empirically.The results attained from the study estab-lish that numerous hosts can be shut down,while at the same time,the user's workload requirement can also be met.The proposed method is energy-efficient in workload consolidation,saves cost and time,and leverages active hosts.展开更多
There are different types of Cyber Security Attacks that are based on ICMP protocols. Many ICMP protocols are very similar, which may lead security managers to think they may have same impact on victim computer system...There are different types of Cyber Security Attacks that are based on ICMP protocols. Many ICMP protocols are very similar, which may lead security managers to think they may have same impact on victim computer systems or servers. In this paper, we investigate impact of different ICMP based security attacks on two popular server systems namely Microsoft’s Windows Server and Apple’s Mac Server OS running on same hardware platform, and compare their performance under different types of ICMP based security attacks.展开更多
This paper studies customer joining behavior and system regulation strategy in nonexhaustive visible M/M/m queues with synchronous vacations of a part of the servers.Once this part of the servers are idle,they take mu...This paper studies customer joining behavior and system regulation strategy in nonexhaustive visible M/M/m queues with synchronous vacations of a part of the servers.Once this part of the servers are idle,they take multiple vacations simultaneously(vacation period).Until there are customers waiting in the queue,they are reactivated and all servers are busy or idle(busy period).The authors call this part of the servers as“partial servers”.In view of the fully visible queue and the almost visible queue,the authors obtain customers’equilibrium joining threshold strategies and their socially optimal joining threshold strategies,respectively,and observe that customer joining behavior in equilibrium generally makes the system overcrowded,which makes the equilibrium social welfare lower than the optimal social welfare.After regulation,interestingly,for optimizing social welfare,the system manager hopes not only customers arriving in vacation period pay attention to the number of partial servers,but also customers arriving in busy period should care about it rather than ignore.Moreover,arranging more servers for vacation does not necessarily lead to the decrease of social welfare on condition that the number of partial servers is close to m.As for the information advantage of the fully visible case,it is not obvious for increasing social welfare and even unfavorable to servers’profit unless the number of partial servers is big enough.Furthermore,given the different composition of social welfare,there exists the optimal number of partial servers and the optimal arrival rate of customers for maximizing social welfare.展开更多
Rise in inlet air temperature increases the corresponding outlet air temperature from the server. As an added effect of rise in inlet air temperature, some active servers may start exhaling intensely hot air to form a...Rise in inlet air temperature increases the corresponding outlet air temperature from the server. As an added effect of rise in inlet air temperature, some active servers may start exhaling intensely hot air to form a hotspot. Increase in hot air tem- perature and occasional hotspots are an added burden on the cooling mechanism and result in energy wastage in data centers. The increase in inlet air temperature may also result in failure of server hardware. Identifying and comparing the thermal sensi- tivity to inlet air temperature for various servers helps in the thermal-aware arrangement and location switching of servers to minimize the cooling energy wastage. The peak outlet temperature among the relocated servers can be lowered and even be homogenized to reduce the cooling load and chances of hotspots. Based upon mutual comparison of inlet temperature sensitivity of heterogeneous servers, this paper presents a proactive approach for thermal-aware relocation of data center servers. The experimental results show that each relocation operation has a cooling energy saving of as much as 2.1 kW·h and lowers the chances of hotspots by over 77%. Thus, the thermal-aware relocation of servers helps in the establishment of green data centers.展开更多
Parameter server(PS)as the state-of-the-art distributed framework for large-scale iterative machine learning tasks has been extensively studied.However,existing PS-based systems often depend on memory implementations....Parameter server(PS)as the state-of-the-art distributed framework for large-scale iterative machine learning tasks has been extensively studied.However,existing PS-based systems often depend on memory implementations.With memory constraints,machine learning(ML)developers cannot train large-scale ML models in their rather small local clusters.Moreover,renting large-scale cloud servers is always economically infeasible for research teams and small companies.In this paper,we propose a disk-resident parameter server system named DRPS,which reduces the hardware requirement of large-scale machine learning tasks by storing high dimensional models on disk.To further improve the performance of DRPS,we build an efficient index structure for parameters to reduce the disk I/O cost.Based on this index structure,we propose a novel multi-objective partitioning algorithm for the parameters.Finally,a flexible workerselection parallel model of computation(WSP)is proposed to strike a right balance between the problem of inconsistent parameter versions(staleness)and that of inconsistent execution progresses(straggler).Extensive experiments on many typical machine learning applications with real and synthetic datasets validate the effectiveness of DRPS.展开更多
In this paper we consider a queueing network consisting of two parallel servers and threearrival streams generated by independent Poisson sources. Each server has its own queue and receivescustomers from its own arriv...In this paper we consider a queueing network consisting of two parallel servers and threearrival streams generated by independent Poisson sources. Each server has its own queue and receivescustomers from its own arrival stream. A third arrival stream consists of customers which place resourcedemands on both servers, which are handled separately by each server once the request is made. Eachservice time is independent and exponentially distributed. Each customer in the system pays a holdingcost per unit time. The objective is to dynamically determine the optimal scheduling policy to the thirdstream of conupled customers. based on the state of the system, so as to minimize the average cost. Thismodel is new, and has Policy implications for computer or communication networks. A fuzzy approachis presented to solve this problem. Simulation shows that the approach is efficient and promising.展开更多
The high-density server is featured as low power, low volume, and high computational density. With the rising use of high-density servers in data-intensive and large-scale web applications, it requires a high-performa...The high-density server is featured as low power, low volume, and high computational density. With the rising use of high-density servers in data-intensive and large-scale web applications, it requires a high-performance and cost-efficient intra-server interconnection network. Most of state-of-the-art high-density servers adopt the fully-connected intra-server network to attain high network performance. Unfortunately, this solution costs too much due to the high degree of nodes. In this paper, we exploit the theoretically optimized Moore graph to interconnect the chips within a server. Accounting for the suitable size of applications, a 50-size Moore graph, called Hoffman-Singleton graph, is adopted. In practice, multiple chips should be integrated onto one processor board, which means that the original graph should be partitioned into homogeneous connected subgraphs. However, the existing partition scheme does not consider above problem and thus generates heterogeneous subgraphs. To address this problem, we propose two equivalent-partition schemes for the Hoffman-Singleton graph. In addition, a logic-based and minimal routing mechanism, which is both time and area efficient, is proposed. Finally, we compare the proposed network architecture with its counterparts, namely the fully-connected, Kautz and Torus networks. The results show that our proposed network can achieve competitive performance as fully-connected network and cost close to Torus.展开更多
To reduce fetching cost from a remote source,it is natural to cache information near the users who may access the information later.However,with development of 5 G ultra-dense cellular networks andmobile edge computin...To reduce fetching cost from a remote source,it is natural to cache information near the users who may access the information later.However,with development of 5 G ultra-dense cellular networks andmobile edge computing(MEC),a reasonable selection among edge servers for content delivery becomes a problem when the mobile edge obtaining sufficient replica servers.In order to minimize the total cost accounting for both caching and fetching process,we study the optimal resource allocation for the content replica servers’ deployment.We decompose the total cost as the superposition of cost in several coverages.Particularly,we consider the criterion for determining the coverage of a replica server and formulate the coverage as a tradeoff between caching cost and fetching cost.According to the criterion,a coverage isolation(CI) algorithm is proposed to solve the deployment problem.The numerical results show that the proposed CI algorithm can reduce the cost and obtain a higher tolerance for different centrality indices.展开更多
When designing a multimedia server, several things must be decided: which scheduling scheme to adopt, how to allocate multimedia objects on storage devices, and the round length with which the streams will be serviced...When designing a multimedia server, several things must be decided: which scheduling scheme to adopt, how to allocate multimedia objects on storage devices, and the round length with which the streams will be serviced. Several problems in the designing of large-scale multimedia servers are addressed, with the following contributions: (1) a striping scheme is proposed that minimizes the number of seeks and hence maximizes the performance; (2) a simple and efficient mechanism is presented to find the optimal striping unit size as well as the optimal round length, which exploits both the characteristics of VBR streams and the situation of resources in the system; and (3) the characteristics and resource requirements of several scheduling schemes are investigated in order to obtain a clear indication as to which scheme shows the best performance in realtime multimedia servicing. Based on our analysis and experimental results, the CSCAN scheme outperforms the other schemes. It is believed that the results are of value in the design of effective large-scale multimedia servers. Keywords realtime multimedia - storage server - scheduling - data placement - buffer management - variable bit rate This work was supported in part by the University IT Research Center Project and Sunmoon University Research Project.Kyung-Oh Lee is an associate professor in the Faculty of Computer and Information Sciences, Sunmoon University, Korea. He received his B.S., M.S. and Ph.D. degrees in computer science from Seoul National University in 1989, 1994 and 1999, respectively. His current research interests include multimedia system, database, mobile communication. He is a member of KIPS (Korea Information Processing Society).Jungho-Ho Park is a professor in the Divisions of Computer and Information Sciences, Sunmoon University, Korea. He received his M.S. and Ph.D. degrees in computer science from Osaka University in 1987 and 1990, respectively. His current research interests include distributed algorithms, e-learning and electronic commerce. He is a director of KIPS (Korea Information Processing Society) and a vice president of KIPS-IT certification.Yoon-Young Park is an associate professor in the Faculty of Computer and Information Sciences, Sunmoon University, Korea. He received his M.S. and Ph.D. degrees in computer science from Seoul National University in 1985 and 1994, respectively. His current research interests include embedded systems and sensor networks. He is a member of KIPS (Korea Information Processing Society).展开更多
基金supported in part by National Key Research and Development Project (2020YFB1807204)in part by the National Natural Science Foundation of China (U2001213 and 61971191)+1 种基金in part by the Beijing Natural Science Foundation under Grant L201011in part by Jiangxi Key Laboratory of Artificial Intelligence Transportation Information Transmission and Processing (20202BCD42010)
文摘Mobile edge computing(MEC)provides services to devices and reduces latency in cellular internet of things(IoT)networks.However,the challenging problem is how to deploy MEC servers economically and efficiently.This paper investigates the deployment problem of MEC servers of the real-world road network by employing an improved genetic algorithm(GA)scheme.We first use the threshold-based K-means algorithm to form vehicle clusters according to their locations.We then select base stations(BSs)based on clustering center coordinates as the deployment locations set for potential MEC servers.We further select BSs using a combined simulated annealing(SA)algorithm and GA to minimize the deployment cost.The simulation results show that the improved GA deploys MEC servers effectively.In addition,the proposed algorithm outperforms GA and SA algorithms in terms of convergence speed and solution quality.
基金Sponsored by the National Natural Science Foundation of China(Grant No.60973027)Science Projects of China National Ministry of Information Industry(Grant No.01XK230009)
文摘To determine CDN cache servers'placement reasonably,an idea that using graph partitioning to solve the problem was put forward through theoretical analysis and the specific algorithm of partitioning was researched. The concept of graph partitioning for CDN was defined. The conditions of graph partitioning for CDN were demonstrated: the sum of the weights of the nodes in each subarea is as close as possible; edge cut between the subareas is as large as possible; internal nodes in each subarea are connected as far as possible. By reference to light vertex matching algorithm of graph partitioning for network simulation,a multilevel k-way algorithm of graph partitioning for CDN was proposed. The maximized edge cut k-way KL refinement algorithm was discussed. Graph partitioning is a feasible way to solve the problem of CDN servers'placement. Multilevel k-way algorithm is a feasible algorithm for CDN graph partitioning.
基金Supported by the National Natural Science Foun-dation of China (90204008)
文摘The Web cluster has been a popular solution of network server system because of its scalability and cost effective ness. The cache configured in servers can result in increasing significantly performance, In this paper, we discuss the suitable configuration strategies for caching dynamic content by our experimental results. Considering the system itself can provide support for caching static Web page, such as computer memory cache and disk＇s own cache, we adopt a special pattern that only caches dynamic Web page in some experiments to enlarge cache space. The paper is introduced three different replacement algorithms in our cache proxy module to test the practical effects of caching dynamic pages under different conditions. The paper is chiefly analyzed the influences of generated time and accessed frequency on caching dynamic Web pages. The paper is also provided the detailed experiment results and main conclusions in the paper.
文摘The work deals with the development of analytical model of multichannel technical queuing system with unreliable servers and input memory where server failure flows and incoming request flows comply with Poissonian laws, while the flows of failed facilities repairs and flows of incoming requests comply with exponential laws of probability distribution. Random process of system change-over is a Markovian process with continuous time and discrete states. Relations binding basic parameters and output characteristics of the system indicated are obtained as probabilities of system staying in the given moment in one of the possible states. The proposed model is the most generalized compared to some models known in literature which could be considered as special cases of the considered model.
文摘With results on the infinite servers queue systems with Poisson arrivals - M｜G｜∞ queues - busy period, it is displayed an application of those queue systems in the unemployment periods time length parameters and distribution function study. These queue systems are adequate to the study of many population processes, and this quality is brought in here. The results presented are mainly on unemployment periods length and their number in a certain time interval. Also, some questions regarding the practical applications of the outlined formulas are briefly discussed.
文摘In this paper, we conduct research on the high robustness JavaEE enterprise development mode based on Hadoop and cloud servers. The current virtual machine real-time migration can only achieve manual migration, and cannot achieve full-automatic migration. In other words, when the server overload requires the administrator to artificially select a low-load host, and then hit migration command to implement the migration. In recent years, the Hadoop is becoming popular, and the read performance of the data is measured in terms of the time overhead for reading the required data. The key to reducing read time is to optimize that Hadoop cloud data read time and the RDBMS data query time. This paper integrates the mentioned techniques to construct the novel JavaEE enterprise development pattern that will promote the further development of the related techniques.
文摘A prison can be something different as ordinary people imagine. It can be a therapeutic and rehabilitative environment, a place where human beings can learn to cope with the stressful situations they could not deal with in their past. That is what happens in the Rencheng prison under the administration of Lineng Prison Sub-Bureau of East China’s Shandong Province.
基金partially supported by the computing power networks and new communication primitives project under Grant No. HC-CN-2020120001the National Natural Science Foundation of China under Grant No. 62102066Open Research Projects of Zhejiang Lab under Grant No. 2022QA0AB02
文摘In distributed machine learning(DML)based on the parameter server(PS)architecture,unbalanced communication load distribution of PSs will lead to a significant slowdown of model synchronization in heterogeneous networks due to low utilization of bandwidth.To address this problem,a network-aware adaptive PS load distribution scheme is proposed,which accelerates model synchronization by proactively adjusting the communication load on PSs according to network states.We evaluate the proposed scheme on MXNet,known as a realworld distributed training platform,and results show that our scheme achieves up to 2.68 times speed-up of model training in the dynamic and heterogeneous network environment.
基金This work was supported by the Ministry of Science and Technology of Taiwan,R.O.C.,under Grant MOST 110-2622-E-468-002 and 110-2218-E-468-001-MBK.
文摘The Internet of Vehicles(IoV)has evolved as an advancement over the conventional Vehicular Ad-hoc Networks(VANETs)in pursuing a more optimal intelligent transportation system that can provide various intelligent solutions and enable a variety of applications for vehicular traffic.Massive volumes of data are produced and communicated wirelessly among the different relayed entities in these vehicular networks,which might entice adversaries and endanger the system with a wide range of security attacks.To ensure the security of such a sensitive network,we proposed a distributed authentication mechanism for IoV based on blockchain technology as a distributed ledger with an ouroboros algorithm.Using timestamp and challenge-responsemechanisms,the proposed authentication model can withstand several security attacks such asMan-in-Middle(MiM)attacks,Distributed Denial of Service(DDoS)attacks,server spoofing attacks and more.The proposed method also provides a solution for single-point failure,forward secrecy,revocability,etc.We exhibit the security of our proposed model by using formal(mathematical)analysis and informal analysis.We used Random Oracle Model to perform themathematical analysis.In addition,we compared the communication cost,computation cost,and security of the proposed model with the related existing studies.We have verified the security of the model by using AVISPA tool simulation.The security analysis and computation analysis show that the proposed protocol is viable.
文摘In cloud data centers,the consolidation of workload is one of the phases during which the available hosts are allocated tasks.This phenomenon ensures that the least possible number of hosts is used without compromise in meeting the Service Level Agreement(SLA).To consolidate the workloads,the hosts are segregated into three categories:normal hosts,under-loaded hosts,and over-loaded hosts based on their utilization.It is to be noted that the identification of an extensively used host or underloaded host is challenging to accomplish.Thresh-old values were proposed in the literature to detect this scenario.The current study aims to improve the existing methods that choose the underloaded hosts,get rid of Virtual Machines(VMs)from them,andfinally place them in some other hosts.The researcher proposes a Host Resource Utilization Aware(HRUAA)Algorithm to detect those underloaded and place its virtual machines on different hosts in a vibrant Cloud environment.The mechanism presented in this study is contrasted with existing mechanisms empirically.The results attained from the study estab-lish that numerous hosts can be shut down,while at the same time,the user's workload requirement can also be met.The proposed method is energy-efficient in workload consolidation,saves cost and time,and leverages active hosts.
文摘There are different types of Cyber Security Attacks that are based on ICMP protocols. Many ICMP protocols are very similar, which may lead security managers to think they may have same impact on victim computer systems or servers. In this paper, we investigate impact of different ICMP based security attacks on two popular server systems namely Microsoft’s Windows Server and Apple’s Mac Server OS running on same hardware platform, and compare their performance under different types of ICMP based security attacks.
基金supported by the National Natural Science Foundation of China under Grant No.71971188the Humanities and Social Science Fund of Ministry of Education of China under Grant No.22YJCZH086the Hebei Natural Science Foundation under Grant Nos.G2023203008 and G2022203003。
文摘This paper studies customer joining behavior and system regulation strategy in nonexhaustive visible M/M/m queues with synchronous vacations of a part of the servers.Once this part of the servers are idle,they take multiple vacations simultaneously(vacation period).Until there are customers waiting in the queue,they are reactivated and all servers are busy or idle(busy period).The authors call this part of the servers as“partial servers”.In view of the fully visible queue and the almost visible queue,the authors obtain customers’equilibrium joining threshold strategies and their socially optimal joining threshold strategies,respectively,and observe that customer joining behavior in equilibrium generally makes the system overcrowded,which makes the equilibrium social welfare lower than the optimal social welfare.After regulation,interestingly,for optimizing social welfare,the system manager hopes not only customers arriving in vacation period pay attention to the number of partial servers,but also customers arriving in busy period should care about it rather than ignore.Moreover,arranging more servers for vacation does not necessarily lead to the decrease of social welfare on condition that the number of partial servers is close to m.As for the information advantage of the fully visible case,it is not obvious for increasing social welfare and even unfavorable to servers’profit unless the number of partial servers is big enough.Furthermore,given the different composition of social welfare,there exists the optimal number of partial servers and the optimal arrival rate of customers for maximizing social welfare.
文摘Rise in inlet air temperature increases the corresponding outlet air temperature from the server. As an added effect of rise in inlet air temperature, some active servers may start exhaling intensely hot air to form a hotspot. Increase in hot air tem- perature and occasional hotspots are an added burden on the cooling mechanism and result in energy wastage in data centers. The increase in inlet air temperature may also result in failure of server hardware. Identifying and comparing the thermal sensi- tivity to inlet air temperature for various servers helps in the thermal-aware arrangement and location switching of servers to minimize the cooling energy wastage. The peak outlet temperature among the relocated servers can be lowered and even be homogenized to reduce the cooling load and chances of hotspots. Based upon mutual comparison of inlet temperature sensitivity of heterogeneous servers, this paper presents a proactive approach for thermal-aware relocation of data center servers. The experimental results show that each relocation operation has a cooling energy saving of as much as 2.1 kW·h and lowers the chances of hotspots by over 77%. Thus, the thermal-aware relocation of servers helps in the establishment of green data centers.
基金supported by the National Key R&D Program of China(2018YFB1003404)the National Natural Seience Foundation of China(Grant Nos.62072083,U1811261,61902366)+2 种基金Basal Research Fund(N180716010)Liao Ning Revitalization Talents Program(XLYC1807158)the China Postdoctoral Science Foundation(2020T130623).
文摘Parameter server(PS)as the state-of-the-art distributed framework for large-scale iterative machine learning tasks has been extensively studied.However,existing PS-based systems often depend on memory implementations.With memory constraints,machine learning(ML)developers cannot train large-scale ML models in their rather small local clusters.Moreover,renting large-scale cloud servers is always economically infeasible for research teams and small companies.In this paper,we propose a disk-resident parameter server system named DRPS,which reduces the hardware requirement of large-scale machine learning tasks by storing high dimensional models on disk.To further improve the performance of DRPS,we build an efficient index structure for parameters to reduce the disk I/O cost.Based on this index structure,we propose a novel multi-objective partitioning algorithm for the parameters.Finally,a flexible workerselection parallel model of computation(WSP)is proposed to strike a right balance between the problem of inconsistent parameter versions(staleness)and that of inconsistent execution progresses(straggler).Extensive experiments on many typical machine learning applications with real and synthetic datasets validate the effectiveness of DRPS.
文摘In this paper we consider a queueing network consisting of two parallel servers and threearrival streams generated by independent Poisson sources. Each server has its own queue and receivescustomers from its own arrival stream. A third arrival stream consists of customers which place resourcedemands on both servers, which are handled separately by each server once the request is made. Eachservice time is independent and exponentially distributed. Each customer in the system pays a holdingcost per unit time. The objective is to dynamically determine the optimal scheduling policy to the thirdstream of conupled customers. based on the state of the system, so as to minimize the average cost. Thismodel is new, and has Policy implications for computer or communication networks. A fuzzy approachis presented to solve this problem. Simulation shows that the approach is efficient and promising.
基金supported by the Strategic Priority Research Program of the Chinese Academy of Sciences under Grant No.XDA06010401the National Natural Science Foundation of China under Grant Nos.61202056,61331008,61221062the HuaweiResearch Program of China under Grant No.YBCB2011030
文摘The high-density server is featured as low power, low volume, and high computational density. With the rising use of high-density servers in data-intensive and large-scale web applications, it requires a high-performance and cost-efficient intra-server interconnection network. Most of state-of-the-art high-density servers adopt the fully-connected intra-server network to attain high network performance. Unfortunately, this solution costs too much due to the high degree of nodes. In this paper, we exploit the theoretically optimized Moore graph to interconnect the chips within a server. Accounting for the suitable size of applications, a 50-size Moore graph, called Hoffman-Singleton graph, is adopted. In practice, multiple chips should be integrated onto one processor board, which means that the original graph should be partitioned into homogeneous connected subgraphs. However, the existing partition scheme does not consider above problem and thus generates heterogeneous subgraphs. To address this problem, we propose two equivalent-partition schemes for the Hoffman-Singleton graph. In addition, a logic-based and minimal routing mechanism, which is both time and area efficient, is proposed. Finally, we compare the proposed network architecture with its counterparts, namely the fully-connected, Kautz and Torus networks. The results show that our proposed network can achieve competitive performance as fully-connected network and cost close to Torus.
基金supported by NSFC(61571055)fund of SKL of MMW(K201815)Important National Science and Technology Specific Projects(2017ZX03001028)
文摘To reduce fetching cost from a remote source,it is natural to cache information near the users who may access the information later.However,with development of 5 G ultra-dense cellular networks andmobile edge computing(MEC),a reasonable selection among edge servers for content delivery becomes a problem when the mobile edge obtaining sufficient replica servers.In order to minimize the total cost accounting for both caching and fetching process,we study the optimal resource allocation for the content replica servers’ deployment.We decompose the total cost as the superposition of cost in several coverages.Particularly,we consider the criterion for determining the coverage of a replica server and formulate the coverage as a tradeoff between caching cost and fetching cost.According to the criterion,a coverage isolation(CI) algorithm is proposed to solve the deployment problem.The numerical results show that the proposed CI algorithm can reduce the cost and obtain a higher tolerance for different centrality indices.
文摘When designing a multimedia server, several things must be decided: which scheduling scheme to adopt, how to allocate multimedia objects on storage devices, and the round length with which the streams will be serviced. Several problems in the designing of large-scale multimedia servers are addressed, with the following contributions: (1) a striping scheme is proposed that minimizes the number of seeks and hence maximizes the performance; (2) a simple and efficient mechanism is presented to find the optimal striping unit size as well as the optimal round length, which exploits both the characteristics of VBR streams and the situation of resources in the system; and (3) the characteristics and resource requirements of several scheduling schemes are investigated in order to obtain a clear indication as to which scheme shows the best performance in realtime multimedia servicing. Based on our analysis and experimental results, the CSCAN scheme outperforms the other schemes. It is believed that the results are of value in the design of effective large-scale multimedia servers. Keywords realtime multimedia - storage server - scheduling - data placement - buffer management - variable bit rate This work was supported in part by the University IT Research Center Project and Sunmoon University Research Project.Kyung-Oh Lee is an associate professor in the Faculty of Computer and Information Sciences, Sunmoon University, Korea. He received his B.S., M.S. and Ph.D. degrees in computer science from Seoul National University in 1989, 1994 and 1999, respectively. His current research interests include multimedia system, database, mobile communication. He is a member of KIPS (Korea Information Processing Society).Jungho-Ho Park is a professor in the Divisions of Computer and Information Sciences, Sunmoon University, Korea. He received his M.S. and Ph.D. degrees in computer science from Osaka University in 1987 and 1990, respectively. His current research interests include distributed algorithms, e-learning and electronic commerce. He is a director of KIPS (Korea Information Processing Society) and a vice president of KIPS-IT certification.Yoon-Young Park is an associate professor in the Faculty of Computer and Information Sciences, Sunmoon University, Korea. He received his M.S. and Ph.D. degrees in computer science from Seoul National University in 1985 and 1994, respectively. His current research interests include embedded systems and sensor networks. He is a member of KIPS (Korea Information Processing Society).
