Weijiang Liu

Xen Live Migration with Slowdown Scheduling Algorithm

With the increasing number of technology areas using Virtual Machine (VM) platforms, challenges exist in Virtual Machine migrating from one physical host to another. However, the complexity of these virtualized environments presents additional management challenges. Unfortunately, many traditional approaches may be either not effective well for reducing downtime or migration time, or not suitable well for Xen VMs platforms. This paper presents the design and implementation of a novel Slowdown Scheduling Algorithm (SSA) for Xen live VM migration. In our SSA methodology, the CPU resources which have been assigned to migration domain are decrease properly. That is, the dirtying page rate is reduced according to the decrease of CPU activity. Experimental results illustrate that our SSA approach can shorten both the total migration time and downtime obviously under high dirty page rate environment.

Resource preprocessing and optimal task scheduling in cloud computing environments

Cloud computing came into being and is currently an essential infrastructure of many commerce facilities. To achieve the promising potentials of cloud computing, effective and efficient scheduling algorithms are fundamentally important. However, conventional scheduling methodology encounters a number of challenges. During the tasks scheduling in cloud systems, how to make full use of resources and how to effectively select resources are also important factors. At the same time, communication delay also plays an important role in cloud scheduling, which not only leads to waiting between tasks but also results in much idle interval time between processing units. In this paper, a fuzzy clustering method is used to effectively preprocess the cloud resources. Combining the list scheduling with the task duplication scheduling scheme, a new directed acyclic graph based scheduling algorithm called earliest finish time duplication algorithm for heterogeneous cloud systems is presented. Earliest finish time duplication attempts to insert suitable immediate parent nodes of the current selected node in order to reduce its waiting time on the processor. The case study and experimental results illustrate that the algorithm proposed in this paper is better than the popular heterogeneous earliest finish time algorithms. Copyright

DAG Cluster Scheduling Algorithm for Grid Computing

During the tasks scheduling in Grid, communication delay plays an important role in Grid scheduling, which not only leads to waiting between tasks, but also results in much idle interval time between processing units. At the same time, how to make use of resources and how to select resources are also important factors. In this paper, combining the list scheduling with the task duplication scheduling scheme, a new DAG (Directed Acyclic Graph) cluster algorithm called CFTD (Cluster Earliest Finish Time Duplication) for heterogeneous Grid systems is presented. CFTD attempts to insert suitable immediate parent nodes of the current selected node in order to reduce its waiting time on the processor. The case study and experimental results show that the algorithm proposed in this paper is better than HEFT algorithm. Keywords-Grid computing; DAG; cluster algorithm; Heuristic algorithm

Detection of Superpoints Using a Vector Bloom Filter

Internet attacks, such as distributed denial-of-service attacks and worm attacks, are increasing in severity and frequency. Identifying and mitigating realtime attacks are an important and challenging task for network administrators. An infected host can make a large number of connections to distinct destinations during a short time. Such a host is called a superpoint. Detecting superpoints can be utilized for traffic engineering and anomaly detection. This paper proposes a novel data streaming method for detecting superpoints and proves guarantees on its accuracy with low memory requirements. The superior performance of this method comes from a new data structure, called vector bloom filter (VBF), which is a variant of standard BF. The VBF consists of six hash functions, four of which take some consecutive bits from the input string as the corresponding value, respectively. The information of superpoints is obtained by using the overlapping of hash bit strings of the VBF. Theoretical analysis and experimental results show that the proposed method can detect superpoints precisely and efficiently through comparison with other existing approaches.

Identifying Elephant Flows Using a Reversible MultiLayer Hashed Counting Bloom Filter

Identifying elephant flows is very important for many applications, such as differentiated services, load balancing and network management. Existing work requests relatively high burden. In this paper, we propose a new method to identify elephant flows. The proposed idea is based on a novel data structure called Reversible MultiLayer Hashed Counting Bloom Filter(RML-HCBF). An RML-HCBF includes a few of hash functions which select some consecutive bits from the original string as its function values. Although RML-HCBF does not preserve any flow identifier (ID) explicitly, the flow ID of an elephant can be reconstructed by using the overlapping of the hash bit strings. RML-HCBF can identify elephant flows without storing flow ID and performing flow ID lookup. We evaluate the performance of RML-HCBF through theoretical analysis and experiments on real network traffic traces. The results show that RML-HCBF can identify elephant flows accurately and efficiently.

Minimum Control Latency of SDN Controller Placement

SDN is known as a decoupled architecture separatingthe control and data planes. The controller is in charge ofnetwork nodes through a southbound interface. However, it iscostly to impose all controllers to control each node separately. Meanwhile, the layout of controllers affects the networks abilityto respond to network events. Therefore, the controller placementis an important problem in SDN community. When latencyand reliability metrics are both considered, there is usuallyno single best controller placement solution, but a trade-offinstead. Most of deployment schemes based on latency are mainlyfocused on transmission delay (TD) or Propagation delay (PD) atpresent. However, to our best knowledge, no existing works haveconsidered the influence of control latency on SDN controllerplacement problem while control latency makes great sense inquick response to network events. Thus, in this paper, a minimum-control-latency optimized algorithmbased on greedy controlling pattern design is proposed. Theideas and mechanisms are illustrated using the Internet2 OS3Etopology compared to average-latency-optimized placement andworst-case-optimized placement. Whats more, the accuracy oflatency measurement is also a main concern in SDN research. Itis not precise enough when measuring only based on Ping results. In this paper, a novel method based on active latency measurementis also presented to obtain more accurate control latency. Extensive experiments have shown that our minimum-controllatency-optimized can improve the imbalance when partitioningSDN domains and achieve the maximum number of nodes percontroller controlled.

Detecting superpoints through a reversible counting Bloom filter

Internet attacks such as distributed denial-of-service (DDoS) attacks and worm attacks are increasing in severity. Identifying realtime attack detection and mitigation of Internet traffic is an important and challenging problem. For example, a compromised host doing fast scanning for worm propagation often makes an unusually high number of connections to distinct destinations within a short time. We call such a host a superpoint, which are sources that connect to a large number of distinct destinations. Detecting superpoints is very important in developing effective and efficient traffic engineering schemes. We propose two novel schemes for detecting superpoints and prove guarantees on their accuracy and memory requirements. These schemes are implemented by introducing a reversible counting Bloom filter (RCBF), a special counting Bloom filter. The RCBF consists of 4 hash functions which projectively select some consecutive bits from original strings as function values. We obtain the information of superpoints using the overlapping of hash bit strings of the RCBF. The theoretical analysis and experiment results show that our schemes can precisely and efficiently detect superpoints.

A Novel Method for Estimating Flow Length Distributions from Double-Sampled Flow Statistics

Since the generation of detailed traffic statistics does not scale well with link speed, increasingly passive traffic measurement employs sampling at the packet or flow level. Sampling has become an attractive and scalable means to measure flow data on high-speed links. However, knowing the length distributions of traffic flows passing through a network link is useful for some applications such as inferring traffic demands, characterizing source traffic, and detecting traffic anomalies. Passive traffic measurement increasingly makes inferences from sampled network traffic. However, previous work has shown the inaccuracy of estimating flow length distributions from sampled traffic when the sampling is performed at the packet level. In this paper, we propose a novel method that uses flow statistics formed from double-sampled packet stream to infer the absolute frequencies of lengths of flows in the unsampled stream. We achieve this through statistical inference and by exploiting heavy-tailed feather. The method allow us to recover the complete flow length distribution.

Sparse Matrix Prediction Filling in Collaborative Filtering

Collaborative filtering is one of the most successful techniques that attempts to recommend items (such as music, movies, web sites) that are likely of interest to the people. However, Existing CF technique may work poorly due to the sparse attribute inherent to the rating data. In this paper, a new mechanism that combines the user-based rating and item attribute-based is presented. First, we use the inherent item attributes to construct Boolean matrix. Second, we propose a novel blank unrated element prediction approach to compute the similarity of items by comparing the Euclidean distance between two items. Case studies show that our approach contributes to predict the unrated blank data for sparse matrix. The filling-in accuracy is also acceptable and reasonable.

A Data Streaming Algorithm for Detection of Superpoints With Small Memory Consumption

A superpoint is a host that communicates with a large number of distinct destinations (sources) within a measurement period. Identifying superpoints is an important and meaningful task for network security and monitoring. To keep up with the line speed in a high-speed network, fast memory is indispensable for detecting superpoints. Moreover, the memory is also expensive and size-limited. In this letter, we propose a new data streaming algorithm for detecting superpoints, called Snare, which can work in tight memory space and yield good performance. Its accuracy and efficiency come from a new data structure snare and the compensation mechanism for the number of lost flows. Theoretical analysis and experimental results show that Snare can detect superpoints accurately and efficiently.