Many important cloud services require replicating massive data from one datacenter (DC) to multiple DCs. While the performance of pair-wise inter-DC data transfers has been much improved, prior solutions are insufficient to optimize bulk-data multicast, as they fail to explore the rich inter-DC overlay paths that exist in geo-distributed DCs, as well as the remaining bandwidth reserved for online traffic under fixed bandwidth separation scheme. To take advantage of these opportunities, we present BDS+, a near-optimal network system for large-scale inter-DC data replication. BDS+ is an application-level multicast overlay network with a fully centralized architecture, allowing a central controller to maintain an up-to-date global view of data delivery status of intermediate servers, in order to fully utilize the available overlay paths. Furthermore, in each overlay path, it leverages dynamic bandwidth separation to make use of the remaining available bandwidth reserved for online traffic. By constantly estimating online traffic demand and rescheduling bulk-data transfers accordingly, BDS+ can further speed up the massive data multicast. Through a pilot deployment in one of the largest online service providers and large-scale real-trace simulations, we show that BDS+ can achieve 3- 5× speedup over the provider's existing system and several well-known overlay routing baselines of static bandwidth separation. Moreover, dynamic bandwidth separation can further reduce the completion time of bulk data transfers by 1.2 to 1.3 times.

Weather radar images provide critical information for mesoscale weather nowcasting which plays significant roles in a range of fields including civil aviation and navigation. Differed from traditional radar exploration methods, this paper presents a novel prediction model based on a deep recurrent neural network (DeepRNN). The approach converts the task of nowcasting to a task of image series prediction. We first design a new loss function that pays more attention to the changes of images in the input sequence. In the mean while, an image discriminator is incorporated into the model to improve the visual quality of predicted images. Furthermore, optical flow is explored to preserve the the motion information. The prediction results are evaluated based on widely used statistic scores. The experimental results show that the proposed model leads to significant improvement in tasks of 2 hours forecasting of radar echo.

Many cellular responses to surrounding cues require temporally concerted transcriptional regulation of multiple genes. In prokaryotic cells, a single-input-module motif with one transcription factor regulating multiple target genes can generate coordinated gene expression. In eukaryotic cells, transcriptional activity of a gene is affected by not only transcription factors but also the epigenetic modifications and three-dimensional chromosome structure of the gene. To examine how local gene environment and transcription factor regulation are coupled, we performed a combined analysis of time-course RNA-seq data of TGF-β treated MCF10A cells and related epigenomic and Hi-C data. Using Dynamic Regulatory Events Miner (DREM), we clustered differentially expressed genes based on gene expression profiles and associated transcription factors. Genes in each class have similar temporal gene expression patterns and share common transcription factors. Next, we defined a set of linear and radial distribution functions, as used in statistical physics, to measure the distributions of genes within a class both spatially and linearly along the genomic sequence. Remarkably, genes within the same class despite sometimes being separated by tens of million bases (Mb) along genomic sequence show a significantly higher tendency to be spatially close despite sometimes being separated by tens of Mb along the genomic sequence than those belonging to different classes do. Analyses extended to the process of mouse nervous system development arrived at similar conclusions. Future studies will be able to test whether this spatial organization of chromosomes contributes to concerted gene expression.

Many important cloud services require replicating massive data from one datacenter (DC) to multiple DCs. While the performance of pair-wise inter-DC data transfers has been much improved, prior solutions are insufficient to optimize bulk-data multicast, as they fail to explore the capability of servers to store-and-forward data, as well as the rich inter-DC overlay paths that exist in geo-distributed DCs. To take advantage of these opportunities, we present BDS, an application-level multicast overlay network for large-scale inter-DC data replication. At the core of BDS is a fully centralized architecture, allowing a central controller to maintain an up-to-date global view of data delivery status of intermediate servers, in order to fully utilize the available overlay paths. To quickly react to network dynamics and workload churns, BDS speeds up the control algorithm by decoupling it into selection of overlay paths and scheduling of data transfers, each can be optimized efficiently. This enables BDS to update overlay routing decisions in near realtime (e.g., every other second) at the scale of multicasting hundreds of TB data over tens of thousands of overlay paths. A pilot deployment in one of the largest online service providers shows that BDS can achieve 3-5 x speedup over the provider's existing system and several well-known overlay routing baselines.

Cross-DR WAN (Datacenter Region Wide Area Network) with various services are deployed to provide timely data information and analytics for users in a wide range of geographical locations. For its reliability and performance, data duplication synchronization is essential among different IDCs (Internet datacenters). However, this problem poses a challenge. First, data duplication requires huge amount of bandwidth whereas the bandwidth of cross-DR links and the upload/download rates of server interfaces are limited. Second, data transmissions are time sensitive, but the current network cannot complete such tasks in a timely manner. In this work, we present PieBridge, a cross-RD data duplicate transmission platform that accommodates hundreds of TBs of data generated from user applications online data analytics. We deployed PieBridge on the IDCs of Baidu and obtained promising performance results in comparison with the prevalent approaches.

It is long known attackers may use forged source IP address to conceal their real locations. To capture the spoofers, a number of IP traceback mechanisms have been proposed. However, due to the challenges of deployment, there has been not a widely adopted IP traceback solution, at least at the Internet level. As a result, the mist on the locations of spoofers has never been dissipated till now. This paper proposes passive IP traceback (PIT) that bypasses the deployment difficulties of IP traceback techniques. PIT investigates Internet Control Message Protocol error messages (named path backscatter) triggered by spoofing traffic, and tracks the spoofers based on public available information (e.g., topology). In this way, PIT can find the spoofers without any deployment requirement. This paper illustrates the causes, collection, and the statistical results on path backscatter, demonstrates the processes and effectiveness of PIT, and shows the captured locations of spoofers through applying PIT on the path backscatter data set. These results can help further reveal IP spoofing, which has been studied for long but never well understood. Though PIT cannot work in all the spoofing attacks, it may be the most useful mechanism to trace spoofers before an Internet-level traceback system has been deployed in real.

IP spoofing is a well-known security threat on the Internet. Though there have been a number of spoofing prevention mechanisms, due the diversity of networks and management objectives, the operators may prefer a framework which enables easy installation and modification of the IP spoofing prevention solution, rather than a single mechanism. In this article, a lightweight and efficient framework for route-based IP spoofing filtering, named SEFA, is proposed. Through providing a collective view of the network and decoupling the filtering rule generation from network devices, SEFA enables easily installation of spoofing filtering application. SEFA mainly resolves the challenge that how to build network abstraction without taking full controllability. SEFA has been implemented based on slightly modifying commercial routers and an open source controller. Based on experiments, SEFA is found to be able to reduce the overhead and the latency of filtering rule generation and installation, while keeping off the complexity and latency of generating forwarding rules by the controller.

No abstract available.

In this paper, a potential threat to reliability of Software Defined Networking (SDN) is disclosed: the cascading failures of controllers. Current SDN designs have widely utilized multiple controllers and the load of a failed controller can be redistributed to the other controllers. However, simply utilizing multiple controllers cannot protect SDN networks from a single point of failure: the load of the controllers which carry the load of the failed controller can exceed the capacity of them, and then cascading failures of controllers will happen. In this article, at first we propose a model for such failures and present simulation results based on the model. Strategies for initial load balance and load redistribution after failure are designed to prevent such failures. The simulation result shows the strategies can significantly increase the resistance of SDN networks to cascading failures.

Current Internet is lack of validation on source IP address, resulting in many security threats. The future Internet can face the similar routing locator spoofing problem without careful design. The current in-progress source address validation standard, i.e., SAVI, is not of enough protection due to the solution space constraint. In this article, a mechanism named VAVE is proposed to improve the SAVI solutions. VAVE employs OpenFlow protocol, which provides the de facto standard network innovation interface, to solve source address validation problem with a global view. Significant improvements can be found from our evaluation results.

In IPv6 network, before configuring any address, a node must perform Duplicate Address Detection (DAD) to ensure the address is unique on link. However, original DAD is unreliable and vulnerable. In this article, a pull model DAD is designed, which achieves improvements both in reliability and security through changing the solicitation model. Comparing with SEcure Neighbor Discovery (SEND), this proposal has advantage in lightweight overhead and flexibility of address generation. Through evaluation, it is found to be feasible and cost effective.

IP traceback can be used to find the origin of anonymous traffic; however, Internet-scale IP traceback systems have not been deployed due to a need for cooperation between Internet Service Providers (ISPs). This article presents an Internet-scale Passive IP Trackback (PIT) mechanism that does not require ISP deployment. PIT analyzes the ICMP messages that may scattered to a network telescope as spoofed packets travel from attacker to victim. An Internet route model is then used to help re-construct the attack path. Applying this mechanism to data collected by Cooperative Association for Internet Data Analysis (CAIDA), we found PIT can construct a trace tree from at least one intermediate router in 55.4% the fiercest packet spoofing attacks, and can construct a tree from at least 10 routers in 23.4% of attacks. This initial result shows PIT is a promising mechanism.

In this paper, we present a novel source address spoofing prevention method for IPv6 access network called CGA based source address authentication (CSAA). It makes use of CGA (cryptographically generated address) to generate an unspoofable identifier of host without PKI, and bind it to the authorized address of the host. Then all the packets sent out by the host can be validated in the first-hop router by a light-weight method.

The transcription factor Myc plays a central role in regulating cell-fate decisions, including proliferation, growth, and apoptosis. To maintain a normal cell physiology, it is critical that the control of Myc dynamics is precisely orchestrated. Recent studies suggest that such control of Myc can be achieved at the post-translational level via protein stability modulation. Myc is regulated by two Ras effector pathways: the extracellular signal-regulated kinase (Erk) and phosphatidylinositol 3-kinase (PI3K) pathways. To gain quantitative insight into Myc dynamics, we have developed a mathematical model to analyze post-translational regulation of Myc via sequential phosphorylation by Erk and PI3K. Our results suggest that Myc integrates Erk and PI3K signals to result in various cellular responses by differential stability control of Myc protein isoforms. Such signal integration confers a flexible dynamic range for the system output, governed by stability change. In addition, signal integration may require saturation of the input signals, leading to sensitive signal integration to the temporal features of the input signals, insensitive response to their amplitudes, and resistance to input fluctuations. We further propose that these characteristics of the protein stability control module in Myc may be commonly utilized in various cell types and classes of proteins.

Summary: We describe a database and information discovery system named DIG (Duke Integrated Genomics) designed to facilitate the process of gene annotation and the discovery of functional context. The DIG system collects and organizes gene annotation and functional information, and includes tools that support an understanding of genes in a functional context by providing a framework for integrating and visualizing gene expression, protein interaction and literature-based interaction networks. Availability: The DIG system is freely accessible at http://dig.cagp.duke.edu/ Contact: [email protected] Supplementary information: http://data.cgt.duke.edu.dig