Search

The recent emergence of ultra high-speed networks up to 100 Gb/s has posed numerous challenges and has led to many investigations on efficient protocols to saturate 100 Gb/s links. However, end-to-end data transfers involve many components, not only protocols, affecting overall transfer performance. These components include disk I/O subsystem, additional computation associated with data streams, and network adapters. For example, achievable bandwidth by TCP may not be implementable if disk I/O or CPU becomes a bottleneck in end-to-end data transfer. In this paper, we first model all the system components involved in end-to-end data transfer as a graph. We then formulate the problem whose goal is to achieve maximum data transfer throughput using parallel data flows. We also propose a variable data flow GridFTP XIO stack to improve data transfer with data compression. Our contributions lie in how to optimize data transfers considering all the system components involved rather than in accurately modeling all the system components involved. Our proposed formulations and solutions are evaluated through experiments on the ESnet 100G testbed and a wide-area cluster-to-cluster testbed. The experimental results on the ESnet 100G testbed show that our approach is several times faster than Globus Online—8 × faster for datasets with many 10 MB files and 3–4 × faster for other datasets of larger size files. The experimental results on the cluster-to-cluster testbed show that our variable data flow approach is up to 4 × faster than a normal cluster data transfer. [ABSTRACT FROM AUTHOR]

Abstract: The Data Distribution Service (DDS) middleware has recently been standardized by the OMG. Prior to data communication, a discovery protocol had to locate and obtain remote DDS entities and their attributes. Specifically, DDS discovery matches the DataWriters (DWs) and DataReaders (DRs) entities (Endpoints) situated in different network nodes. DDS specification does not specify how this discovery is translated “into the wire”. To provide interoperability and transparency between different DDS implementations, the OMG has standardized the DDS Interoperability Wire Protocol (DDS-RTPS). Any compliant DDS-RTPS implementation must support at least the SDP (Simple Discovery Protocol). The SDP works in relatively small or medium networks but it may not scale as the number of DDS Endpoints increases. This paper addresses the design and evaluation of an SDP alternative–which uses Bloom Filters (BF)–that increases DDS scalability. BFs use Hash functions for space-efficient probabilistic data set representation. We provide both analytical and experimental studies. Results show that our approach can improve the discovery process (in terms of network load and node resource consumption), especially in those scenarios with large Endpoint per Participant ratios. [Copyright &y& Elsevier]

Abstract: In many wireless sensor network applications, nodes are mobile, while many protocols proposed for these networks assume a static network. Thus, it is important to evaluate if a traditional protocol designed for a static network can tolerate different levels of mobility. This paper provides an analytic model to investigate the effect of mobility on a well-known cluster-based protocol, LEACH. The model evaluates data loss after construction of the clusters due to node mobility, which can be used to estimate a proper update interval to balance the energy and data loss ratio. Thus, the results can help the network designer to adjust the topology update interval given a value of acceptable data loss threshold. A practical approach to increase the mobility tolerance of the protocol is applying a buffer zone to the transmission ranges of the nodes. The model is extended in order to consider the effect of buffer zone. To validate the analytic evaluations, extensive simulations are conducted and correctness of the evaluations is tightly verified. [Copyright &y& Elsevier]

We define a finite-state message-passing model using guarded commands. This model is particularly appropriate for defining and reasoning about self-stabilizing protocols, due to the well-known result that self-stabilizing protocols on unbounded-channel models must have infinitely many legitimate states. We argue that our model is more realistic than other models, and demonstrate its use with two simple examples. We then give a translation from this model to a lower-level model that uses a notion of time. We argue that this latter model is very close to a real network. We conclude by discussing how this translation might be used to implement self-stabilizing protocols on actual networks. [Copyright &y& Elsevier]