Future demands on storage systems are expected to stress current storage architectures on ail dimensions: amount of aggregate storage is expected to grow by 2 to 3 orders of magnitude, while data-granularity is expected to fall by at least an order of magnitude. Further, the network access bandwidths are also increasing at least by two orders of per link, and several times that in aggregate. We present a transactional filer, a extensible scalable file-system capable of "petabyte" and "petabit-per-sec" performance. This filer is built using, as its components, an all-silicon grid storage server, whose scalability derives from a no-software "server-less server" design. The grid storage server implements multiple layers of a protocol stack which are compiled into a hardware engine that processes all layers concurrently on-chip, without any software involvement. The protocol engine chip has a vanishingly low price point, at which it is possible to directly network-enable each individual hard-disk drive. This "serverless" design allows it to scale up to match "terabit" network speeds and "terabyte" disk capacities per node, aggregating to petabyte-petabit capacities across the large scale grid application. The transactional filer is a file-system is built upon this distributed silicon protocol engine architecture; and benefits from the advantages of reconfigrable computing. The file-system primitives which are implemented in the hardware protocol engine are simple and hence silicon-friendly, yet they can be combined together in ways to provide a very rich transactional versioned file-system interface to the end-application. While this newer grid-friendly file-system interface is exposed for new applications, the same hardware primitives can be used in a backward compatible POSIX manner to support legacy applications

Currently, there is no distributed data management system for computational grids that provides transparent data distribution and replication. This paper introduces three methods to enhance an algorithm for data distribution and replication from the field of peer-to-peer computing named "content-addressable network". They enable this algorithm to being a basis for a fully transparent distributed data management system for grids. The first enhancement provides multiple dataset queries which allow the user to search for all datasets matching a given combination of key and mask. Another optimization reduces existing query latencies by introducing a system of specifically selected short cuts. The third enhancement enables the algorithm to keep the amount of data stored on a certain node more equally balanced and as a result optimizes the algorithms stability while working with high data fill-levels

Grid is an ambitious and exciting global effort to develop an environment in which individual users can access computers, databases and experimental facilities simply and transparently, without having to consider where those faculties are located. Monitoring the behavior of the resources of a distributed system, especially grid, is necessary, both for determining the cause of performance problems, and for tuning the system, which results in optimized performance. However writing monitoring and management applications for grid resource is a complex task. This is due to the geographic distribution of resources and heterogeneous environments that are owned by different organizations. We have proposed architecture for autonomous grid node monitoring agent, to address the issues of complex resource monitoring and management. The architecture is based on grid monitoring architecture (GMA) and is autonomous as it provides functionality of equality, locality and self-containment, which are main components of autonomic controllability in any sub system as part of autonomous decentralized system (ADS) (K. Mori, 1993). The evaluation results of the proposed system proved the validity of the concept for achieving autonomous resource management in dynamic systems like grid

Resource-sharing is one of the grid intrinsic properties enabling efficient usage of services by clients from many virtual organizations. This is particularly true for network infrastructures, which are typically designed to serve a large number of heterogeneous users. In such an environment, the support of network quality of service requires the establishment of service level agreements between users and network service providers, admission control as well as network control planes for automatic or semi-automatic service provisioning. The availability of guarantees allows flexible usage of the distributed infrastructure and enables application-controllable performance. Nevertheless, many network providers today only offer best-effort services and do not support service level agreement signalling. In this paper we analyze requirements and introduce two bandwidth-oriented services for enhanced grid data transfer: the guaranteed delivery file transfer and virtual leased line services. We propose a complete architecture - fully integrated with grid middleware - for network service level agreement management and inter-domain network service provisioning. This is based on the functional division into agreement and service provider layers proposed by the Grid Resource Allocation and Agreement Protocol working group of the Global Grid Forum

In order to provide cost effective transport services for highly demanding data-intensive grid applications, National Research Networks (NRNs) are considering additional types of access to their network infrastructures. Next to traditional IP access, NRNs like to provide automated, grid application driven access to their underlying connection-oriented network infrastructure. This combination is called hybrid networking. Recently, both NRNs and grid communities started to acquire their own global optical network connections. A lambda grid was created when these organizations decided to interconnect these links via interconnection points such as Starlight and Netherlight. Apart from supporting scientific applications, the lambda grid allows network research. Within this context, this paper will present a novel token-based path selection mechanism that will enable authorized access to lambda grid links. The token-based approach allows temporal separation of the path authorization process from obtaining access to a lambda grid link. The path authorization process may involve many parties and complex, time consuming decisions whereas path access requires a fast real-time implementation. We will describe the application of the token-based approach for an interconnection point between a hybrid network and a lambda grid

In hybrid internetworks, the plurality of service options and compositions poses several cross-layer operational challenges. The growing popularity of such internetworks among researchers and advanced users - many of which are grid users - warrants a complexity relief effort in support of the internetworks' continued scalability, all the way to planet scale. This paper proposes the use of semantic Web techniques to automate the processing of networking meta-data in a hybrid internetwork, with the two-fold outcome of raising the degree of confidence in the internetwork and lowering its operating costs. By virtue of latching onto the semantic Web curve, a new breed of power tools - bots, compilers, browsers - will be delivering on the aforementioned outcomes, while fully leveraging COTS and open source building blocks. Furthermore, the investment in the semantic Web is thought to future proof a hybrid internetwork and readies it for times when service metrics other than connectivity (e.g., resiliency, protocol transparency, security etc.) will be evaluated end-to-end. The approach appears to be applicable to the global lambda integrated facility (GLIF) as well as other advanced, experimental communities

Grid computing systems are expected to be more scalable, more survivable from partial systems failures and more adaptive to dynamic network environments in order to improve user experience, expand system's operational longevity and reduce maintenance cost. Based on the observation that many biological systems have already overcome these requirements, the proposed network architecture, called SymbioticSphere, applies biological concepts and mechanisms to design grid systems (application services and middleware platforms). In SymbioticSphere, each application service and middleware platform is designed as an artificial biological entity, analogous to an individual bee in a bee colony. Application services and middleware platforms implement biological concepts and mechanisms such as decentralization, energy level, healthy level, energy exchange between species, environment sensing, migration, replication and death. Like in biological systems, desirable system characteristics such as scalability, survivability and adaptability emerge from the collective actions and interactions of application services and middleware platforms. This paper presents the architectural design of SymbioticSphere, and describes how application services and middleware platforms act and interact with each other. Preliminary simulation results show that application services and middleware platforms collectively adapt to dynamic changes in the network (e.g. user location, network traffic and resource availability)

The distributed optical testbed (DOT) is a wide area experimental grid infrastructure designed to facilitate the creation of new techniques for the efficient execution of distributed applications, supported by agile optical networks. The architecture of the testbed enables the close investigation of new methods for supporting distributed heterogeneous environments that are interconnected by high performance, application-addressable lightpaths. Currently, almost all grid implementations are supported through communication services based on non-adaptive, non-deterministic packet-routed data networks. The DOT architecture allows grid applications to take advantage of flexible, reconfigurable, and deterministic optical channels. By directly provisioning lightpaths, applications can enhance data communications by supplementing or by passing traditional data transport services. Experiments conducted on this testbed demonstrate that a significant potential exists for supporting grid applications through adaptive optical networking

We study optical lightpath networking for distributed computing environment. Lightpath networks are suitable to network-wide distributed computing because the bandwidth is assured once a lightpath is established. In this paper, we dynamically configure a unidirectional optical ring that is used for multipoint-to-multipoint communications among hosts in a group. The ring consists of multiple unidirectional lightpaths. Each host in the group is an endpoint of the lightpath. The routing of the network is usually specific to point-to-point communications. We propose a ring composition algorithm that finds the least-cost set of lightpaths of the optical ring under this routing. In this algorithm, a parent host finds a destination host of which route from the host is the least-cost route among routes to hosts in the group. We also propose backward signaling methods that are specific to the establishment of the lightpaths for the ring composition. Since the signaling for lightpath establishment is well adjusted to RSVP-TE for GMPLS, an implementation of the ring composition can be accelerated. Through simulation experiments, we find the proposed ring composition method generates lightpaths for the optical ring with less wavelength resources. The smart signaling accelerates the efficiency of network resources. The clear reduction of blocking probability of establishing a ring for a group is found from the simulation result

This paper presents the adaptive reliable optical burst switching (AR-OBS) paradigm that employs two-way reservation, and edge burst aggregation to guarantee deterministic data latency and loss rate under a distributed control plane. AR-OBS employs the adaptive robust fast optical reservation protocol (A-RFORP) to provide adaptive blocking recovery control with offset delay adjusted to take into account of blocking recovery delay. And delayed reservation is employed to ensure wavelengths are reserved just in time before the data bursts arrive to maximize the wavelength utilization. Performance of AR-OBS is discussed in terms of burst reservation delay, burst aggregation overflow probability, burst reservation blocking probability and wavelength utilization. It is found that performance of AR-OBS is compared favorably with other OBS schemes based on two-way reservation and edge burst aggregation

Grid computing provides a flexible, secure, sharing of resources among dynamic collections of individuals, institutions, and resources. Optical burst switching (OBS) is one of the most promising new optical transport paradigms for efficiently transporting data over an all-optical network. In this paper, we discuss several mechanisms for deploying a reliable OBS network using a range of loss minimization and loss recovery mechanisms. We also propose two new loss recovery mechanisms for deploying a reliable OBS network, namely, forward error correction and composite burst assembly. Our simulation shows that the two proposed mechanisms significantly reduce the loss experienced by loss-sensitive connection-oriented traffic through an OBS network. We have also shown how the reduction of loss leads to increased throughput of TCP-over-OBS network. We also propose a comprehensive OBS service architecture, and provide guidelines for implementing a reliable connection-oriented photonic transmission control protocol (PTCP) over OBS-based grid networks

We propose and evaluate a novel scheme, referred to as bimodal burst switching (BBS) scheme, for optical burst switching (OBS) star networks, which can be used in grid networks. The main feature of this scheme is that it has zero burst loss and a large geographical coverage. Using simulation techniques we show that the proposed scheme provides high bandwidth and outperforms OBS

This paper presents a novel architecture of synchronous stream optical burst switching (SS-OBS), which employs a unified traffic control plane to enable a wavelength-routed optical network with integrated sub-wavelength grooming services to support sub-wavelength streaming traffic of constant bit rate (CBR) and variable bit rate (VBR) with guaranteed end-to-end delay and data loss rate. SS-OBS shapes sub-wavelength CBR streams into periodic burst trains; aggregates multiple sub-wavelength VBR streams with the same source and destination into a statistically multiplexed periodic burst train. By adopting periodic advanced scheduling of burst trains, SS-OBS substantially decreases recursive signaling processing in supporting sub-wavelength streaming traffic. SS-OBS employs burst train reservation blocking recovery by time-shifting a blocked burst train. Performance of SS-OBS is analyzed in terms of wavelength utilization and data loss. It is found that the performance of the SS-OBS under each type of traffic is compared favorably with other single-service-based OBS schemes

Dynamic optical networks hold the potential of satisfying very large bandwidth requirements of many of the grid applications. However, encapsulation of optical network elements into manageable grid resources and dynamic provisioning of lightpaths is necessary to meet the complex demand patterns of the grid applications and to optimize usage of optical network components. In this paper, we first present a scalable algorithm for an NP-hard problem of scheduling on-demand and advance reservation requests for lightpaths. We then investigate in detail the effect of proportion of advance reservations, laxity and distribution of the size of data transfer requests on performance through extensive experimentation. The paper also investigates that how much improvement in performance can be gained by segmenting large data transfer requests into multiple requests of smaller sizes and up to what percentage of overheads is segmentation justified in scheduling of lightpaths. We demonstrate how laxity can be exchanged for segmentation to achieve high utilization of lightpaths

Dynamic assignment of complex applications' resource need on a dynamically aggregated computing infrastructure assembled via a network is central to autonomic grid computing and several other netcentric computing paradigms. In this paper we present an algorithm for active flow assignment in autonomic grid system. Active flow refers to the case where the bandwidth requirement between the flow processing elements can change due to active transformation/processing of the flow. We present the algorithm with the help of a dynamically deployable video grid application that is capable of negotiating local network state based rate and let the stream propagate over, grid network with highly asymmetric link and node capacities with minimum delay involved, in two diverse models, MISD and SIMD. The algorithm uses network connectivity graph factorization and thus adheres to the hierarchical nature of the autonomous networks

For resource intensive tasks, effective utilization of resources is essential to achieve scalability and a satisfactory quality of service. The number of resources required and how the resources should be grouped depends on many factors, such as, network delays, the availability of resources and their current utilizations. Making such decisions at runtime using information, such as, current system state can improve performance. This paper proposes an adaptive architecture in which resource groupings and their patterns of interactions can be changed dynamically based on the task to be performed and the system state. A prototype implementation, valid for a certain type of tasks, is described. This architecture can be extended for other types of tasks using the concepts discussed in this paper

We make a comparative analysis of two proposals for setting up a facilities-based metropolitan network to serve the education and research community in the city of Belem do Para in Brazil, using the "do-it-yourself" (DIY) model of building infrastructure, and benefiting around a dozen research and education institutions. The first alternative is to build a fiber optic network, using Gigabit Ethernet technology. The second one uses a metropolitan wireless network, based on the IEEE 802.16 standard, studying the feasibility of the use of this technology in a context with the particular characteristics of the Amazon region. In both cases, we show that the capital cost for each institution is feasible, and that the total cost in the medium term is lower than the recurrent costs of maintaining the traditional model based on telecommunications provider services

The Brazilian National Research and Education Network - RNP - is currently engaged in a nationwide program to upgrade the capacity of the communications networks used by the Education and Research institutions in the country. An essential component of this program is the deployment of metropolitan optical networks in the major population centers. The article describes the background to this initiative, the methodology adopted and the expected results

Datacasting is a new service offering available as a result of the deployment of digital television (DTV). Several services are available to users of datacasting in the educational, business and public safety arena. Using the ATSC A/90 Standard Model, datacasting has the ability to reach the public in areas traditionally not support by existing high-speed data transmission services

This paper presents the work that is taking place in the Region of Western Greece in order to develop state-of-the-art broadband infrastructure. The current status of broadband infrastructure in the region is being analyzed and the main principles on which the deployment of the networks will be based on is being described, by presenting the metropolitan area networks and the wireless access networks that will be developed. Two case studies of such municipal networks are also presented in this paper

Emerging multiple service broadband networks will support applications with diverse performance objectives and traffic characteristics. The evolution from the current "best effort" service, to QoS-enabled IP services should be performed in a smooth and scalable way. MUSE architecture exploits the combination of class-based forwarding and per-flow admittance in order to yield a promising, effective and scalable solution. In order to support also migration scenarios, the proposed architecture is generic enough to cover different access technologies with multiple link rates. Although multiple mechanisms on different planes should be incorporated, the goal is the same; to provide service provisioning with quantitative and qualitative measures for customers' satisfaction. Our attention is focused on data plane and especially on the behavior of the residential gateway (RGW). This paper gives a simulative performance analysis of the proposed traffic differentiation and buffer settings effects when applied on the QoS block of the RGW. Performance results are derived from real network traffic simulations on a model based on the proposed architecture and functionality of the gateway for relatively slow access lines. The results will show the special features of the applied MUSE architecture and will highlight its capability to provide predefined qualitative metrics on the RGW node so that the operators could build a well-engineered network, achieving QoS provisioning in the access network

In some communities, we observe a single vertically integrated FTTP network owner and service provider that can either be profit maximizing or welfare maximizing. In other cases, either out of choice or regulation, we observe the network owner (the wholesaler) leasing facilities to competing service providers (retailers), who then provide voice, video and data service over the shared network (wholesale-retail split). The network owner can either wholesale dark fiber or "lit" transport (and can be profit maximizing or welfare maximizing). In this paper, using a simplified model where subscribers always buy voice service and choose between either broadband data service or video service or the video and data bundle, we show that such a wholesale-retail split interferes with the ability of the network owner to price discriminate. We conclude that for a bivariate correlated normal distribution of willingness to pay for data and video services, the supply side economics of FTTP (i.e., the fact that the marginal cost of provisioning the bundle is less than the sum of the marginal costs of provisioning the individual services) ensures that the bulk of the extractable economic surplus resides in the bundle and not in the individual services. Since the wholesale-retail split interferes with wholesaler's ability to extract economic surplus only from an individual service (video service) and not from the bundle, the implications (the difference between the profits of a profit maximizing wholesaler and a profit maximizing vertically integrated entity) are modest, at best. Our analysis also shows that municipalities or communities that build out FTTP and choose to be wholesalers (i) can realize sustainable prices, (ii) are likely to create greater welfare and (iii) are just as likely to recover costs (vis-a-vis vertically integrated entities)

This article proposes a new architecture, which once implemented, would help in achieving end-to-end quality of service (QoS) requirements of an application which is being served in an interworking system of IEEE 802.16/WiMAX and IEEE 802.11e/WiFi networks. Our approach strives at mapping the QoS requirements of an application originating in IEEE 802.11e network to a serving IEEE 802.16 network and assuring the transfer of data having appropriate QoS back to the application in IEEE 802.11e network. We discuss how an application flow specifies its QoS requirements, either in an IEEE 802.11e or IEEE 802.16 network and the mechanisms that ensure that these requirements are known to the serving network. We identify the necessary parameters, as per advice in the standards, that could stipulate the QoS requirements for an application depending upon traffic type it represents. We propose the mapping of various parameters for different kinds of flows which would ultimately make sure that an application receives the QoS it requested. The resulting architecture would work as a hybrid of two different kinds of networks

Ethernet passive optical network (EPON) has drawn considerable research attention in recent years. EPON can be implemented in different topologies in access network environment. Recently, the tree topology was the focus of the researchers. In this paper, we propose a cost effective novel ring-based EPON architecture with associated dynamic bandwidth allocation (DBA) algorithm as an alternative to standard EPON architectures. This architecture, compared with conventional ring, allows optical line terminal (OLT) receiver to function with narrower dynamic range through minimization of near-far problem. In addition, the proposed DBA algorithm outperforms other DBA algorithms proposed in the literature. This proposed architecture is proved to be fault-tolerant and cost effective. Furthermore, this novel approach also has the potential of inter-optical network units (ONU) connectivity, which can bring LAN support in access environment

Despite its imperfections, GSM security has stood well the test of time. In part, this security success has relied on closed platforms that prevent the end-user from tampering with the GSM protocol stacks. While it is possible to build phones that do not have such restrictions, this is difficult due to, e.g., legislation and technical complexity. Unlicensed Mobile Access (UMA) is a new technology that provides access to GSM services over Wireless LAN or Bluetooth. It also challenges the assumption of closed platforms, since it is relatively easy to implement a UMA phone purely in software running on standard PC hardware and operating systems. This paper examines the security implications of UMA for GSM security, focusing especially on the impact of open terminal platforms. We identify several areas where open platforms may increase risks to both honest users and network operators, and propose countermeasures for mitigating these risks.