Enabling Scalable Information Sharing for Distributed Applications Through Dynamic Replication

Replication;Distributed middleware system;Information dissemination

Chang, Tianying ; Computing

University:Georgia Institute of Technology

Department:Computing

关键词: Replication;    Distributed middleware system;    Information dissemination;   

Others  :  https://smartech.gatech.edu/bitstream/1853/7589/1/chang_tianying_200512_phd.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

As broadband connections to the Internet become more common, newinformation sharing applications that provide rich services todistributed users will emerge. Furthermore, as computing devicesbecome pervasive and better connected, the scalability requirementsfor Internet-based services are also increasing. Distributed objectmiddleware has been widely used to develop such applications sinceit made it easier to rapidly develop distributed applications forheterogeneous computing and communication systems. As theapplication's scale increases, however, the client/serverarchitecture limits the performance due to the bottleneck at thecentralized servers. The recent development in peer-to-peertechnologies creates a new opportunity for addressing scalabilityand performance problems for services that are used by many nodes.In a peer-to-peer system, peer nodes can contribute a fraction oftheir resources to the system, enabling more flexible and extendedsharing between the entities in the system. When peer nodes arerequired to contribute their resources by replicating a service forself and others, however, several new challenges arise.Our thesis is that non-dedicated resources in a distributed systemcan be utilized to replicate shared objects dynamically so that thequality and scalability of a distributed service can be achievedwith lower cost by replicating the objects at right places andupdates to those shared objects can be disseminated efficiently andquickly. The following are the contributions of our work that hasbeen done to validate the thesis.1. A new fair and self-managing replication algorithm thatallows distributed non-dedicated resources to be used to improveservice performance with lower cost.2. A multicast grouping algorithm that is used to disseminateupdates to the shared objects among a large set of heterogeneouspeer nodes to keep consistent view for all peer nodes. It groupsnodes with similar interests into same group and multicasts all therequired data to the group so that the unwanted data received byeach node can be minimized.3. An overlay construction algorithm that aims at reducing bothnetwork latency and total network traffic when delivering datathrough the built overlay network.4. An implementation of a distributed object framework, GT-RMI,that allows peer nodes to invoke dynamically replicated objectstransparently. The framework can be configured for a particular peernode through a policy file.