Kaashoek, M.Frans, and, Tanenbaum, S. Andrew, "Efficient Reliable Group Communication for Distributed Systems" Dept. of Math and Computer Science, Vrije Universiteit, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....a single response from one member of the current group for successful delivery of a message. k responses: The sending process expects to obtain k responses for the delivered message from the current receiving group of processes. By using k response semantics the groups resilience can be defined [Kaashoek and Tanenbaum 1994]. The resilience of a group is based on the minimum number of processes that must receive and respond to a message. RHODOS November 28, 1997 Page 10 Majority Response: The sending process expects to receive a majority a responses from the current member of the current group of receiving ....

....project into the design and development of distributed operating system and distributed systems. The Amoeba group communication protocol is part of the interprocess communications facility and the current implementation is provided within the kernel of the Amoeba distributed operating system [Kaashoek and Tanenbaum 1994]. Amoeba provides RPC and group communications as the basic interprocess communication primitives. The Amoeba implementation utilises an efficient broadcast protocol as the basic group communication mechanism [Kaashoek et al. 1990] The broadcast protocol utilises the data link broadcast ....

M. F. Kaashoek, and A. S. Tanenbaum., "Efficient Reliable Group Communication for Distributed System," Dept. of Mathematics and Computer Science Technical Report, Vrije Universiteit, Amsterdam.

....of a message) a single response from one member of the group; k responses: The client process expects to obtain k responses for the delivered message The Client Server Model and Systems 47 from the members of the process group. By using k response semantics the group resilience can be defined [Kaashoek and Tanenbaum 1994]. The resilience of a group is based on the minimum number of processes that must receive and respond to a message; Majority Response: The client process expects to receive a majority of responses from the members of the process group; Total Response: The client process requires all members ....

M. F. Kaashoek, and A. S. Tanenbaum., "Efficient Reliable Group Communication for Distributed System," Dept. of Mathematics and Computer Science Technical Report, Vrije Universiteit, Amsterdam.

....parameters and hide communications details from the application programmer. Amoeba uses the traditional client server paradigm, with all objects being managed by object servers to which object invocation requests are sent. Reliable, totally ordered group communication is also provided in Amoeba [52] and is used by the object servers. Amoeba has its own communication protocol, FLIP [51] which operates on specialised hardware. It was designed with both local area network (LAN) and wide area network (WAN) communications in mind. Amoeba also provides a TCP IP server for communicating with more ....

M. Frans Kaashoek and Andrew S. Tanenbaum. Efficient Reliable Group Communication for Distributed Systems. (submitted for publication, 1994), 1994.

....choosing a primary replica is like electing a leader) an efficient form of passive replication. It is also closely related to group communication [Pow96] which (among other uses) provides a powerful basis for implementing active replication. For example, the group communication system in Amoeba [KT91, KT92] uses Garcia Molina s Invitation Algorithm [GM82] to reconfigure a group after crashes. As another example, the group membership algorithms in Horus [FvR95, vRBM96] and Ensemble [Hay97] can be seen as a combination of Garcia Molina s Bully Algorithm (for handling crashes) and Invitation Algorithm ....

M. Frans Kaashoek and Andrew S. Tanenbaum. Efficient reliable group communication for distributed systems. Rapport IR-295 IR-295, Faculteit Wiskunde en Informatica, Vrije Universiteit, 1992. Revised version available from ftp://ftp.cs.vu.nl/pub/papers/amoeba/group94.ps.Z.

....algorithm in a shared whiteboard. Of these three protocols, the multicast NAK protocol was found to have the highest throughput; Hall [9] summarizes a number of applications that include multicast services, including VMTP [4] XTP [26] ISIS [12] MTP [20] MTP2 [2] RAMP [18] and Ameoba [14]. Of particular relevance, Huang, et al. 3] developed a reliable multicast transport service for an Asynchronous Transfer Mode (ATM) network. This service sits atop the unreliable ATM Adaption Layer 5 (AAL5) and uses the vendor supported Application Programming Interface (API) s socket like ....

M. Kaashoek and A. S. Tanenbaum. Efficient Reliable Group Communication for Distributed Systems. (Submitted for publication 1994).

....token has to be transferred L times after the message transmission, and L token holder need to acknowledge the message. This scheme introduces a very long delay before a message can be delivered. It requires also between two and three broadcast messages per atomic broadcast. Kaashoek and al. [50] proposed another type of sequencer based protocol for the Amoeba operating system. A unique broadcast server, the sequencer, imposes a total ordering on all updates originating from all broadcast servers. Their protocol differs from Chang in that the sequencer is fixed. If the sequencer goes ....

M. F. Kaashoek and A. S. Tanenbaum. Efficient reliable group communication for distributed systems. submitted for publication, 1994.

....token has to be transferred L times after the message transmission, and L token holders need to acknowledge the message. This scheme introduces a very long delay before a message can be delivered. It requires also between two and three broadcast messages per atomic broadcast. Kaashoek and al. [52] proposed another type of sequencer based protocol for the Amoeba operating system. A unique broadcast server, the sequencer, imposes a total ordering on all updates originating from all broadcast servers. Their protocol differs from Chang in that the sequencer is fixed. If the sequencer goes ....

M. F. Kaashoek and A. S. Tanenbaum. Efficient reliable group communication for distributed systems. submitted for publication, 1994.

No context found.

Kaashoek, M.Frans, and, Tanenbaum, S. Andrew, "Efficient Reliable Group Communication for Distributed Systems" Dept. of Math and Computer Science, Vrije Universiteit, 1993.

No context found.

M.Frans Kaashoek and Andrew S. Tanenbaum. Efficient Reliable Group Communication for Distributed Systems. The paper is a subset of M. F. Kaashoek's Ph.D. thesis: Group Communication in Distributed Computer Systems, Vrije Universiteit, Amsterdam, 1992.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC