H. Kopetz and W. Merker. The architecture of MARS. In Proceedings of the Fifteenth Symposium on Fault-Tolerant Computing, pages 274--279, Ann Arbor, Mi, Jun 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....communication services have proved to be extremely useful for constructing highly available and dependable distributed applications. Because of their usefulness, a large number of group communication services have been proposed, designed, and implemented. Notable ones include Isis [8, 9] Mars [35, 34], Delta 4 [44] AAS [6, 16] Transis [3] Consul [39] Totem [4] Horus [50] Pinwheel [19] Spread [5, 2] and Bimodal [7] Most of these services have been implemented for an asynchronous distributed computing environment, and do not provide any timeliness guarantees. Specification of these ....

H. Kopetz and W. Merker. The architecture of MARS. In Proceedings of the Fifteenth Symposium on Fault-Tolerant Computing, pages 274--279, Ann Arbor, Mi, Jun 1985.

....failures and joins, so to satisfy the above safety and timeliness properties. Unlike Cristian s protocols, which do not attempt to order membership changes with respect to application messages, the synchronous membership protocol proposed in [Kopetz et al. 1991] implemented in the Mars system [Kopetz Merker 1985, Kopetz et al. 1989] realizes ordering with respect to application messages. This is due to the way in which membership is integrated into the communication system, built on a broadcast bus that is accessed using a time division multiple access (TDMA) strategy based on common global time ....

H. Kopetz and W. Merker, "The Architecture of Mars", in 15th Int. Conf. on Fault-Tolerant Computing, (Ann Arbor, Michigan), pp.274-9, IEEE Computer Society Press, 1985.

....be enforced by all components, based on their local information. In a completely static system, a periodic global calendar is available and each component has its relevant entries referring to its activities in a global time scale. A global activity may only be started according to this schedule (Kopetz Merker, 1985; Kopetz GruK nsteidl, 1994) In a more dynamic system where hard real time, soft real time, and non real time tasks coexist, things are more complicated. If a computing resource is free, a less critical (soft or non real time) task may start computation and request resources. In this case, it ....

Kopetz, H., & Merker, W. (1985). The architecture of MARS. Proceedings of the 15th international symposium on fault tolerant computing (pp. 274}279). Ann Arbor, MI.

....properties are complemented by the upper layer protocol, however, Ethernet can be used for the distributed real time systems. For example, MARS, a distributed real time operating system, attempted to support predictability for Ethernet by implementing TDMA(Time Division Multiple Access) [2,3]. In general TDMA protocol consists of slots and frames where each node is assigned one slot per frame. When TDMA is implemented, the slot time and frame time are determined to be a fixed value, often irrelevant to the characteristics of message streams. As a result, the waste of bandwidth is ....

Hermann Kopetz and W. Merker, "The Architecture of MARS," Proc. 15-th Fault-Tolerant Computing Symposium, pp.274-279, June 1985.

....involves extending the system model of REACT to represent (and analyze) other architectures. REACT currently evaluates the dependability of tightly coupled multiprocessor systems. Although many fault tolerant architectures are multiprocessor based, several have been realized as distributed systems [68, 71, 90] which cannot be modeled within the existing framework of REACT. Given the advantages of distributed computing coupled with recent advances in networking technology, one would expect a growing number of distributed systems to be employed in life and cost critical applications in the near future. ....

Kopetz, H. and Merker, W., "The architecture of MARS," in Proceedings of the 15th International Symposium on Fault-Tolerant Computing, pp. 274-- 279, IEEE, June 1985.

....only a special case of the more general problem to schedule a cooperative distributed computation. There exist several alternative approaches to solve this problem based on the assumptions about the behaviour of the system and the environment. The first approach, known as time triggered approach [[18]] 21] 19] assumes a complete knowledge of all future actions of the system. Hence, during operation these systems exhibit minimal overhead paired with a maximum of predictability and are highly appropriate for safety critical applications which can be modelled by a periodic behaviour. A ....

H. Kopetz and W. Merker: "The Architecture of MARS", IEEE Proceedings of the 15 th Fault Tolerant Computing Systems Symposium, 1985.

....to meet hard deadlines even in worst case anticipated fault and load scenarios. Since resource reservation is based on pessimistic assumptions, reserved resources are often unused. Moreover, there are usually some non reserved resources remaining. In a strictly time triggered system like MARS [4], 1] it is not possible to exploit system resources, without reserving them by the offline scheduler. In contrast, a dynamic system, e.g. MARUTI [8] can achieve optimal resource utilization by Earliest Deadline First (EDF) scheduling mechanism. In a complex real time application, like ....

....hard real time is unnecessary, and may lead to bandwidth shortage. In a completely static system, a global calendar is available and each node has its relevant entries referring to its message transmission times in a global time scale. A message may only be transmitted according to this schedule [4], 1] 6] In a more dynamic system where hard real time, soft real time, and non real time messages coexist, things are more complicated. If the bus is free, a soft real time message may be sent. In this case, it must be guaranteed that it does not cause a timing failure of a hard real time ....

[Article contains additional citation context not shown here]

H. Kopetz and W. Merker, "The Architecture of MARS", Proc. of 15th Fault Tolerant Computing Symposium, pp. 274-279, Ann Arbor, Michigan, 1985.

....a high resolution global time base, each node knows exactly when to send or receive a message. However, due to its static nature, the TDMA approach is not suitable for dynamic real time systems. Moreover, the TDMA protocol is applied preferrably on highbandwidth busses, like Ethernet or FDDI [7][11]. The deadline monotonic priority assignment [16] achieves meeting deadlines as guaranteed by an off line feasibility test for a static system with periodic tasks. Although static systems can be scheduled easily by this approach, it does not allow scheduling of dynamic systems, where an offline ....

....has to be enforced by all participants, based on their local information. In a completely static system, a global calendar is available and each participant has its relevant entries referring to its activities in a global time scale. A local activity may only be started according to this schedule[11]. In a more dynamic system where critical, essential, soft real time and non real time tasks coexist, things are more complicated. If a computing resource is free, a less critical task may start computation and request resources. In this case, it must be guaranteed that it does not block an ....

H. Kopetz and W. Merker: "The Architecture of MARS", IEEE Proceedings of the 15 th Fault Tolerant Computing Systems Symposium, 1985.

....service is realized using logical clocks [Lam78] and is used by multicast to consistently order messages. Finally, we note that, while this division of functionality into abstract services is somewhat arbitrary, a large number of systems use these services or variants thereof [BJ87, BSS91, CDD90, KM85, KDK 89, PSB 88] Further discussion of these services, their interrelationships, and the systems that use them can be found in [MS92] We now turn our attention from the abstract services provided by Consul to the set of protocol modules that realize these services in the substrate. A ....

H. Kopetz and W. Merker. The architecture of MARS. In Proceedings of the Fifteenth Symposium on Fault-Tolerant Computing, pages 274--279, Ann Arbor, Mi, Jun 1985.

....design of the protocol. In contrast, other protocols wait until all the processes have formed their sf groups before removing the failed process. 7.3 Fault tolerant Systems We compare the system design of Consul with some of the recent fault tolerant systems being developed. These include MARS [28, 26], AAS [12] DELTA 4 [39] and ISIS [6] Both MARS and AAS are distributed real time systems that employ synchronized clocks to implement various fault tolerant services provided by the system. MARS is a system designed for distributed real time process control applications, while AAS is designed ....

H. Kopetz and W. Merker. The architecture of mars. In 15th Annual Symposium on Fault-Tolerant Computing, pages 274--279, Ann Arbor, Michigan, Jun 1985.

....distributed systems that have been designed and or implemented over the past decade. Here, we briefly outline some representative examples: the Advanced Automation System (AAS) BDD 89, CDD90] Argus [LS83, Lis88] Consul [MPS91] Delta 4 [PSB 88] ISIS [BJ87, BSS91] and MARS [KM85, KDK 89] our specific emphasis is on describing how each fits into the framework developed above. Others systems of interest include ADS [IM84] ANSA [Tea91, OOW91] Arjuna [SDP91] Avalon [DHW88] Chorus [BFG 85] and Clouds [LW85] 10.1 Advanced Automation System AAS is a ....

H. Kopetz and W. Merker. The architecture of MARS. In Proceedings of the Fifteenth Symposium on Fault-Tolerant Computing, pages 274--279, Ann Arbor, Mi, Jun 1985.

....using logical clocks [9] and is used by multicast to consistently order messages. Finally, it should be emphasized again that these services are widely recognized as fundamental to the construction of fault tolerant distributed systems, with variants being used in a large number of systems [10, 3, 4, 11, 12, 13]. The dependencies are fundamental as well since they are induced by the properties of the services and not by the specific way in which they are realized in Consul. Further discussion of fault tolerant services, their interrelationships, and the systems that use them can be found in Reference ....

H. Kopetz and W. Merker, "The architecture of MARS," in Proceedings of the Fifteenth Symposium on Fault-Tolerant Computing, (Ann Arbor, Mi), pp. 274--279, Jun 1985.

....The methods of N modular redundancy and N version programming are special cases of this class of fault tolerance methods. Examples of systems using active replication include the new ISIS system (ISIS 2 ) Birman87] Circus [Cooper85, Cooper84] CHORUS [Banino85, Banino82] MP [Gait85] MARS [Kopetz85b], FT Concurrent C [Cmelik88] PRIME [Fabry73] SIFT [Wensley78] and Yang and York s work on the Intel iAPX 432 [Yang85] These methods are well suited for use in real time systems, since failure recovery is essentially immediate. However, this ability requires extra processors to be dedicated to ....

H. Kopetz and W. Merker. The architecture of MARS. In The Fifteenth Annual International Symposium on Fault-Tolerant Computing: Digest of Papers, pages 274--279. IEEE Computer Society, June 1985.

....membership, or transactions. Here, we examine the basic outline of some of these systems, without addressing in detail any of the particular services provided. Since the emphasis is on configurable systems, only two static systems are described: Isis [Bir85a, BJ87, BC91, BSS91, BR94] and Mars [KM85, KDK 89, KG94] The configurable systems presented illustrate different approaches taken towards customization in fault tolerant computing. A large number of other projects are not addressed in detail here, including ADS [IM84] Amoeba [RST89, KT91] AMp and xAMp [VRB89, RV91, RV92] Argus ....

....after the computation is finished. The services provided by Isis are not configurable, but the system includes a collection of multicast services with different ordering and reliability semantics. 2.1.1. 2 Mars Mars is a system for building fault tolerant distributed real time applications [KM85, KDK 89, KG94] Mars is targeted for hard real time applications, where missing a deadline can be catastrophic. Therefore, Mars places extremely strict requirements on the timeliness of communication and task execution, as well as the overall reliability of the system. Mars provides various ....

H. Kopetz and W. Merker. The architecture of Mars. In Proceedings of the 15th Symposium on Fault-Tolerant Computing, pages 274--279, Ann Arbor, MI, Jun 1985.

....in that period. Amidst all these research activities, the work on the TimeTriggered Architecture (TTA) started in 1979 at the Technical University of Berlin with the MARS project. A first report on the MARS project [19] appeared in 1982 and was later published at the IEEE FTCS conference in 1985 [20]. After 1982, different versions of the MARS architecture have been implemented at the Vienna University of Technology [21] 22] and it became clear that a hardware supported fault tolerant clock synchronization is a fundamental building block of a timetriggered architecture. At about that time ....

H. Kopetz and W. Merker. The Architecture of MARS. In Proceedings of the 15th International Symposium on Fault-Tolerant Computing, pages 274--279, 1985.

No context found.

H. Kopetz and W. Merker, `The architecture of MARS', Proceedings of the Fifteenth Symposium on Fault-Tolerant Computing, Ann Arbor, MI, June 1985, pp. 274--279.

No context found.

Kopetz, H. and Merker, W., "The Architecture of MARS", Proceedings 15th FTCS, June 1985, pp. 274-279.

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