Finding consistent global checkpoints of a distributed computation is important for analyzing, testing, or verifying properties of these computations. In this paper we present a theoretical foundation for finding consistent global checkpoints. Given an arbitrary set S of local checkpoints, we prove exactly which sets of other local checkpoints can be combined with S to build consistent global checkpoints, and we present an algorithm for finding all such global checkpoints. The minimal and maximal consistent global checkpoints are presented as special cases. The results are based on the notion of zigzag paths introduced by Netzer and Xu [14]. We also present a method for finding zigzag paths using the rollback-dependency graph introduced by Wang [17, 16]. Keywords: causality, distributed checkpointing, consistent global checkpoint collection, failure recovery, fault tolerance. 1 Introduction Consistency of global states is a recurring theme in distributed systems [10]. A globa...

We introduce concurrent time-stamping, a paradigm that allows processes to temporally order concurrent events in an asynchronous shared-memory system. Concurrent time-stamp systems are powerful tools for concurrency control, serving as the basis for solutions to coordination problems such as mutual exclusion, l-exclusion, randomized consensus, and multiwriter multireader atomic registers. Unfortunately, all previously known methods for implementing concurrent timestamp systems have been theoretically unsatisfying since they require unbounded-size time-stamps -- in other words, unbounded-size memory. This work presents the first bounded implementation of a concurrent time-stamp system, providing a modular unbounded-to-bounded transformation of the simple unbounded solutions to problems such as those mentioned above. It allows solutions to two formerly open problems, the boundedprobabilistic -consensus problem of Abrahamson and the fifo-l-exclusion problem of Fischer, Lynch, Burns and...

Time and message complexities are measures commonly used to characterize distributed computations. However, these measures only address quantitative aspects of the computation. This paper presents several qualitative measures which quantify the degree of concurrency in a distributed computation by analyzing its synchronization constraints. We introduce two abstractions, called cone and cylinder, and define concurrency measures using these two abstractions. Then, we present simple ways to compute these measures; therefore, these measures can be easily incorporated into any system to analyze distributed executions. 1 Introduction Time and space complexities are the quantitative measures generally used to characterize the efficiency of sequential algorithms and programs. In the field of parallel applications, numerical computations in particular, the notion of "speed-up" has been introduced to measure the performance gain obtained by the parallelism with respect to the best possible seq...

Mutual exclusion is a topic that Leslie Lamport has returned to many times throughout his career. This article, which is being written in celebration of Lamport's sixtieth birthday, is an attempt to survey some of his many contributions to research on this topic.

. This paper examines synchronization support of two generations of SCI adapters from Dolphin Interconnect Solutions and compares the functionality to similar support on Digital's Memory Channel. Memory Channel enforces sequential consistency across the interconnect, while SCI allows store reordering. This gives SCI a potential performance payoff by allowing more flexible pipelining of data through the interconnect. The lower number of ordering constraints also reduces hardware complexity, but moves the complexity to software. For a straightforward implementation of message passing this overhead is significant. A new software algorithm, the valid flag algorithm, is introduced to improve this situation. A new hardware lock support mechanism is proposed to facilitate efficient locks in absence of lock support on the I/O bus. Performance of the simple message passing protocol is compared to the suggested valid flag protocol. The valid flag protocol reduces latency of a small message by 50...

This paper presents the RatchetFT distributed faulttolerant mutual exclusion algorithm for processor rings. RatchetFT is self-stabilizing, in that if mutual exclusion is lost due to any sequence of on-line failures and repairs of processors, mutual exclusion will eventually be regained. This research demonstrates that self-stabilization can be achieved in the presence of faulty processors, provided that these faulty processors always appear to behave incorrectly. Self-stabilization is achievable even if faulty processor behavior is not restricted to transient failures or other simple failure models. The key results of the paper include the specification of RatchetFT and a detailed sketch of its correctness proof. 1 Introduction Self-stabilizing systems have primarily been studied for fault-free environments [8,10,5,14] in which, given any arbitrary initial state, a self-stabilizing system will eventually reach, and remain in, a legitimate state. Transient failures affecting either sy...

this paper to exemplify steps that need to be taken in order to overcome this deficit. We first discuss choices that need to be made when designing a suitable real-time execution model for SDL and Estelle and proceed to present two remedies to the inexpressiveness problem: First, we introduce the concept of complementary real-time specification by reconciling the semantic models of Metric Temporal Logic and SDL and showing how both languages can be used in a complementary fashion. Second, we suggest a language extension and the corresponding semantic interpretation for Estelle. While we present examples from the domain of multimedia and broadband systems, the applicability of our specification methods extends to hard real-time systems. Finally, we discuss extensions of our techniques to capture QoS stochastic properties, and we allude to formal requirements verification and automatic implementation based on our techniques. Condensed version to appear in Computer Networks and ISDN Systems. 21 August 1997 1 Introduction The specification of requirements on the observable behavior of distributed, communicating real-time system is an important step in the engineering of these systems. Requirements specifications help avoiding inconsistencies in the requirements, they are the basis for deriving correct system designs, they are essential in establishing the system's correctness by serving as a basis for testing and formal verification, and they are important in the documentation of system requirements. [20] suggests that proper requirements engineering, of which requirements specification is an important step, is pivotal in avoiding pitfalls of what is called the `software crisis'. Facets of the software crisis include systems that have low reliability or even unusability a...

this paper. Motivated by their first work Moir and Anderson developed renaming algorithms, in the read/write model, when such a bound on the maximum number of processes is known in advance. This led to a sequence of works on the renaming problem in this model [MA95, MG96, BGHM95] that lead to a long-lived (2K \Gamma 1)-renaming algorithm with O(K ) step complexity and O(K space complexity [Moi98]. These works employed various variants of the splitter building block which is a descendant of Lamport's adaptive mutual exclusion algorithm, however the last one [Moi98] depends on an additional work which is the first long-lived renaming algorithm by Burns and Peterson [BP89]

ion of Computation Actions ............................ 3 1.1.2 Program and its Execution ................................................... 5 1.2 The Debugging Process .............................................................. 6 1.2.1 Block Triangular Solver Example ....................................... 7 1.2.2 Different Parts of the Debugging Process............................ 9 1.3 Overview of the Unified Debugger .......................................... 12 1.3.1 The Actual Execution Behavior......................................... 13 1.3.2 Restricting Execution to Selected Actions......................... 14 1.3.3 Predicate/Model of the Expected Behavior ....................... 14 1.3.4 Automatic Checking of the Expected Behavior ................ 15 1.3.5 Display of the Unexpected Behavior ................................. 15 1.3.6 Cyclical Debugging ........................................................... 16 1.3.7 Support for Interactive Debugging ......................

60dB 4.20 512 96 ETSI-A ETSI-A ETSI-1 60dB 4.20 1536 512 AWGN-140 AWGN-140 Draft Recommendation G.992.2 140 14 T1.601 #9 1536kbps 256kbps 49 Annex A G.992.2 15 T1.601 #9 1536kbps 256kbps 24 DSL 16 Shortened T1.601#7 1536kbps 256kbps 24 HDSL Table 47. Extended Reach Test Cases NOTE1: A goal of future enhancements of this Recommendation is to make the "Extended Reach Cases" mandatory. NOTE2: Performance levels do not reflect the effect of customer premise wiring, which is expected to reduce data rate.G.992.2G.992.2G.992.2 Draft Recommendation G.992.2 139 ANNEX D D.1 System Performance for North America All test loops specified in this section shall be used for G.992.2 and testing shall confirm to the following: . No power cutback on upstream transmitter. . Margin=4 dB . BER=10 -7 . Background noise = -140 dBm/Hz . Rates, except where noted,