We present a solution to the problem of information integrity protection in distributed systems which is robust against malicious parties, is space and communication efficient, and uses cryptography in a minimal way. ...
详细信息
ISBN:
(纸本)0897916131
We present a solution to the problem of information integrity protection in distributed systems which is robust against malicious parties, is space and communication efficient, and uses cryptography in a minimal way. Our solution builds on Rabin's information dispersal algorithm (IDA). While the IDA scheme is able to deal with missing pieces of information, here we solve the more general secure information dispersal problem, in which recovery of information is possible even against modification of information shares by a possibly malicious adversary. Previous solutions to this problem suffer from space and communication blowup, or use costly cryptographic tools that limit the usefulness of the scheme. In contrast, our scheme uses cryptography in a `minimal' way. It gets rid of the need of private and public key systems and, actually, requires no secret keys at all. It permits recovery of the distributed information by any party in the system, and at the same time prevents any modification or loss of information, as long as a honest majority of parties exist. The proposed solution is space optimal and flexible enough to replace the basic IDA algorithm in most applications that contemplate general faults. Our solution introduces a new cryptographic tool called distributed fingerprints, which consists of public fingerprints for data integrity having the `paradoxical' property that everyone in the system can compute them (using the same function and no secrets!) but no one can forge them. distributed fingerprints may replace some of the (integrity) functions provided by signatures in distributed systems, but at a lower cost.
This paper investigates a new shared-memory model called immediate atomic snapshot memory. It is an extension of atomic snapshot memory in which a write operation in addition to writing, also returns an atomic snapsho...
详细信息
ISBN:
(纸本)0897916131
This paper investigates a new shared-memory model called immediate atomic snapshot memory. It is an extension of atomic snapshot memory in which a write operation in addition to writing, also returns an atomic snapshot of the memory. Unlike regular atomic snapshot, immediate snapshot is guaranteed to closely follow the write operation. This model was previously used to obtain an impossibility result. Here we investigate its utility for the design of algorithms. We first implement the one-shot version in the read-write model. We then use the model to design a new renaming algorithm. The renaming algorithm we obtain is simple and requires at worst n cycles of one-shot immediate snapshots. Since our implementation of the one-shot immediate snapshot requires O(n2) primitive read-write operations, we obtain an O(n3) renaming algorithm. Currently the best renaming algorithm is exponential. Our implementation of the immediate snapshot relies on a priori knowledge of the bound on the number of snapshots to be taken. We were not able to dispense with this condition and thus were unable to obtain an implementation of the long-lived object.
The problem of achieving optimal clock synchronization in a communication network with arbitrary topology and perfect clocks (that do not drift) is studied. A novel modular presentation of the problem is described whi...
详细信息
ISBN:
(纸本)0897916131
The problem of achieving optimal clock synchronization in a communication network with arbitrary topology and perfect clocks (that do not drift) is studied. A novel modular presentation of the problem is described which allows to deal with different assumptions for the delay of messages. We present a definition of clock synchronization under arbitrary delay assumptions, and present an optimal clock synchronization algorithm for general systems. We then show that in local systems (where delays on each link are independent of the other links) the inputs for the clock synchronization algorithm can be computed from the maximum local shifts for each pair of processors sharing a link. The maximum local shift for two processors depends only on their views. This allows our theory to deal with systems where different links adhere to different assumptions, or the same link satisfies several sets of assumptions;such mixtures are quite likely in practice. In particular, we show how to compute the maximum local shifts from the views, and hence provide optimal algorithms for systems where some links may have upper and/or lower bounds on the delay, some may have a bound on the difference between the delay in both directions, some may have both kinds of bounds and some may have no bounds. Previous results dealt only with the case where upper and lower bounds were known for all links. We introduce a new notion of optimality, that requires an algorithm to achieve the best possible precision on each instance;this notion is stronger than the previously used notion of worst case optimality. In contrast to the worst case approach, the new notion handles models where the worst-case behavior of any clock synchronization algorithm is inherently unbounded.
We describe computation migration, a new technique that is based on compile-time program transformations, for accessing remote data in a distributed-memory parallel system. In contrast with RPC-style access, where the...
详细信息
ISBN:
(纸本)0897915895
We describe computation migration, a new technique that is based on compile-time program transformations, for accessing remote data in a distributed-memory parallel system. In contrast with RPC-style access, where the access is performed remotely, and with data migration, where the data is moved so that it is local, computation migration moves part of the current thread to the processor where the data resides. The access is performed at the remote processor, and the migrated thread portion continues to run on that same processor;this makes subsequent accesses in the thread portion local. We describe an implementation of computation migra^ tion that consists of two parts: an implementation that migrates single activation frames, and a high-level language annotation that allows a programmer to express when migration is desired. We performed experiments using two applications;these experiments demonstrate that computation migration is a valuable alternative to RPC and data migration.
The symposium materials contain 25 papers. The topics covered include distributed priority algorithms, connection-based communication in dynamic networks, adaptive packet routing, computing with faulty shared memory, ...
详细信息
ISBN:
(纸本)0897914953
The symposium materials contain 25 papers. The topics covered include distributed priority algorithms, connection-based communication in dynamic networks, adaptive packet routing, computing with faulty shared memory, fault-tolerant coordination, self-stabilization, shared-memory multiprocessors, fast mutual exclusion, fast network decomposition, the weakest failure detector for solving consensus, leader election in complete networks, randomized coordinated attack protocols, distributed edge coloring, and proving probabilistic correctness statements.
This paper addresses problems which arise in the synchronization and coordination of distributed systems which employ unreliable shared memory. We present algorithms which solve the consensus problem, and which simula...
详细信息
ISBN:
(纸本)0897914953
This paper addresses problems which arise in the synchronization and coordination of distributed systems which employ unreliable shared memory. We present algorithms which solve the consensus problem, and which simulate reliable shared-memory objects, despite the fact that the available memory objects (e.g. read/write registers, test-and-set registers, read-modify-write registers) may be faulty.
Analysis and design of distributed algorithms and protocols are difficult issues. An important cause for those difficulties is the fact that the logical structure of the solution is often invisible in the actual imple...
详细信息
ISBN:
(纸本)0897914953
Analysis and design of distributed algorithms and protocols are difficult issues. An important cause for those difficulties is the fact that the logical structure of the solution is often invisible in the actual implementation. We introduce a framework that allows for a formal treatment of the design process, from an abstract initial design to an implementation tailored to specific architectures. A combination of algebraic and axiomatic techniques is used to verify correctness of the derivation steps. This is shown by deriving an implementation of a distributed minimum weight spanning tree algorithm in the style of [GHS].
We determine what information about failures is necessary and sufficient to solve Consensus in asynchronous distributed systems subject to crash failures. In [CT91], we proved that ♦W, a failure detector that provides...
详细信息
ISBN:
(纸本)0897914953
We determine what information about failures is necessary and sufficient to solve Consensus in asynchronous distributed systems subject to crash failures. In [CT91], we proved that ♦W, a failure detector that provides surprisingly little information about which processes have crashed, is sufficient to solve Consensus in asynchronous systems with a majority of correct processes. In this paper, we prove that to solve Consensus, any failure detector has to provide at least as much information as ♦W. Thus, ♦W is indeed the weakest failure detector for solving Consensus in asynchronous systems with a majority of correct processes.
Certain types of routing, scheduling and resource allocation problems in a distributed setting can be modeled as edge coloring problems. We present fast and simple randomized algorithms for edge coloring a graph, in t...
详细信息
ISBN:
(纸本)0897914953
Certain types of routing, scheduling and resource allocation problems in a distributed setting can be modeled as edge coloring problems. We present fast and simple randomized algorithms for edge coloring a graph, in the synchronous distributed point-to-point model of computation. Our algorithms compute an edge-coloring of a graph G with n nodes and maximum degree Δ with at most (1.6+Ε)Δ+log2+δn colors with high probability (arbitrarily close to 1), for any fixed Ε, δ>0. To analyze the performance of our algorithms, we introduce new techniques for proving upper bounds on the tail probabilities of certain random variables. Chernoff-Hoeffding bounds are fundamental tools that are used very frequently in estimating tail probabilities. However, they assume stochastic independence among certain random variables, which may not always hold. Our results extend the Chernoff-Hoeffding bounds to certain types of random variables which are not stochastically independent. We believe that these results are of independent interest, and merit further study.
暂无评论