Constant-time snapshots with applications to concurrent data structures

Wei, Yuanhao; Ben-David, Naama; Blelloch, Guy E.; Fatourou, Panagiota; Ruppert, Eric; Sun, Yihan

doi:10.1145/3437801.3441602

Cited by 19 publications

(27 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An important open problem is to generalize our algorithm to a multi-writer version, or to allow snapshots of stronger primitives, such as CAS objects (similar to [21]).…”

Section: Discussionmentioning

confidence: 99%

“…A more flexible multi-writer snapshot specification was proposed by Wei, Ben-David, Blelloch, Fatourou, Ruppert and Sun [21]. Their object allows a process to take a snapshot of multiple CAS objects, and returns a handle to that snapshot.…”

Section: Related Solutionsmentioning

confidence: 99%

“…As far as we know, our algorithm is the only solution to the partial snapshot problem, where snapshots of o(n/log n) components can be obtained in o(n) time. A comparison of our implementation with the partial snapshot objects of [8,21] is shown in Table 1.…”

Section: Related Solutionsmentioning

confidence: 99%

See 2 more Smart Citations

An Efficient Adaptive Partial Snapshot Implementation

Bashari

Woelfel

2021

Proceedings of the 2021 ACM Symposium on Principles of Distributed Computing

View full text Add to dashboard Cite

The standard single-writer snapshot type allows processes to obtain a consistent snapshot of an array of n memory locations, each of which can be updated by one of n processes. In almost all algorithms, a Scan() operation returns a linearizable snapshot of the entire array. Under realistic assumptions, where hardware registers do not have the capacity to store many array entries, this inherently leads to a step complexity of Ω(n).In this paper, we consider an alternative version of the snapshot type, where a Scan() operation stores a consistent snapshot of all n memory locations, but does not return anything. Instead, a process can later observe the value of any component of that snapshot using a separate Observe() operation. This allows us to implement the type from fetch-and-increment and compare-and-swap objects, such that Scan() operations have constant step complexity and Update() and Observe() operations have step complexity O(log n).

show abstract

“…An important open problem is to generalize our algorithm to a multi-writer version, or to allow snapshots of stronger primitives, such as CAS objects (similar to [21]).…”

Section: Discussionmentioning

confidence: 99%

Section: Related Solutionsmentioning

confidence: 99%

See 1 more Smart Citation

An Efficient Adaptive Partial Snapshot Implementation

Bashari

Woelfel

2021

Proceedings of the 2021 ACM Symposium on Principles of Distributed Computing

View full text Add to dashboard Cite

show abstract

“…Wei et al [33] describe a technique for implementing a Scan operation for a collection of CAS objects. The technique involves augmenting each CAS object with an unbounded version list.…”

Section: Related Workmentioning

confidence: 99%

“…Hence, these implementations are impractical for scannable objects with many components, even if the components are small. Other implementations of scannable objects either use unbounded sequence numbers [1,9,15,20], or require significantly more than base objects [5,8,24,33,36]. It is natural to ask how much extra space (i.e.…”

Section: Introductionmentioning

confidence: 99%

The Space Complexity of Scannable Binary Objects

Ovens

2021

Proceedings of the 2021 ACM Symposium on Principles of Distributed Computing

View full text Add to dashboard Cite

Determining a consistent view of the contents of a shared array while its components are being modified is a fundamental task in distributed algorithm design. Snapshot objects address this problem when read/write registers are components of the array. A natural generalization of a snapshot object is a scannable object, an object with multiple components that each support and some other operations. Scannable objects also support the operation, which provides an instantaneous view of all the object's components. This raises the question: How many instances of objects in some set O are required to implement a scannable object whose components are in O?In this paper, we consider scannable binary objects, whose components are arbitrary 1-bit objects. If each component is a test-and-set object, there is a simple implementation using test-and-set objects. However, when each component is non-monotonic (i.e. can change from 0 to 1 and from 1 to 0), we prove that more objects are needed. Specifically, when ≤ 2 −1 + 2, we show a lower bound of + − 2 base objects, even for obstruction-free, single-updater implementations. When ≥ 2 − 2, we prove that 2 − 1 base objects are required for obstruction-free, single-updater implementations. When 2 −1 + 2 < < 2 − 2, we show a gradual transition between these two lower bounds. We show that there is a lockfree implementation of an -process, scannable binary object with components from + instances of 1-bit base objects, nearly matching our lower bound when ≤ 2 −1 + 2. There is a known wait-free, single-updater implementation that uses 2 base objects, nearly matching our lower bound when ≥ 2 − 2. CCS CONCEPTS• Theory of computation → Distributed algorithms.

show abstract