Skip Hash: A Fast Ordered Map Via Software Transactional Memory

Abstract: Scalable ordered maps must ensure that range queries, which operate over many consecutive keys, provide intuitive semantics (e.g., linearizability) without degrading the performance of concurrent insertions and removals. These goals are difficult to achieve simultaneously when concurrent data structures are built using only locks and compare-and-swap objects. However, recent innovations in software transactional memory (STM) allow programmers to assume that multi-word atomic operations can be fast and simple. This paper introduces the skip hash, a new ordered map designed around that assumption. It combines a skip list and a hash map behind a single abstraction, resulting in $O(1)$ overheads for most operations. The skip hash makes use of a novel range query manager -- again leveraging STM -- to achieve fast, linearizable range queries that do not inhibit scalability. In performance evaluation, we show that the skip hash outperforms the state of the art in almost all cases. This places the skip hash in the uncommon position of being both exceedingly fast and exceedingly simple.

• The paper introduces Skip Hash, integrating STM to achieve O(1) overhead for most operations in a concurrent ordered map.
• The paper presents a novel range query management system that ensures fast, linearizable queries using global versioning.
• The paper demonstrates that Skip Hash outperforms traditional lock-based methods, achieving high throughput in lookup and mixed-operation workloads.

Skip Hash: A Fast Ordered Map Via Software Transactional Memory

The paper presents the "skip hash," an innovative ordered map structure that combines a skip list with a hash map, leveraging modern software transactional memory (STM) to enhance performance and scalability. This work addresses the challenges posed by concurrent operations on ordered maps, particularly focusing on achieving fast, linearizable range queries without disabling the scalability of insertions and deletions.

Key Contributions

The skip hash employs STM to facilitate multi-word atomic operations efficiently, leading to significant performance improvements over traditional implementations that rely solely on locks and compare-and-swap techniques. The following are the main contributions of the paper:

1. STM Integration: The skip hash employs STM for synchronization, enabling atomic operations across both the skip list and hash map. This integration results in $O(1)$ overheads for most operations, a considerable enhancement over conventional methods.
2. Range Query Management: A novel range query manager is introduced to ensure fast, linearizable range queries. The manager maintains a global versioning system to facilitate transactions and prevent conflicts, often occurring in high-contention environments.
3. Performance Evaluation: Empirical results demonstrate that the skip hash consistently outperforms state-of-the-art concurrent ordered maps. The optimization employed allows the skip hash to achieve higher throughput, exhibiting both rapid execution and simplicity in design.

Numerical Results and Claims

The paper supports its claim with strong numerical results, showing that the skip hash not only compares favorably with existing solutions but frequently surpasses them regarding speed and scalability:

• In scenarios composed entirely of lookup operations, the skip hash achieves more than twice the throughput of other state-of-the-art structures.
• The skip hash also demonstrates resilience in mixed-operation environments (containing updates and range queries) by maintaining performance stability and avoiding degradation.

Implications and Future Work

The implications of these advancements are significant for both theoretical and practical applications:

• Theoretical Implications: The skip hash challenges existing assumptions about the limitations of STM, proposing that STM can efficiently support concurrent data operations with minimal contention-related overheads. This establishes a new baseline for future research in concurrent data structures.
• Practical Applications: In data-intensive environments where rapid concurrent access to ordered maps is required, such as databases and real-time analytics, the skip hash's efficient range queries and updates provide a robust solution.

Future work includes exploring modular clock mechanisms to further reduce contention on range query coordinator metadata, extending concepts to distributed or non-volatile memory environments, and continuing to refine STM optimizations to make transactional memory even more appealing for large-scale applications.

In summary, the skip hash represents a meaningful progression in the development of concurrent data structures, showcasing the power of combining skip lists with hash maps through the use of STM. This composition not only simplifies the design and implementation process but also significantly enhances performance metrics, which is critical for advancing the field of parallel computing.