Scaling HTM-Supported Database Transactions to Many Cores

Abstract

So far, transactional memory-although a promising technique-suffered from the absence of an efficient hardware implementation. Intel's Haswell microarchitecture introduced hardware transactional memory (HTM) in mainstream CPUs. HTM allows for efficient concurrent, atomic operations, which is also highly desirable in the context of databases. On the other hand HTM has several limitations that, in general, prevent a one-to-one mapping of database transactions to HTM transactions. In this work we devise several building blocks that can be used to exploit HTM in main-memory databases. We show that HTM allows to achieve nearly lock-free processing of database transactions by carefully controlling the data layout and the access patterns. The HTM component is used for detecting the (infrequent) conflicts, which allows for an optimistic, and thus very low-overhead execution of concurrent transactions. We evaluate our approach on a 4-core desktop and a 28-core server system and find that HTM indeed provides a scalable, powerful, and easy to use synchronization primitive.

Scalable and robust latches for database systems

“…A special form of optimistic locking is Intel's hardware-supported speculative locking [18,19,21]. Speculative locking allows multiple threads to hold the same lock as long as their operations do not conflict [8].…”

DHTM: Durable Hardware Transactional Memory

“…While the database community has developed a plethora of techniques to guarantee ACID efficiently, these techniques have predominantly been developed with slow block-based media in mind. When applied to in-memory settings, such techniques tend to spend a significant amount of time on concurrency control [6], [7], [8] and logging [6], [9], [10]. This leads us to ask the question: How fast can we enforce ACID in the presence of fast persistent memory?…”