Construction method of high-efficiency transactional file system based on non-volatile main memory

Abstract

The invention discloses an efficient transaction file system construction method based on a non-volatile main memory. The method comprises the steps that a file system provides a corresponding file operation transaction interface; when a transaction begins, the file system allocates a transaction ID to the transaction; when the transaction is executed, the file system divides a log region into a metadata log region and a data log region, and different log processing methods are adopted according to metadata update and data update; when the transaction is submitted, the file system forces metadata content relevant to the transaction and data log content relevant to the transaction to be persisted into the non-volatile main memory, and submitted log entries are written to indicate that the transaction is completed; and the submitted data log content is periodically written back into the file system, and a concurrent selection write-back mechanism is adopted for write-back of the data log content. Through the method, design of an upper-level application program can be simplified through the transaction interface provided by the file system, and high efficiency and correctness in the transaction processing process can be guaranteed.

Description

technical field [0001] The invention relates to the technical field of non-volatile main memory storage, in particular to a non-volatile main memory-based high-efficiency transaction file system construction method. Background technique [0002] Emerging fast non-volatile memory (Non-Volatile Memory, NVM) technology, such as phase change memory (Phase Change Memory, PCM), spin-torque memory (Spin-Torque Transfer RAM, STT-RAM), and resistive memory (Resistive RAM, RRAM), with non-volatile, high access speed, byte addressing and other characteristics. Connecting these non-volatile memories to the main memory bus will generate non-volatile main memory (Non-Volatile Main Memory, NVMM), and enable storage devices to have memory-level performance. [0003] Applications in related technologies (such as text editors, traditional relational databases, and key-value storage systems, etc.) all need to ensure their data consistency. Furthermore, most of these applications are implemen...

AI Technical Summary

Problems solved by technology

However, research at the University of Wisconsin-Madison has shown that the consistency protocol implemented by the application layer is usually very complex and error-prone because the application cannot perceive the persistence characteristics of the underlying file system.

For example, out-of-order persistent operations caused by CPU out-of-order writes or atomic updates in non-volatile file systems that cannot guarantee sector granularity will cause applications to still lose or damage data in the event of system errors

In addition, most applications use disk-optimized logging mechanisms to ensure data consistency. These mechanisms will further hurt the performance of applications running on non-volatile main memory because they will generate excessive data copies.

Images