Adaptive concurrency control using hardware transactional memory and locking mechanism

Patent metadata

Field	Value
Publication number	US-2016299798-A1
Application number	US-201514795339-A
Country	US
Kind code	A1
Filing date	Jul 9, 2015
Priority date	Apr 10, 2015
Publication date	Oct 13, 2016
Grant date	—

Abstract

A method includes the following steps. Runtime statistics related to data transaction processing in a concurrent system are collected. A given request to access shared data in the concurrent system is receive. Based on the collected runtime statistics, the number of reattempts the given request can make to access the shared data prior to access control being switched from a hardware transactional memory to a locking mechanism is adaptively determined.

First claim

What is claimed is:

1 . A method, comprising:
collecting runtime statistics related to data transaction processing in a concurrent system; receiving a given request to access shared data in the concurrent system; and adaptively determining, based on the collected runtime statistics, the number of reattempts the given request can make to access the shared data prior to access control being switched from a hardware transactional memory to a locking mechanism.

2 . The method of claim 1 , wherein the collected runtime statistics comprise a metric indicative of the number of read requests made to the shared data in a given time period.

3 . The method of claim 2 , wherein the metric comprises the number of read requests made to the shared data in the given time period divided by the total number of requests made to the shared data in the given time period.

4 . The method of claim 2 , wherein the number of reattempts the given request can make to access the shared data prior to access control being switched from the hardware transactional memory to the locking mechanism adaptively increases when the number of reads is above a predetermined threshold.

5 . The method of claim 1 , wherein the collected runtime statistics comprise a metric indicative of the number of write requests made to the shared data in a given time period.

6 . The method of claim 5 , wherein the metric comprises the number of write requests made to the shared data in the given time period divided by the total number of requests made to the shared data in the given time period.

7 . The method of claim 5 , wherein the number of reattempts the given request can make to access the shared data prior to access control being switched from the hardware transactional memory to the locking mechanism adaptively decreases when the number of writes is above a predetermined threshold.

8 . The method of claim 1 , wherein the collected runtime statistics comprise a metric indicative of the number of successful requests made to the shared data in a given time period.

9 . The method of claim 8 , wherein the metric is a success ratio comprising the number of successful requests made to the shared data in the given time period divided by the total number of requests made to the shared data in the given time period.

10 . The method of claim 9 , wherein the collected runtime statistics comprise a metric indicative of the number of failed requests made to the shared data in a given time period.

11 . The method of claim 10 , wherein the metric is a failure ratio comprising the number of failed requests made to the shared data in the given time period divided by the total number of requests made to the shared data in the given time period.

12 . The method of claim 11 , wherein the number of reattempts the given request can make to access the shared data prior to access control being switched from the hardware transactional memory to the locking mechanism adaptively increases when the success ratio is greater than the failure ratio.

13 . The method of claim 11 , wherein the number of reattempts the given request can make to access the shared data prior to access control being switched from the hardware transactional memory to the locking mechanism adaptively decreases when the success ratio is less than the failure ratio.

14 . The method of claim 1 , wherein the collected statistics comprise the total number of reattempts made to access the shared data.

15 . The method of claim 1 , wherein the collected statistics comprise the number of operations associated with the given request.

16 . The method of claim 15 , wherein the number of reattempts the given request can make to access the shared data prior to access control being switched from the hardware transactional memory to the locking mechanism adaptively decreases as the number of operations associated with the given request increases.

17 . The method of claim 1 , wherein the collected statistics comprise the amount of data associated with the given request.

18 . The method of claim 17 , wherein locking mechanism is utilized when the amount of data associated with the given request exceeds the size of available transactional memory.

Assignees

• IBM

Inventors

• Bhattacharjee Bishwaranjan
• Canim Mustafa
• Chang Yuan-Chi
• Hamedani Mohammad Sadoghi
• Makni Bassem
• Ross Kenneth Andrew

Classifications

• G06F3/0659

  Command handling arrangements, e.g. command buffers, queues, command scheduling · CPC title

• G06F3/0622 Primary

  in relation to access · CPC title

• G06F9/467 Primary

  Transactional memory (G06F9/528 takes precedence) · CPC title

• G06F2212/1052

  Security improvement · CPC title

• G06F12/1466

  Key-lock mechanism · CPC title