SYSTEM AND METHOD FOR SUPPORTING AN ADAPTIVE SELF-TUNING LOCKING MECHANISM IN A TRANSACTIONAL MIDDLEWARE MACHINE ENVIRONMENT

• US 20150317191A1
• Filed: 06/19/2014
• Published: 11/05/2015
• Est. Priority Date: 03/07/2012
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Shen, Xugang, Zhang, Qingsheng, Jin, Yongshun
• Granted Patent: US 9,846,603 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 11/073 in a memory management context, e.g. virtual memory or cache management memory management G06F12/00; testing of static memory units G11C29/00
• CPC Class Codes: G06F 11/0784 Routing of error reports, e.g. with a specific transmission path or data flow
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40
• CPC Class Codes: G06F 11/34 Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
• CPC Class Codes: G06F 9/52 Program synchronisation; Mutual exclusion, e.g. by means of semaphores
• CPC Class Codes: G06F 9/526 Mutual exclusion algorithms

Abstract

A system and method can support an adaptive self-tuning locking mechanism in a transactional middleware machine environment. The system allows each process in a plurality of processes to perform one or more test-and-set (TAS) operations in order to obtain a lock for data in a shared memory. Then, the system can obtain a spin failed rate for a current tuning period, wherein a spin failure happens when a process fails to obtain the lock after performing a maximum number of rounds of TAS operations that are allowed. Furthermore, the system can adaptively configuring a spin count for a next tuning period based on the obtained spin failure rate, wherein the spin count specifies the maximum number of rounds of TAS operations that are allowed for the next tuning period.

Claims

1. A method for supporting an adaptive locking mechanism in a transactional middleware machine environment, comprising:

performing, via each process in a plurality of processes, one or more test-and-set (TAS) operations in order to obtain a lock for data in a shared memory;

obtaining a spin failed rate for a current tuning period, wherein a spin failure happens when a process fails to obtain the lock after performing a maximum number of rounds of TAS operations that are allowed; and

adaptively configuring a spin count for a next tuning period based on the obtained spin failure rate, wherein the spin count specifies the maximum number of rounds of TAS operations that are allowed for the next tuning period.

2. The method of claim 1, further comprising:

using an assembly component to perform said one or more TAS operations.

3. The method of claim 1, further comprising:

preconfiguring the spin count using metadata.

4. The method of claim 1, further comprising:

determining whether an idle CPU rate for the current tuning period is more than a user-configured minimum idle CPU rate and whether a spin failed ratio for the current tuning period meets a user-configured target.

5. The method of claim 4, further comprising:

when an idle CPU rate for the current tuning period is more than a user-configured minimum idle CPU rate and a spin failed ratio for the current tuning period does not meet a user-configured target, increasing the spin count

if the spin failed ratio for the current tuning period is better than a last good spin failed ratio, or

if the spin failed ratio for the current tuning period becomes worse without tuning.

6. The method of claim 5, further comprising:

dividing a range for possible spin count values into several intervals, and

using a different formula to calculate a tuned spin count value in each different interval.

7. The method of claim 1, further comprising:

decreasing the spin count, if at least one of

an application is idle,

the idle CPU ratio is less than a user-configured minimum idle CPU rate and the user CPU ratio is sufficient, and

the spin count has been kept stable for a period of time.

8. The method of claim 1, further comprising:

keeping the spin count, if at least one of

the spin failed ratio for the current tuning period meets the requirement,

the spin count for the current tuning period reaches the top or bottom limitation, and

the spin failed ratio for the current tuning period is kept stable.

9. The method of claim 1, further comprising:

waiting, via a process with a spin failure, in a semaphore waiting queue for a release of a lock on the data in the shared memory; and

accessing the data in the shared memory, when a lock owner releases the lock on the data.

10. The method of claim 9, further comprising:

applying extra spin counts on each process performing the TAS operation, when the semaphore waiting queue is not empty.

11. A system for providing an adaptive locking mechanism in a transactional middleware machine environment, comprising:

one or more processors;

a plurality of processes, wherein each process in the plurality of processes operates to perform one or more test-and-set (TAS) operations in order to obtain a lock for data in a shared memory;

a transactional server, running on the one or more processors, operate to

obtain a spin failed rate for a current tuning period, wherein a spin failure happens when a process fails to obtain the lock after performing a maximum number of rounds of TAS operations that are allowed; and

adaptively configure a spin count for a next tuning period based on the obtained spin failure rate, wherein the spin count specifies the maximum number of rounds of TAS operations that are allowed for the next tuning period.

12. The system of claim 11, wherein:

an assembly component is used to perform said one or more TAS operations.

13. The system of claim 11, wherein:

the spin count is preconfigured using metadata.

14. The system of claim 11, wherein:

the managing component operates to determine whether an idle CPU rate for the current tuning period is more than a user-configured minimum idle CPU rate and whether a spin failed ratio for the current tuning period meets a user-configured target.

15. The system of claim 14, wherein:

when an idle CPU rate for the current tuning period is more than a user-configured minimum idle CPU rate and a spin failed ratio for the current tuning period does not meet a user-configured target, the managing component operates to increase the spin count

if the spin failed ratio for the current tuning period is better than a last good spin failed ratio, or

if the spin failed ratio for the current tuning period becomes worse without tuning.

16. The system of claim 15, wherein:

a range for possible spin count values is divided into several intervals, and the managing component operates to use a different formula to calculate a tuned spin count value in each different interval.

17. The system of claim 11, wherein:

the managing component operates to decrease the spin count, if at least one of an application is idle,

the idle CPU ratio is less than a user-configured minimum idle CPU rate and the user CPU ratio is sufficient, and

the spin count has been kept stable for a period of time.

18. The system of claim 11, wherein:

the managing component operates to keep the spin count, if at least one of

the spin failed ratio for the current tuning period meets the requirement,

the spin count for the current tuning period reaches the top or bottom limitation, and

the spin failed ratio for the current tuning period is kept stable.

19. The system of claim 11, wherein:

the managing component operates to

put a process with a spin failure to wait in a semaphore waiting queue for a release of a lock on the data in the shared memory;

access the data in the shared memory, when a lock owner releases the lock on the data; and

apply extra spin counts on the processes performing the TAS operation, when the semaphore waiting queue is not empty.

20. A non-transitory machine readable storage medium having instructions stored thereon that when executed cause a system to perform the steps comprising:

performing, via each process in a plurality of processes, one or more test-and-set (TAS) operations in order to obtain a lock for data in a shared memory;

obtaining a spin failed rate for a current tuning period, wherein a spin failure happens when a process fails to obtain the lock after performing a maximum number of rounds of TAS operations that are allowed; and

adaptively configuring a spin count for a next tuning period based on the obtained spin failure rate, wherein the spin count specifies the maximum number of rounds of TAS operations that are allowed for the next tuning period.