Method to reduce the latency caused by garbage collection in a java virtual machine

• US 20060248042A1
• Filed: 04/17/2006
• Published: 11/02/2006
• Est. Priority Date: 04/29/2005
• Status: Abandoned Application

• Current Assignee: Alcatel SA (Nokia Corporation)
• Original Assignee: Alcatel SA (Nokia Corporation)
• Inventors: Anna Janssens, Nico Alfred
• Application Number:
  US11/404,820
• Publication Number:
  US 20060248042A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 12/0253 Garbage collection, i.e. reclamation of unreferenced memory

Abstract

A method to reduce the latency caused by garbage collection in a Java Virtual Machine (JVM) wherein different machines (MC1, MC2, MC3) are running consecutive tasks (P1, P2, P3) of an application. Because instances of a garbage collector (GC1, GC2, GC3) have to run on each machine and interrupt the task running then on that machine, their starts are synchronized to optimize the impact on the latency of the application'"'"'s response. In other words, because the tasks of the application run successively on different machines, the garbage collector (GC) affects the operation of only one machine by starting all its instances simultaneously on all the machines, i.e. by synchronizing the instances of the GC tasks. Each machine may be of the multi-processor type (CPU1, CPU2, CPU3) whereby the operation of all the processors is interrupted when an instance of the garbage collector is running on one of these processors.

Claims

1. A method to run tasks (P1, P2, P3) of an application on different machines (MC1, MC2, MC3) within a Java Virtual Machine (JVM) environment, while running instances of a garbage collector (GC1, GC2, GC3) at predetermined time periods on each of said machines, and wherein each instance of said garbage collector interrupts the task running on the same machine,

characterized in that said method comprises the step of starting all the instances of said garbage collector (GC1, GC2, GC3) simultaneously on all the machines (MC1, MC2, MC3).

2. The method according to claim 1, characterized in that said application comprises a plurality of consecutive tasks (P1, P2, P3) distributed over the different machines (MC1, MC2, MC3), wherein a first task (P1) completed on a first machine (MC1) is followed (T1) by a second task (P2) running on a second machine (MC2).

3. The method according to claim 1,

characterized in that each machine (MC1, MC2, MC3) comprises a plurality of processors (CPU1, CPU2, CPU3),

and in that the operation of all the processors of a same machine are simultaneously interrupted by the instance of said garbage collector (GC1, GC2, GC3) running on one of the processors of said machine.

4. The method according to claim 1,

characterized in that an instance of said garbage collector (GC1, GC2, GC3) comprises at least a portion that interrupts the task running on the same machine,

and in that said method comprises the step of starting all the interrupting portions of the instances of said garbage collector simultaneously on all said machines (MC1, MC2, MC3).