Transmission of higher-order objects across a network of heterogeneous machines

• US 5,745,703 A
• Filed: 07/18/1995
• Issued: 04/28/1998
• Est. Priority Date: 07/18/1995
• Status: Expired due to Term

• Current Assignee: NEC Corporation
• Original Assignee: NEC Research Institute, Inc. (NEC Corporation)
• Inventors: Jagannathan, Suresh, Cejtin, Henry, Kelsey, Richard A.
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): HOMERE, JEAN RAYMOND
• Application Number:
  US08/503,566
• Time in Patent Office: 1,015 Days
• Field of Search: 395/610, 395/617, 395/618, 395/200.03, 395/200.13, 395/200.15, 395/200.68, 395/200.43, 395/200.75, 395/200.31, 707/10, 707/201, 707/202
• US Class Current: 709/238
• CPC Class Codes: G06F 9/546 Message passing systems or structures, e.g. queues

Abstract

The system comprises a collection of address spaces within which potentially many concurrent lightweight perceptible threads may execute. The address space is uniformly distributed among different nodes in a network of heterogeneous machines. Address spaces are first-class and may be generated dynamically. Threads within an address space may communicate with one another via shared memory; communication between address spaces takes place using explicit message-passing.

Claims

1. An apparatus for sending messages across machines comprising:

a plurality of data objects, code objects, threads and globally unique identifiers in at least one virtual machine in a network of machines; and

a plurality of address spaces in said at least one virtual machine, each address space having a unique identifier and a list of pairs of input and output channels, said address spaces being capable of creating and destroying said channels;

whereby a thread sends data objects, code objects, thread objects or global unique identifiers in the form of a message to a selected address space, said selected address space intercepts and reconstructs incoming messages.

2. An apparatus as set forth in claim 1, where said address space reconstructs messages dynamically so that a new message can be intercepted by said address space without said address space stopping intercepting incoming messages.

3. An apparatus as set forth in claim 1, where a plurality of threads in the same virtual machine simultaneously sends messages to a plurality of selected address spaces.

4. An apparatus as set forth in claim 1, further comprising a shared memory comprising unique identifiers, where each unique identifier is associated with a data object, code object or thread.

5. An apparatus as set forth in claim 1, where additional address space is added dynamically to a running computation without causing ongoing computations to halt.

6. An apparatus as set forth in claim 1, where said at least one virtual machine resides in a heterogeneous network of machines.

7. An apparatus as set forth in claim 6, where said network comprises a local area network.

8. An apparatus as set forth in claim 6, where said network comprises a wide area network.

9. An apparatus as set forth in claim 6, where said network resides in a single machine.

10. A method of sending messages across virtual machines comprising the steps of:

providing a plurality of data objects, code objects, threads and globally unique identifiers in at least one virtual machine;

providing a plurality of address spaces in said at least one virtual machine, each address space having a unique identifier and a list of pairs of input and output channels, said address space being capable of creating and destroying said channels; and

sending via a thread, data objects, code objects, thread objects or global unique identifiers in the form of a message to a selected address space, said selected address space intercepting and reconstructing incoming messages.

11. A method as set forth in claim 10, where said reconstructing is performed dynamically so that a new message can be intercepted without said selected address space stopping intercepting incoming messages.

12. A method as set forth in claim 10, where said sending via a thread sends messages simultaneously to a plurality of selected address spaces.

13. A method as set forth in claim 10, further comprising providing a shared memory comprising unique identifiers, where each unique identifier is associated with a data object, code object or thread.

14. A method as set forth in claim 10, further comprising dynamically adding address space to a running computation without causing ongoing computations to halt.

15. A method as set forth in claim 10, where said providing at least one virtual machine is in a heterogeneous network of machines.

16. A method as set forth in claim 10, further comprising garbage collecting performed weakly synchronously across a plurality of virtual machines.