Electrical computers and digital processing systems: multicomput – Computer-to-computer data routing – Least weight routing
Reexamination Certificate
1997-12-11
2001-03-13
Banankhah, Majid (Department: 2151)
Electrical computers and digital processing systems: multicomput
Computer-to-computer data routing
Least weight routing
C709S219000, C709S223000
Reexamination Certificate
active
06202080
ABSTRACT:
MICROFICHE APPENDIX
A microfiche appendix having one page of microfiche with a total of 50 frames of a computer program set out in at least three software objects constitutes a part of the specification of this invention pursuant to 37 C.F.R. §§ 1.77 and 1.96, the microfiche appendix being incorporated by reference herein for all purposes.
A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by any one of the patent disclosure, as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all copyrights whatsoever.
TECHNICAL FIELD
The present invention relates to computer job workload distribution on a network, and in particular, to an apparatus and method for job workload distribution on a network utilizing a network file system.
BACKGROUND OF THE INVENTION
Although computer data is commonly shared with networked workstations, the collective computer power of a distributed-systems network remains under-utilized for computer job workload distribution. A distributed-systems network comprises a plurality of networked computer workstations. Each workstation is commonly referred to as a node. The nodes are in communication with each other through network communications protocols. A job is a computer workstation command. For each command execution, an allotment of Central Processing Unit (“CPU”) and electronic memory resources is required. “Compile,” “halt,” and “extract” are some examples of workstation commands.
Commonly, computer jobs requested are submitted in batch-queue or job-set fashion, to a single node. The jobs are typically executed by a node on a “first-come first-served” basis. When a computer job is completed, then the next job is queued for execution. The job execution process continues until the batch queue is completed. Accordingly, batch-queue submissions take a single node many hours to complete.
Reducing the execution time for multiple job requests is particularly important for telecommunication switch testcase environments. The faster the test jobs can be executed, the faster telecommunications switches can be ready for installation. These testcase environments automate test jobs for telecommunication switch “burn-in.” Examples of test jobs are database testcase extractions, package testcases for execution efficiency, testcase compilations, testcase executions, testcase execution result analysis, and testcase result database storage. These jobs can be combined and submitted to a node as a testcase set, which is a form of batch-queue submission.
To decrease the time to process job batch-queues, job workload distribution programs have been developed. But these programs are complex and cumbersome to use. Such programs have required a high-level system access to the root file system of a UNIX-based network. The root file system is an upper-level configuration layout of the network nodes. Such job workload distribution programs must query the root file system to determine whether data storage is available and whether distribution is even permitted to certain nodes.
Also, because root-level access is needed, these programs require hard-coded and application-specific network configurations. That is, the program must be tailored to the idiosyncracies of the network before it can be used. Further, root-level access is restricted to a single system administrator. When the network structure changes, due to additions or deletions of nodes, not only must the system administrator modify the network root file, but also reconfigure the distribution program to mirror the network changes.
Furthermore, when a program demands root-level access, the overall “controllability” and “granularity” characteristics of the network are adversely affected. That is, system level commands have been limited to large “work units” (low granularity). Large work units restrict the flexibility for a lower-level user—such as an application level user—to manipulate a work unit (low control level).
Workload distribution programs have also relied on electronic messaging formats to distribute jobs to other network nodes. But such electronic messaging is not robust because of susceptibility to power fluctuations. Data message records are not typically saved to a disk. Thus, a power failure or a node crash on the network jeopardizes the in-transit message-data because the information is not maintained.
Thus, a need exists for a simplified workload distribution method and apparatus that does not require root-level access to distribute job workloads across a network. Also, a workload distribution method and apparatus is desired that has a greater robustness, or ability to function or continue functioning in unexpected situations.
SUMMARY OF THE INVENTION
These and other disadvantages are overcome by the present invention, which comprises an improved method and apparatus for job workload distribution in a computer network.
In a first aspect of the invention, provided is a job distribution system implemented by a computer network. The computer network has a plurality of nodes, with each of the nodes having a local directory that is cross-mounted with other network node directories. With cross-mounted directories, the networking capacity is invoked automatically, and the physical layout of the network is hidden from the user. The result is, in virtual terms, a single, large, distributed computer system.
The job workload distribution system has a job balance object, a node daemon object, and an updatable node distribution table. The job distribution system is on a first node of a plurality of nodes. This node can be either a local node, which holds the network server application, one of the remaining or remote nodes, or a combination local/remote node. The node daemon object is on each remaining node of the plurality of nodes, and the job balance object has access to the local directory of each of the remaining nodes. The updatable node distribution table has a plurality of distribution data structures. Each of these data distribution structures corresponds to each of the remaining nodes, and are updated by the node daemon object with a job status value, such as “complete,” “aborted,” or “failed.” The job balance object distributes the plurality of jobs among the remaining nodes according to an average job status of each of the remaining nodes.
In another aspect of the invention, a workload distribution method for a computer network is provided. The workload distribution method is for a network having at least one local node and at least one remote node. The local directories of each node are cross-mounted with each other. The workload distribution method provides at least an updatable node distribution table with a plurality of distribution data structures. Each of the distribution data structures corresponds to a network node. A plurality of job requests are submitted by a resident client object. Each of the plurality of nodes updating the distribution data structure with a job status that corresponds to that remote node. The job requests are balanced among the plurality of nodes when a balancing criteria as a function of the job status value of each node. The distribution data structures relating to each of the remote nodes are updated as job requests are executed.
REFERENCES:
patent: 5032979 (1991-07-01), Hecht
patent: 5053950 (1991-10-01), Naganuma et al.
patent: 5283897 (1994-02-01), Georgiadis et al.
patent: 5675791 (1997-10-01), Bhide et al.
patent: 5845265 (1998-12-01), Woolston
patent: 5923850 (1999-07-01), Barroux
patent: 6026425 (2000-02-01), Suguri et al.
patent: 6047323 (2000-04-01), Krause
Asava Anand
Juan Shao-Min
Lu James
Banankhah Majid
Carr & Storm, L.L.P.
Nortel Networks Limited
LandOfFree
Apparatus and method for computer job workload distribution does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for computer job workload distribution, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for computer job workload distribution will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2548823