Load balancing, error recovery, and reconfiguration control in a data movement subsystem with cooperating plural queue processors

• US 5,459,864 A
• Filed: 02/02/1993
• Issued: 10/17/1995
• Est. Priority Date: 02/02/1993
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Panner, Christine R., Dewkett, Thomas J., Scalzi, Casper A., Brent, Glen A.
• Primary Examiner(s): Kriess, Kevin A.
• Assistant Examiner(s): PAYNE, MATTHEW
• Application Number:
  US08/012,187
• Time in Patent Office: 987 Days
• Field of Search: 395/650
• US Class Current: 718/105
• CPC Class Codes: G06F 11/1471 involving logging of persistent data for recovery
• CPC Class Codes: G06F 11/2017 where memory access, memory control or I/O control functionality is redundant redundant communication control functionality G06F11/2005; redundant storage control functionality G06F11/2089
• CPC Class Codes: G06F 11/2035 without idle spare hardware
• CPC Class Codes: G06F 11/2043 where the redundant components share a common memory address space
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0647 Migration mechanisms
• CPC Class Codes: G06F 9/505 considering the load

Abstract

Provides load balancing, recovery and reconfiguration control for a data move subsystem comprised of a plurality of interconnected and cooperating data move processors (DMPs). Each DMP processor has an associated queue for receiving queue elements (QEs) from central processing units of a data processing system which specify data move requirements of the data processing system. QEs can be transferred between queues of other DMPs or a common queue to achieve load balancing, recovery and reconfiguration control.

Claims

1. A data move subsystem for use in a data processing system including one or more central processing units (CPUs), electronic memories, and input/output (I/O) devices where the CPUs execute program instructions generating work requests including data move (DM) work requests for moving data within the electronic memories or I/O work requests for moving data between I/O devices and the electronic memories, the data move subsystem being connected to the data processing system and I/O devices and including:

a plurality of data move processors (DMPs) each including control means, each said DMP characterized as being busy or non-busy executing a CPU generated work request;

a plurality of queues, each queue being associated with and accessible to one of said DMPs and including means for receiving one or more queue elements (QEs) from a CPU, each said QE specifying a DM work request or an I/O work request;

connecting means, said connecting means providing communication of control signals between said DMPs and transfer of QEs from one of said queues to another of said queues;

said control means of each said DMP including,

queue testing means, operative during a DM work request, for detecting the presence of a new QE specifying a DM work request in the associated and accessible one of said queues,

means for detecting the presence of a non-busy DMP, and

means, responsive to the detection of a non-busy DMP, for transferring the new QE on said connecting means to said queue associated with and accessible to said non-busy DMP.

2. A data move subsystem in accordance with claim 1 further including:

a common queue connected to and accessible to all of said DMPs;

said control means of each said DMP further including,

DMP status testing means, responsive to the detection of the absence of a non-busy DMP, for transferring the new QE to said common queue, and

means, operative at the completion of movement of all sub-units of data specified by a QE, for transferring a new QE from said common queue to said queue associated with and accessible to said DMP.

3. A data move subsystem in accordance with claim 1 wherein said queue testing means of said control means of each said DMP further includes:

means responsive to the detection of a new QE specifying an I/O work request for initiating the transfer of data between the electronic memories and an I/O device.

4. A data move subsystem in accordance with claim 2 wherein:

said common queue includes;

a common queue header including a plurality of status fields, each said status field associated with one of said DMPs for indicating the busy/non-busy status of the associated said DMP, and

said DMP status testing means of said control means includes;

means for examining the busy/non-busy indication of all of said status fields in said common queue header.

5. A data move subsystem in accordance with claim 4 wherein:

each said QE includes;

checkpoint data indicating the location of data elements to be moved, indicating the number of sub-units of data to be moved, and indicating the number of sub-units of data successfully moved for the DM work request, and

said control means of each said DMP includes;

means for updating the number of sub-units of data successfully moved at the completion of each sub-unit of data moved.

6. A data move subsystem in accordance with claim 5 wherein:

each said plurality of queues includes a queue header, said queue header including an enabled/disabled indicator signalling whether or not the associated said queue and said DMP is operative to receive new QEs;

said control means of each said DMP further includes,

reconfiguration signal testing means, operative during a DM work request and prior to operation of said queue testing means, for detecting the presence on said connecting means of a reconfiguration command signal, said reconfiguration command specifying this DMP as a Reconfigure DMP, said reconfiguration command specifying another one of said DMPs and whether said specified one of said DMPs is to be deleted from the system (Deleted DMP) or added to the system (Added DMP),

means for setting said enabled/disabled indicator associated with a Deleted DMP to the disabled state, or of an Added DMP to the enabled state,

means for transferring any said QE from said common queue to said queue associated with said Added DMP,

means for transferring any said QE in said queue associated with said Deleted DMP to said queue associated with this said Reconfigure DMP, and

means for setting said status field of said common queue header associated with said Deleted DMP to indicate a non-busy status.

7. A data move subsystem in accordance with claim 6 wherein said control means of each said DMP further includes:

error detection means for detecting error conditions in said DMP and for generating a recovery signal on said connecting means identifying this DMP as an Error DMP;

recovery signal testing means, operative during a DM work request and prior to operation of said queue testing means, for detecting the presence on said connecting means of said recovery signal said recovery signal specifying this DMP as a Recovery DMP;

means for setting said enabled/disabled indicator in said queue header associated with said Error DMP to the disabled state;

means for transferring any said QEs in said queue associated with said Error DMP to said queue associated with said Recovery DMP.