Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
1998-12-11
2002-05-14
Lee, Thomas (Department: 2182)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C710S021000, C714S011000, C714S048000
Reexamination Certificate
active
06389554
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a concurrent write duplexing device with extension of memory bus in a tightly coupled fault tolerance system, and more particularly, to a concurrent write duplexing device with extension of memory bus, to maintain memory data consistency within a duplexing operating processor module in a failure sensing control system.
2. Discussion of Related Art
Generally, all of systems which are manufactured by the adaptation of human being's technology always contain the possibility that various kinds of troubles caused due to their designers's mistake, failure of components and the like may occur. If, however, such the troubles occur in the system used for a specific purpose which should prevent the troubles, such as medical equipments, flight control systems, satellites, weapon systems, switching systems, a normal operation is impossible, which results in a serious problem. A fault tolerance system means a non-stop system in a system level which is constructed to operate in a designed order regardless of generation of hardware failure or software error. In the case where any trouble in the fault tolerance system occurs, the fault tolerance system fundamentally includes a redundancy module which can back up a system function and varies its own embodied method in accordance with the number and type of additional redundancy modules.
In case of encountering any fault in a switching system, the fault can be repaired at a proper time by an operator. Therefore, the switching system does not need a large amount of hardware redundancy which is required in the medical equipments, flight control systems, satellites, and weapon systems. Typically, the switching system is comprised of a module which executes a system function and a standby module which backs up the system function, which is embodied in a duplexing manner. The switching system which operates under high reliability and availability, supports a fault tolerancy function for some important parts in the duplexing manner. A control part as one of important parts in the switching system operates an active module for one portion and a standby module for the other portion by using the same processor module. In a preferred embodiment of the present invention, the data consistency of memory are keeped to be same in the active module and the standby module, and if the fault is produced in the active module, the standby module receives only state information of the active module and changes its own state to be the state of the active module, so that the system can be operated in a non-stop manner in a system level. It is of course important that the data consistancy between the two modules should be maintained to be same as each other. To this end, therefore, a concurrent write method is embodied in the preferred embodiment of the present invention. With the concurrent write method which is applied in the fault tolerance system, a memory write operation in the active module which is implemented in the standby module to constantly maintain the same memory contents, and then if any fault occurs in the active module, the standby module executes the system function in the defective active module and continuously proceeds the function in the system level regardless of the fault in the active module.
Examples of the conventional duplexing devices in which the concurrent write method is embodied are a duplexing data channel matching device using a tightly coupled data transmission method and a duplexing data channel matching device using decoupled data transmission method. Since these devices are achieved by extension of a system bus and alteration of most hardware according to change of a central processing unit (CPU), i.e., the system bus, they exhibit low compatibility and should have a long period of time in driving the hardware development. With the improvement of the performance of the CPU, the conventional device is not useful due to clock increment of the system bus and does not ensure a reliable application in hundreds of MHz of a currently commercial high performance microprocessor. In addition, since the duplexing data channel matching device using the coupled data transmission method should receive answer signals from the two modules to proceed the next operation, the device exhibits serious performance deterioration. Meanwhile, although the duplexing data channel matching device using the decoupled data transmission method separates a memory write operation and a concurrent write operation in the active module by using a first-in first-out (FIFO) to thereby solve the performance deterioration in the duplexing data channel matching device using the coupled data transmission method, the device overcomes the troubles generated in a complicated manner and also exhibits a high fault generation probability.
FIG. 1
is a block diagram illustrating a data transmission channel where a coupled data transmission system which does not separate a memory write operation and a concurrent write operation in an active module is employed, in a duplexing device in which a conventional concurrent write method is embodied.
As shown in the figure an active module
10
a
and a standby module
10
b
respectively include a central processing unit (CPU)
11
, a main memory
12
, a data transmission channel
13
and an input/output (I/O) matching device
14
.
The solid line as shown in
FIG. 1
indicates the memory write operation in the active module
10
a
which is executed to the main memory
12
by the CPU
11
or the I/O matching device
14
, and the dotted line indicates an answer signal process to inform the CPU
11
or the I/O matching device
14
that specific data is stored to each main memory
12
of the active module
10
a
and the standby module
10
b
. In this case, for the duplexing operation, the main memory write operation in the active module
10
a
is extended to a local bus in the standby module
10
b
through the data transmission channel
13
to be thereby transmitted to the main memory
12
in the standby module
10
b
, such that the data in the main memory
12
within the active module
10
a
and the standby module
10
b
are constantly maintained to be same as each other. In the data transmission channel where the coupled data transmission method is employed, however, there occurs a problem that since an answer signal to write operation completion should be received from the main memory
12
of the standby module
10
b
to complete the write operation to a specific region of the corresponding main memory
12
, no following operation can be executed before the answer signal is received from the standby module
10
b.
Accordingly, the overhead caused due to the waiting time for the answer signal from the standby module forcibly renders system performance in the duplexing device in which the conventional coupled transmission system is employed to be deteriorated, such that the data transmission channel using the conventional coupled transmission system can not be well employed in the system having a high performance processor.
FIGS. 2A and 2B
are block diagrams illustrating a data transmission channel in which a conventional decoupled data transmission system which is designed to minimize performance deterioration caused in
FIG. 1
is employed, where a main memory write operation within an active module and a concurrent write as a main memory write operation within a standby module through a data transmission channel are independently separated and operated.
As shown in the figure, an active module
20
a
and a standby module
20
b
respectively include a central processing unit (CPU)
21
, a main memory
22
, an input/output (I/O) bus matching device
23
, an SCSI Ethernet, miscellaneous I/O matching device
24
, and a high speed data transmission channel
25
. Further, a buffer for separating the operation in the two modules is disposed in the interior of each of the modules. However, in the conventional decoupled data transmission system, there
Jung Woo Sug
Kwon Bo Sub
Song Kwang Sug
Cohen & Pontani, Lieberman & Pavane
Electronics and Telecommunications Research Institute
Lee Thomas
Mai Rijue
LandOfFree
Concurrent write duplex device does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Concurrent write duplex device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Concurrent write duplex device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2869084