KR100310285B1 - Group communication method supporting fault-tolerant service in the real-time object-oriented distributed platform - Google Patents

Abstract

The present invention relates to a group communication method for supporting a fault tolerance service in a real-time object-oriented distributed platform, and a recording medium. The present invention manages membership within a group of processors and a plurality of groups in a distributed computing environment, and based on the concept of object redundancy. In order to provide a group communication method and a recording medium for supporting a fault tolerance service on a real-time object-oriented distributed platform, all the messages are arranged in the same order in the receiving processing unit of all the objects in the group to ensure data integrity. Delivering; Retransmitting the missing message while circulating the token message between the group objects in the token processing unit, and managing the received message in a sorted and complete form by newly assigning a sequence number; Determining whether or not to transmit a message for each group model in order to improve the efficiency of the message transmission, and managing the transmission buffer for fault tolerance; And determining a new group member by a failure handling unit when a failure occurs in some objects in the group, and used for a real-time system having a distributed platform.

Description

Group communication method supporting fault-tolerant service in the real-time object-oriented distributed platform

The present invention relates to a group communication method for supporting a fault tolerance service in an object-oriented distributed platform in a distributed computing environment, and to a computer readable recording medium having recorded thereon a program for realizing the method.

The object-oriented distributed platform has a standard specification called Common Object Request Broker Architecture (CORBA), and is currently enacting additional specifications regarding real-time and fault tolerance services.

The fault tolerance service in CORBA is based on the concept of object redundancy, in which a plurality of duplicate objects are formed into duplicate object groups so that each member performs the same function. .

On the other hand, communication between object groups or within an object group is one-to-many or many-to-many communication as multicasting. In order to obtain the same result in all objects, a message in which all objects in a group are arranged in a causal order in the same order It is essential to execute them in turn, and this condition is called causal ordering or total ordering.

Much research has been done in this area because it is a very complex problem to consistently meet the alignment conditions while tolerating the failures in the objects in the group.

Conventional group communication has used methods in which the objects in a group do not need to be duplicate objects and are applied under a general situation in which one object belongs to any plurality of groups.

On the other hand, in the replication object group for fault-tolerant service of object-oriented distributed platform, all members in the group have the same code to perform the same task, and one object belongs to exactly one group, and requests between client / server connection between groups It operates according to the response procedure, so a much more efficient way can be devised. Therefore, an efficient communication method capable of supporting a fault tolerance service on a real-time object-oriented distributed platform is required.

The present invention has been proposed to meet the above requirements, and manages membership within a group of processors and a plurality of groups in a distributed computing environment, and provides fault tolerance services on a real-time object-oriented distributed platform based on the concept of object redundancy. It is an object of the present invention to provide a group communication method for supporting a computer readable recording medium recording a program for realizing the method.

1 is an exemplary configuration of a real-time object-oriented distributed system to which the present invention is applied.

FIG. 2 is a diagram illustrating an embodiment of functions of duplicate object group models supported by a group communication module used in the present invention. FIG.

3 is an explanatory diagram showing a message flow between groups used in the present invention.

4 is an exemplary internal configuration of a group communication module used in the present invention.

5 is a diagram for explaining an internal function of a group communication module used in the present invention.

FIG. 6 is a flowchart illustrating a process performed by a receiving processor in a group communication module in a group communication method supporting a fault tolerance service according to the present invention; FIG.

7 is a flowchart illustrating a process performed by a token processing unit in a group communication module in a group communication method supporting a fault tolerance service according to the present invention;

8 is a flowchart illustrating a process performed by a transmission processing unit in a group communication module in a group communication method supporting fault tolerance service according to the present invention.

9 is a flowchart illustrating a process performed by a failure processing unit in a group communication module in a group communication method supporting a fault tolerance service according to the present invention;

* Explanation of symbols for the main parts of the drawings

1: distributed object 2: real-time distributed object-oriented platform

3: duplicate object group 4: network

5: group communication module

In order to achieve the above object, the present invention provides a method for implementing fault tolerance through object duplication by forming a duplicate object group in an object-oriented distributed platform providing fault tolerance service in a networked real-time distributed system. In a real-time group communication method, a reception processor for receiving an inter-object message uses a reception buffer and an alignment buffer to queue a message having no reception sequence number in the reception buffer, and to store the message with the reception sequence number in the alignment buffer. A first step of arranging all messages in the same order by the receiving processing unit of all objects in the group in order to ensure data integrity; In order to recover lost messages from each object in the group and to give a consistent message receiving sequence number, the token processing unit circulates the token message between group objects, retransmits the missing message, and assigns a new sequence number to sort the received message. And second step of managing in complete form; A third step of determining whether or not to transmit a message for each group model and managing a transmission buffer for fault tolerance in order to increase the efficiency of message transmission in a transmission processor that receives a transmission message from the platform and transmits the message to an destination group; And a fourth step of determining a new group member by the failure processing unit when a failure occurs in some objects in the group.

The present invention provides a real-time distributed system having a processor, in which a receiving processor that receives an inter-object message waits for a message in which a receiving sequence number is not determined using a receiving buffer and an alignment buffer, in the receiving buffer. A first function of waiting for the assigned message in the sorting buffer, and in order to ensure data integrity, the reception processing unit of all objects in the group lists all the messages in the same order and delivers them to the platform; In order to recover lost messages from each object in the group and to give a consistent message receiving sequence number, the token processing unit circulates the token message between group objects, retransmits the missing message, and assigns a new sequence number to sort the received message. Second function of managing in complete form; A third function of determining whether to send a message for each group model and executing the transmission buffer in a transmission processing unit receiving the transmission message from the platform and transmitting the message to the destination group, and managing the transmission buffer for fault tolerance; And a computer-readable recording medium having recorded thereon a program for realizing a fourth function of determining a new group member by a failure processing unit when a failure occurs in some objects in the group.

The present invention relates to a lower communication method that is most important in providing fault tolerance services in a real-time object-oriented distributed platform, and forms a plurality of duplicate objects into a duplicate object group so that each member performs the same function. Even if it occurs, the rest of the objects are based on the concept of object redundancy, which normally proceeds normally. In other words, communication between object groups or intra-object groups is one-to-many or many-to-many communication with multicasting. In order to achieve the same result in all objects, messages are arranged in order in the same order in order.

To this end, the present invention is a group communication module between a distributed platform and a network, and has a function of storing a request / response multicast message received by a receiving processor in a buffer, and a token processing unit having a token corrects a processing sequence of a received message. The ability to maintain data integrity, reliably transmit request / response multicasts, form new groups through replicated object groups, and reach consensus on member members to return to normal It is provided.

In this way, the present invention provides a group communication module between the platform and the network at the time of group communication support in efficiently providing various fault tolerant services in a real-time object-oriented distributed platform, while maintaining its own fault tolerance function and stopping service and By supporting real-time group communication without delay due to object locking, it can be introduced and used in real-time system with distributed platform.

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is an exemplary configuration diagram of a real-time object-oriented distributed system to which the present invention is applied. In the drawing, '1' is a distributed object, '2' is a real-time distributed object-oriented platform, '3' is a replicated object group, and '4' is The network and '5' represent group communication modules, respectively.

As shown in FIG. 1, the real-time object-oriented distributed system to which the present invention is applied is a real-time distributed object-oriented platform (2) that supports function calls between distributed objects (1), a set of physically distributed replica objects. A replication object group (3) configured to implement functionality within a fault tolerance, a network (4) that physically conveys messages between platform components, and a replication object group (3), in order to allow a distributed platform to function as an organism It consists of a group communication module 5 which is responsible for multicasting communication efficiently.

2 is a diagram illustrating an example of functions of a replica object group model supported by a group communication module used in the present invention, in which '6' is an activity-waiting group, '7' is an activity object, and '8' '9' is a waiting object, '9' is a request message, '10' is a response message, '11' is a multicast transmission of periodic status information transmission, '12' is a representative group, '13' is a representative object, '14 'Represents the active object,' 15 'represents the entire active group, and' 16 'represents the active object.

In the activity-waiting group 6, one of the objects in the group is selected as the activity object 7 and the other is operated as the standby object 8. When the request message 9 arrives at the activity-waiting group 6, only the activity object 7 processes the message and sends the processing result as the response message 10. The activity object 7 periodically sends 11 its current state information to the objects in the group.

The representative group 12 consists of one representative object 13 and an active object 14, all of which execute a request message, and only the representative object 13 transmits the execution result to the request object. At this time, the representative object 13 does not need to periodically multicast the state information.

In the whole active group 15, all member objects are active objects 16, unlike in the representative group 12, all members send a response message to the requesting object.

3 is an explanatory diagram showing message flow between groups used in the present invention, in which '17' is a transmission group, '18' is a reception group, '19' is a transmission message, '20' is a transmission sequence number, ' 21 'represents a table indicating a target group to which a transmission sequence number and a reception sequence number should be managed in each group, respectively.

As shown in FIG. 3, when there is a message 19 transmitted from one group 17 to another group 18, the former 17 transmits a transmission sequence number 20 to a message sent to the latter 18. The latter 18 records the sequence number in the arrived message as a reception sequence number for the former 17.

The table 21 displays target groups for which the transmission sequence number and the reception sequence number should be managed in each group.

4 is an exemplary internal configuration of a group communication module used in the present invention, in which '22' is a receiving processor, '23' is a newly arrived object message, '24' is a receiving buffer, '25' is an alignment buffer, '26' delivers the first message in the sort buffer, '27' receives the token processing unit, '28' receives the token message, '29' selects candidates from the waiting messages and moves them to the sort buffer, and '30' renews them. The token is sent to the next object in the group, with '31' being the sender, '32' being sent from the platform, '33' being the send buffer, and '34' being the first message in the send buffer. The transmission and '35' respectively indicate the arrival of an ACK message confirming receipt of the message from the destination group.

The reception processor 22 stores the newly arrived object message 23 in the reception buffer 24, and delivers the head message 26 in the alignment buffer 25 to the platform only when data integrity is guaranteed.

The token processing unit 27 waits until the token message is received 28, and when the token message arrives, the sort buffer 25 is updated using the information in the token, and among the messages waiting in the reception buffer 24 The candidate is selected and moved to the alignment buffer 25 (29). After a period of time, the token is sent 30 to the next object in the group with the newly updated information.

The transmission processing unit 31 stores the transmission message arriving 32 from the platform in the transmission buffer 33. After the message 34 at the head of the transmission buffer 33 has been transmitted 34 to the destination group, it waits until an ACK message arrives 35 confirming receipt of a message from the destination group.

FIG. 5 is a diagram illustrating an internal function of a group communication module used in the present invention, in which '36' is an object message received from a network, '37' is a receiving buffer, and '38' is a receiving sequence number. List, '39' is the list of incoming sequence numbers in the module, '40' is the sort buffer, '41' is the multicast transmission of management messages, '42' is the message leading to the sort buffer to the platform, and '43' Is a list of completed sequence numbers, '44' is a send message delivered from the platform, '45' is a send buffer, '46' is a send sequence list, and '47' is the first message in the send buffer. '48' indicates a transmission sequence number list in the token message, and '49' indicates a transmission completion sequence number list, respectively.

The object message 36 received from the network is stored in the receive buffer 37 and waits to be moved to the alignment buffer 40.

When the module receives the token, it first compares the received sequence number list 38 in the token and the received sequence number list 39 in the module to show the received message sequence numbering status of the previous token module and sends out missing messages. Ask the old token module.

Next, from among the messages remaining in the reception buffer 37, the transmission sequence number is selected to be one greater than the transmission group corresponding reception sequence number in the reception sequence number list, and the reception sequence number of the transmission group in the reception sequence number list is increased by one. After attaching the list value to the message and adding it to the last position of the sort buffer 40, multicast transmission of the ACK message with the received sequence number list to the member objects in the group to which the module belongs and the message sending group 41).

The message at the head of the sort buffer 40 is delivered to the platform 42 upon recognizing that an absolute majority of the members in the group to which the module belongs are received (42), and the delivery complete sequence list (43) is also a message. It is set to the list value in.

The outgoing message 44 delivered from the platform is stored in the outgoing buffer 45, and the object to which the current module belongs is an activity object of an activity-representative group, a representative object of a representative group, or an object of all active types. Finds the sequence number last given to the message at the head of the transmission buffer from the transmission sequence number 46, increments 1, attaches the value to the message as the transmission sequence number, and sends it to the destination (47). After updating the list value, it waits for an ACK message from the destination group.

In the case of an object that cannot transmit a message, that is, a standby object of an active-waiting group or an active object of a representative group, the transmission message is stored in the transmission buffer 45 to be used for message recovery in case of failure. If the transmission message is safely transmitted through the transmission sequence number list 48 in the token message, the corresponding messages are deleted in the transmission buffer 45, and the transmission sequence number 46 in the module is also updated. If there is no corresponding message, the transmission sequence number list value in the token is copied to the transmission completion sequence number list 49 in the module for further processing.

FIG. 6 is a flowchart illustrating an example of a process performed by a receiving processor in a group communication module in a group communication method supporting fault tolerance service according to the present invention.

As shown in FIG. 6, in the group communication method supporting the fault tolerance service according to the present invention, the processing in the reception processing unit in the group communication module first initializes (601) the variables used in the reception processing unit and the object message. If a message arrives, it checks whether it has received the same message in the past and discards it if it is a duplicate message, and adds it to the reception buffer if it is a new message (603).

When an ACK message is received from the token objects in the same group (604), the message is found from the receive buffer, and a list of received sequence numbers in the ACK message is attached to the sort buffer. Then, the sequence number in the ACK message is compared with the received sequence number in the module, and if there is a missing message, the token object is requested to be sent (605).

Next, it analyzes whether the message exists in the sort buffer (606), if there is a message, compares the sequence number of the message in the buffer header with the list of completed sequence numbers in the module, and if there is another message in between, requesting the transmission of the message. If the token object is found, the header message is transmitted to the platform when the header message is received by more than half of the object modules in the group, and the list of completed sequence numbers in the module is updated (607). Thereafter, the process returns to step 602 again to repeat the above process.

7 is a flowchart illustrating a process performed by a token processing unit in a group communication module in a group communication method supporting a fault tolerance service according to the present invention.

As illustrated in FIG. 7, in the group communication method supporting the fault tolerance service according to the present invention, the processing in the token processing unit in the group communication module first initializes the variables related to the token processing unit (701) and the token message. Wait until 702 arrives (702).

Then, when a token message arrives, it uses the list of receive sequence numbers in the token to request and receive messages missing from the receive and sort buffers from the previous token object, then cleans up the receive buffer and places the sort buffer the same as the previous token object. Own them (703).

Next, if the group to which the current module belongs is an activity-waiting or representative type, the transmission sequence number list in the token is copied to the transmission completion sequence number list in the module (704). After that, for a certain period of time (707), the candidate message that can move from the receiving buffer to the sorting buffer is selected (705), and the corresponding sequence number is increased by 1, attached to the message, moved to the sorting buffer, and the corresponding list value is updated. Thereafter, an ACK message is transmitted to the object that has sent the message and the group to which the current module belongs (706).

After a certain period of time (707), when the token is passed to the next object, the token message is transmitted to the next object in the group after the new value in the token is newly updated (708).

8 is a flowchart illustrating a process performed by a transmission processing unit in a group communication module in a group communication method supporting fault tolerance service according to the present invention.

As shown in FIG. 8, in the group communication method supporting the fault tolerance service according to the present invention, the processing in the transmission processing unit in the group communication module first initializes the relational variables (801), and the current module sends the transmission message. It determines whether it is eligible to send (802), if it is not eligible, removes the messages already sent from another object from the send buffer (803), and if it is entitled to send the send message (804), it sends the message at the head of the send buffer. Multicasting is sent to the target group (805).

Thereafter, after waiting for an ACK message from the target group (806), the internal sequence number is updated (807), and a check is made whether the transmission message from the platform is waiting (808), and if there is a transmission message, it is delivered from the platform. Receive the data, and store the data in the transmission buffer (809) and return to step 802 again to repeat the above steps.

9 is a flowchart illustrating a process performed by a failure processing unit in a group communication module in a group communication method supporting a fault tolerance service according to the present invention.

As shown in FIG. 9, in the group communication method supporting the fault tolerance service according to the present invention, the processing in the failure processing unit in the group communication module first initializes the relationship variables (901) and determines whether a failure has occurred (902). Or if the failure recovery message is received (903), and if it detects a failure, it becomes a leading object and performs steps (904 to 912) for failure recovery, and when a failure recovery message is received, it becomes a late object. In accordance with the map of the lead object, the system returns to the normal state through the failure recovery process 913 to 916.

The lead object writes a list of failed objects in the recovery message and sends it to the objects in the group (904), then waits for recovery reply messages (905), compares the failure object list in the reply message with the list in the module ( 906) If the contents do not match, the fault list in the module is updated as the sum of the two lists (907), and the process proceeds to step '904' to start the recovery procedure again.

On the other hand, it is checked whether all the reply messages have arrived within a predetermined time (908), if there is a reply message that has not arrived within the predetermined time (909), the object is considered as a failure and added to the failure object list (910) Go to step 904 'and start the recovery procedure again.

If a message is received from all normal objects within a specified time period, if the number of normal objects does not exceed half of the total number of members (911), all normal objects not in the failure object list are stopped, and if more than half of them are new, After the group member is confirmed and sends a member message to each member (912), it returns to normal operation.

In the case of the late object, the received recovery message is combined with the existing failure object list, the existing list is updated, recorded in the recovery reply message, and transmitted to the lead object (913). In addition, when another recovery message arrives (914), the process repeats the step '913', and when the member message is received (915), the group member list is updated (916), and then it returns to the normal operation.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains, and the above-described embodiments and accompanying It is not limited to the drawing.

In the present invention as described above, in the efficient real-time object-oriented distributed platform to provide a variety of fault tolerance services in the group communication support when installing the group communication module between the platform and the network own the own fault tolerance while stopping the service And by supporting the real-time group communication with no delay due to object locking (Locking) there is an effect that can be introduced and used in a real-time system with a distributed platform.

Claims (6)

In a real-time group communication method for supporting a method of implementing fault tolerance through object redundancy by forming a duplicate object group in the object-oriented distributed platform provides fault tolerance services in a networked real-time distributed system, A reception processor receiving a message between objects uses a reception buffer and an alignment buffer to wait for a message having no reception sequence number in the reception buffer, and wait for a message having a reception sequence number in the alignment buffer, while maintaining data integrity. A first step of arranging all messages in the same order in the reception processing unit of all objects in the group to ensure the same; In order to recover lost messages from each object in the group and to give a consistent message receiving sequence number, the token processing unit circulates the token message between group objects, retransmits the missing message, and assigns a new sequence number to sort the received message. And second step of managing in complete form; A third step of determining whether or not to transmit a message for each group model and managing a transmission buffer for fault tolerance in order to increase the efficiency of message transmission in a transmission processor that receives a transmission message from the platform and transmits the message to an destination group; And The fourth step of determining a new group member by the fault handling unit when a failure occurs in some objects in the group Group communication method for supporting fault tolerance services in a real-time object-oriented distributed platform, including. The method of claim 1, The first step is, A fifth step of receiving, by the transmission processor for processing a received object message to the platform and receiving a group object message and storing a new message in the received buffer; Receiving a deficiency message from the token object upon receiving an ACK message from the token object and leading the alignment buffer to the completed array; And A seventh step for all objects to be identically aligned and delivered to the platform in an order to ensure data integrity in the messages in the sort buffer Group communication method for supporting fault tolerance services in a real-time object-oriented distributed platform, including. The method of claim 1, The second step, A fifth step of assigning a correct receiving sequence number to a receiving object message and recovering the deficient message by using the information in the token when the token message arrives; A sixth step of arranging previously transmitted messages among the messages in the transmission buffer; A seventh step of selecting a message, assigning a sequence number, and moving the message to the alignment buffer to ensure data integrity in the reception buffer; And Eighth step of transmitting the token recording the newly updated information to the next object Group communication method for supporting fault tolerance services in a real-time object-oriented distributed platform, including. The method of claim 1, The third step, A fifth step of removing, from the transmission buffer, messages already transmitted in the non-transmission object by the transmission processing unit managing the transmission message; A sixth step of transmitting a head message from the transmittable object to the target group and acknowledging the receipt; And A seventh step of moving a transmission message from the platform to the transmission buffer Group communication method for supporting fault tolerance services in a real-time object-oriented distributed platform, including. The method of claim 1, The fourth step, A fifth step of, when the failure processing unit, which is in charge of determining new object members and recovering to a normal state when an object failure occurs, receives a failure occurrence detection or a failure recovery message, performs recovery steps as a leading object and a late object, respectively; In the case of the leading object, transmitting a recovery message to the late object to derive consensus on new group members; In the case of the late object, after receiving a recovery message, reflecting the opinion of the late object; An eighth step, when the opinions of all the objects agree to form a new object group, when the normal number of object members is less than a majority of the absolute number of all objects in the group, entering an idle state to provide message consistency; And A ninth step of taking the procedure for returning to the normal group when the number is more than half Group communication method for supporting fault tolerance services in a real-time object-oriented distributed platform, including. In a real-time distributed system with a processor, A reception processor receiving a message between objects uses a reception buffer and an alignment buffer to wait for a message having no reception sequence number in the reception buffer, and wait for a message having a reception sequence number in the alignment buffer, while maintaining data integrity. A first function of arranging all the messages in the same order in the reception processing unit of all the objects in the group to ensure the delivery; In order to recover lost messages from each object in the group and to give a consistent message receiving sequence number, the token processing unit retransmits the missing message while circulating the token message between the group objects and assigns a new sequence number to sort the received message. Second function of managing in complete form; A third function of determining whether to send a message for each group model and executing the transmission buffer in a transmission processing unit receiving the transmission message from the platform and transmitting the message to the destination group, and managing the transmission buffer for fault tolerance; And The fourth function of determining a new group member by the fault handling unit when a failure occurs in some object in the group A computer-readable recording medium having recorded thereon a program for realizing this.