CN1558638A - An inter-processor communication method based on message - Google Patents
An inter-processor communication method based on message Download PDFInfo
- Publication number
- CN1558638A CN1558638A CNA2004100007740A CN200410000774A CN1558638A CN 1558638 A CN1558638 A CN 1558638A CN A2004100007740 A CNA2004100007740 A CN A2004100007740A CN 200410000774 A CN200410000774 A CN 200410000774A CN 1558638 A CN1558638 A CN 1558638A
- Authority
- CN
- China
- Prior art keywords
- processor
- message
- link
- active
- state
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a message-based inter-processor communication method which includes an modified tertiary handshaking method, a message and sliding window based transmission method, a managerial method aiming to process node 1+1 master and standby chain circuits, and non-connection message destination processor master and standby status correction method. The invention can realize reliable data transmission and flow control in distributed systems.
Description
Technical field
The present invention relates to communications field distributed processing system(DPS), relate in particular to the method for transmission of messages between the processor that backs up in the system.
Background technology
Require highly reliable to crucial processing node in the distributed processing system(DPS) that the communications field exists, in many designs these keys are handled two concurrent physical processors of node configuration, be presented as a logical process node, has only a processor at ordinary times in work, it is the main processor of using, another piece is not in the service attitude at ordinary times, is backup processor.Backup processor is followed the tracks of the variation of primary processor, from leading with synchronous incremental data of processor and state; Abandon under the situation of main usefulness with processor with processor fault or master main, rapid switching state continues as other processors and serves.In the distributed system other handle nodes active and standby switch with node in, should be able to upgrade the local active and standby state of using of connection rapidly, make this locality data of going out can arrive correct master and use processor.
Existing most host-host protocol does not have the active and standby of considering processor, active and standby management is to be realized by the application layer more than the agreement, because application layer with respect to protocol layer for the communications status of distributed system understand fewer, upper layer software (applications) for active and standby be disposable with unusual flow processing majority such as state switching, be difficult to abnormal conditions such as losing of critical message in the more complete adaptation reversed process, realize relatively difficulty of activestandby state transition adaptively, such as, when connectionless information sent, application layer was surveyed activestandby state to backup processor after just being difficult to accomplish send the destination packet of having fixed a number.
And application layer is diversified, the state renewal majority of being finished by application is to realize at specific upper layer application and specific underlying protocol, if they go to participate in the active and standby management of communication link, can cause a lot of the repetition but skimble-scamble realization, the variation of the structure of inconvenient software transplanting and adaptation product.
Summary of the invention
The technical problem to be solved in the present invention is to overcome the poor reliability of existing inter-processor communication method existence and the defective of adaptivity difference, a kind of message based inter-processor communication method is provided, be applied to have in the processing node in the distributed processing system(DPS) of active and standby processor, when system's operate as normal, can provide reliable, orderly message transmission and flow control for system; Take place when unusual at processing node, can upgrade with state by adaptive the active and standby of link that carry out.
In order to solve the problems of the technologies described above, the invention provides a kind of message based inter-processor communication method, be applied to have in the logical process node in the distributed processing system(DPS) of active and standby processor, be included in link setup flow process, transmission of messages flow process, active state-maintenance flow process that link layer is realized, wherein:
Described link setup flow process is finished by three-way handshake, may further comprise the steps:
(a1) connect the initiator and initiate handshake for the first time, carry the physical address and the activestandby state of local terminal;
(a2) recipient returns the second handshake signal to the initiator, and carries local terminal physical address and activestandby state according to the active and standby state of using of the activestandby state determined link in the first time handshake;
(a3) after the initiator receives the second handshake signal, the activestandby state that connects link according to its activestandby state that carries decision, the related news that send in the buffer queue are transferred to main transmit queue with connection, send handshake for the third time to the recipient, and beginning sends data on connecting;
(a4) after the recipient received for the third time handshake, beginning is transceive data on connecting;
The sliding window of described transmission of messages flow process employing realizes that the transmitting-receiving of message is synchronous, reliability guarantees and flow control, may further comprise the steps:
(b1) initial bag sequence number and initial window size are determined in two ends when connection is set up, and the message sequence number that the initial maximum of transmit leg can send is that local terminal initial package sequence number adds the initial window size that the opposite end is given;
(b2) transmit leg detects the size of send window, retransmission queue, transmit queue in real time, and the recipient detects the size of receive window, formation in order, out of order formation in real time;
When (b3) transmit leg maximum message sequence number that can send and the maximum message sequence number that sent equate, the send window stop motion; When the sequence number of the receivable maximum message of recipient and the maximum message sequence number that has received overlapped, receive window just stopped operating, and makes the send window stop motion;
(b4) recipient enlarges receive window by routine after upper layer application is handled message, and to opposite end send window update notification, drives the transmitting-receiving flow process again;
Described active state-maintenance flow process comprises that the processor activestandby state initiatively switches and switch routine unusually, and the wherein said routine of initiatively switching may further comprise the steps:
(c1) master begins notice with processor to backup processor transmission masterslave switchover, after backup processor is received, begins to reply with processor transmission masterslave switchover to main;
(c2) main with processor finish self switch processing after, send main the commentaries on classics to backup processor and be equipped with ready message, after backup processor is received, be equipped with ready replying to main commentaries on classics of primary processor response;
(c3) backup processor finish self switch processing after, switch with state to main, and send link-recovery message to all connected other processors;
(c4) after other processors were received described link-recovery message, identification message came source processor, finish in case of necessity active link to the switching of connected link mutually of this processor, it is mutual to recover normal messages.
Described switch unusually routine backup processor know main with processor unusual after, (c3) and (c4) execution set by step.
Said method can have following characteristics: in the described link setup flow process, for the first time and for the third time the handshake processor that sends message by initiator's desire is dealt into the active and standby processor of target processing node, and described second handshake signal is sent to respectively by recipient's active and standby processor and sends the processor of handshake for the first time.
Said method can have following characteristics: described step (a2) afterwards, whether recipient's active and standby processor is also judged to have with another processor of initiator and is connected, if do not have, connects with it by three-way handshake equally.
Said method can have following characteristics: described step (a2) before, the recipient judges at first whether local terminal exists the link setup flow process to the opposite end, if there is no, execution in step (a2), if exist, judge the size of two ends physical address again, according to the relation of the physical address size of setting with link setup, the link setup request of refusal opposite end, perhaps execution in step (a2).
Said method can have following characteristics: in the described transmission of messages flow process, if step (b2) detects when having one to be zero in the non-vanishing and transmit queue of the retransmission queue of transmit leg, the send window, start and retransmit; All non-vanishing with send window and when the link of connection was arranged, initiation message sent at transmit queue.
Said method can have following characteristics: in the described transmission of messages flow process, if step (b2) detects the retransmission queue of transmit leg and transmit queue when being zero, send keep-alive message to the opposite end, carry the active and standby state of using of local terminal processor in the message; The opposite end processor obtains the active and standby with behind the state of local terminal processor, upgrades the active and standby state of using of the link of this logical process node, realizes the migration to new active link of the message that cushions on the link simultaneously.
Said method can have following characteristics: in the described transmission of messages flow process, be zero and transmit queue when non-vanishing starts window and surveys if step (b2) detects the re-transmit queue of transmit leg and send window.
Said method can have following characteristics: described activestandby state is initiatively switched in the routine, and is in step (c2) with (c3), further comprising the steps of:
(c21) the main processor of using sends link switchover orders to all connected other processors;
(c22) after other processors receive described link switchover order, carry out the pre-switch of link, with message transfers to the buffer queue of masterslave switchover.
Said method can have following characteristics: in the described step (c22), other processors also are provided with a masterslave switchover waiting timer, if this timer expiry then is defaulted as overtime message a link-recovery message from current reserve link.
Said method can have following characteristics: described active state-maintenance flow process also comprises link fracture routine: if a link fracture on processor, automatically the master with link switches on another link of this processor with state, and finishes the message that cushions on the link from the migration of old active link to new active link.
Said method can have following characteristics: described active state-maintenance flow process also is included in carries out the routine that activestandby state is safeguarded in the message transmitting procedure, this routine may further comprise the steps:
The destination of carrying message in the service message that sends is main using or the information of backup processor;
When backup processor receives the message that send to main usefulness processor, send active and standby renewal prompting to the informed source processor;
The informed source processor is pointed out according to described renewal, upgrades the active and standby state of using of the link of this logical process node, and finishes the migration that the link buffering goes up message.
Said method can have following characteristics: also comprise the correction flow process that connectionless information is sent the purpose processor, may further comprise the steps:
(d1) processor that sends connectionless information is counted respectively the connectionless information that oneself mails to the current reception processing of each logical process node, when count value reaches set point, to another processor transmit status probe message of purpose logical process node;
(d2) another processor in the logical process node is received and is changed if inquire this processor state after the described probe message, then responds the change message of a main and standby relation to the opposite end;
(d3) after the processor that sends connectionless information receives described change message, carry out this processor on the switching of main/slave link, the processor that changed course sends described change message sends connectionless information, repeats above-mentioned flow process later on.
Said method can have following characteristics: described connectionless information is to mail to the main of logical process node to use processor, and described state detection message is the backup processor that mails to the opposite end, and the concurrent physical processor that comprises the recipient is main using or standby information.
As from the foregoing, link establishment provided by the invention, management method and method for message transmission have guaranteed transfer of data and flow control reliably in the distributed system; By to the active and standby state correction flow process of using of the adaptive communication link of activestandby state transition in the standby system, when 1+1 backup node activestandby state changes in distributed system, make other processors can finish reselecting of link fast, and provide the reliability of transmission of messages in the state renewal process to guarantee.In addition, it is a kind of when 1+1 backup node activestandby state changes that the present invention also provides, for other processors provide a kind of adaptive update method to the selection of the connectionless information transmission destination processor of this node.The present invention realizes at link layer, thereby can adapt to the variation of the structure of product easily.
Description of drawings
Fig. 1 is the schematic diagram of processor physics addressing of the present invention and logic addressing method;
Fig. 2 is a networking schematic diagram of the embodiment of the invention;
Fig. 3 is the schematic flow sheet that connection of the present invention is set up;
Fig. 4 is the handling process schematic diagram of send window of the connection transmit leg of the embodiment of the invention;
Fig. 5 is the connection recipient's of the embodiment of the invention the handling process schematic diagram of receive window;
Fig. 6 is the embodiment of the invention active and standby with under the relation situation of switching in the logical process node, the schematic flow sheet of the active and standby usefulness state correction of the link on other processors;
When being the transmitting-receiving of embodiment of the invention processing connectionless information, corrects Fig. 7 the schematic flow sheet of active with relation.
Embodiment
The inventive method based on system as shown in Figure 1, single processor uses the independent nybble network address to identify in the distributed system, this sign is called physical address here; Use like-identified to address each to the processor that constitutes main and standby relation, this sign is called logical address.Physical processing unit among the figure (including processor) identifies with 1.1.1,1.1.2,1.2.1,1.2.2,1.3.1 and 1.3.2, and the logical process node identifies with 1.1,1.2 and 1.3.The processor that does not back up in the system thinks also that logically there is a back-up processing device in it, and just this processor is not online.Connection between the active and standby processor of other processors and a pair of formation is called the partner mutually and connects.
For the sake of clarity, present embodiment will describe in conjunction with the networking situation among Fig. 2.Among the figure, processor 1,2 and processor 3,4 are active and standby each other, with the active and standby processor in opposite end a link (L1-L2 is arranged respectively, L3-L4) connect, connection link on the each processing unit has active and standby branch, primary, spare transceiver module in the active and standby processor all can receive message or transmission and activestandby state relevant information, but business datum sends from active link.Just often, the sending direction of business datum points to and the main active link that links to each other with processor in opposite end, supposes that current master is processor 1 and 3 with processor, and then the active link of processing unit 1～4 is respectively L1, L3, L1 and L2.If certain in particular cases backup processor 4 exist business datum to need to send, also can only mail to the main of opposite end with processor 1.A transmit queue, the message that the buffer memory upper strata need send are disposed in each connection.
The embodiment of the invention has comprised following method:
Improved three-way handshaking method in order to realizing the foundation of unique link between two processors, and is determined the active and standby state of using of link in the link establishment process.
Based on the transmission method of message and sliding window, realize the transmission of message, and in data transmission procedure, carry out the active and standby maintenance of link with state, make that each processor in the distributed system can be adaptive for holding the mutual of connection message.
At the active and standby link management method of processing node 1+1, in order to the active and standby state of using at link maintenance link in the rise time, and processor is active and standby with the diffusion of state in distributed system.
There is not the correct mutual of connectionless information between the processor of link in the purpose processor activestandby state detection method of connectionless information, two of adaptive maintenances.
To elaborate to the specific implementation of the embodiment of the invention below to above-mentioned every method.
The foundation of link
Set up unique connection between the processor that it is two needs communications in the distributed system that the present invention adopts a kind of improved three-way handshaking method.The link setup process is announced the physical address and the activestandby state of local terminal mutually, adopts the relatively superseded link setup flow process that repeats of method of the two stimulating food reason of link setup address, chooses the active link of opposite end logical process node by specific routine.As shown in Figure 3, the physical address of supposing the connection initiator among the figure is greater than the physical address that connects the recipient, and the link establishment process may further comprise the steps:
Connect the initiator and send connection request, promptly handshake 2.1 for the first time to the active and standby processor of the opposite end that needs to connect, is carried the physical address and the activestandby state of local terminal, step 100;
The active and standby processor that connects the recipient receives after the handshake first time, checks whether local terminal has existed the flow process to the opposite end link setup, if exist, carries out next step, otherwise, execution in step 130, step 110;
Judge that the local terminal physical address is whether greater than the physical address of opposite end, if the rejection connection request is waited for the response of opposite end to the local terminal connection request; Otherwise, carry out next step, step 120;
Receive connection request, change passive link setup state over to, the active and standby state of using according to the opposite end activestandby state determined link in the first time handshake, and the structure connection is accepted to reply, it is second handshake signal 2.2, send to the active and standby processor of opposite end, carry the physical address and the activestandby state of local terminal simultaneously, step 130; After this step, whether recipient's active and standby processor can also be judged to have with another processor of initiator and be connected, if do not have, connects with it by three-way handshake equally.
Connect the initiator after receiving the second handshake signal, be about to the activestandby state of the connection link of formation according to the activestandby state decision of carrying in the signal, transfer to the related news that send in the buffer queue main with on the transmit queue that connects, and send the connection acceptance response to the concurrent physical processor that connects the recipient, i.e. handshake 2.3 for the third time, in the connection of setting up, send data, step 140 then;
After connection recipient's concurrent physical processor obtains this response, think link establishment can begin transceive data on connecting, step 150.
In this process, handshake for the first time all might be sent in two ends, for the handshake 2.4 first time that connects among the figure that the recipient sends, wait for the opposite end processor reply this request during, because link setup request 2.1 arrives, and ask 2.1 come the physical address of source physical address greater than this processor, handshake 2.4 meetings first time that the recipient sends be refused by the opposite end processor, be that the opposite end does not respond this connection request, this processor receives during wait acknowledge after the connection request of opposite end, the link setup process that termination is immediately initiated oneself changes passive link setup state over to; If a side's of any initiation link setup link setup request has obtained the response of opposite end, then the link setup request initiated in the time after in the link setup process of the opposing party all can not get replying because be rejected.
Describe below in conjunction with Fig. 2, suppose that processor 1,2 belongs to the logical process nodes X, processor 3,4 belongs to logical process node Y.
Link setup is driven by data-message, promptly as long as any one processor 1 or 2 among the X has business datum will send to another logical process node Y, can connect to opposite end logical process node Y, suppose processor 1 for connecting the processor of initiating, processor 1 will be initiated the link setup flow process respectively to the processor 3 and 4 among the logical process node Y.The link setup flow process adopts above-mentioned three-way handshake agreement, carried initiator's physical address (being the left position or the information of right position) and be main usefulness or standby information in shaking hands the first time of initiating link setup, the processor 3,4 of the opposite end service message that can send to the logical process nodes X according to having shaken hands clearly the first time that obtains separately is send to processor 1 or 2 like this.Processor 3,4 is received after the link setup request among the logical process node Y, respectively to processor 1 response second handshake, carry physical address of this processor and the information of whether leading usefulness in the handshake information, also should to send to processor 3 with regard to the clear and definite business datum that mails to logical process node Y still be 4 to processor 1 like this; After processor 3, the 4 response second handshakes, judge with the logical process nodes X in another processor 2 whether have and be connected, if do not have, initiate same three-way handshake flow process to processor 2, processor 2 is also clear and definite like this, and service message should send to processor 3 still is 4, even masterslave switchover takes place later on the logical process nodes X, also can continue concurrent physical processor correct in Y and send service message.
The transmission control of message
The present invention is synchronous by the transmitting-receiving that sliding window carries out message, reliability guarantees and flow control is handled.The sequence number of message adopts the double byte signless integer to represent that span is 0～65535.Determine initial bag sequence number and initial window size in the three-way handshake process that connects foundation, thereafter the sequence number of all bags all must increase progressively in initial bag sequence number, the bag sequence number that the initial maximum of transmit leg can send is that local terminal initial package sequence number adds the initial window size that the opposite end is given, and the initial package sequence number of agreement window is not less than 4.
As shown in Figure 4, each of transmit leg connects the transmission buffering area to be described with send window, retransmission queue, duplicate acknowledgment interval and transmit queue, among the figure, the maximum message sequence number 3.1 that can send and the difference of maximum transmitted message sequence number 3.2 are send windows 3.5, sign can also send the counting of how much bar message to the opposite end, is used for flow control; Between the message sequence number 3.3 that does not have to confirm of maximum transmitted message sequence number 3.2 and minimum is retransmission queue 3.6, and the storage organization that reply the Messages-Waiting opposite end that cache user has sent is replied and then removed the item that retransmission queue is replied; Between the minimum message sequence number 3.3 that does not have affirmation and the minimum effective acknowledged sequence number 3.4 is duplicate acknowledgment interval 3.7, if the affirmation sequence number that the affirmation message that transmit leg is received comprises drops on this interval, it then is a duplicate acknowledgment, can do suitable processing as required, perhaps do not deal with, present embodiment does not deal with the affirmation message in this interval; The minimum effective acknowledged sequence number 3.4 and the difference of the maximum message sequence number 3.1 that can send are the sequence number of the message of maximized window, transmit leg keeps this value and is used for judging whether acknowledge message has comprised effective acknowledged sequence number, whether handles this acknowledge message with decision.
Present embodiment connects transmit queue of configuration for each, the storage organization of the message that the buffer memory upper strata need send, when the transmission condition is satisfied in connection, the message in the transmit queue will be sent and be transferred to retransmission queue successively, and the size of transmit queue cushion space can be adjusted as required.
As shown in Figure 5, recipient's reception buffer memory is made up of out of order formation, orderly formation and receive window, among the figure, the difference of receivable maximum message sequence number 4.1 and the maximum message sequence number 4.2 that received is receive windows 4.6, sign can also receive the counting of how much bar message after a certain message, be used to limit opposite end excessive transmission message when local terminal is busy; Between the message sequence number 4.3 that does not arrive in order of maximum message sequence number 4.2 that has received and minimum is out of order formation 4.7, is used for the storage organization that buffer memory arrives the message of local terminal in advance; The minimum message sequence number 4.3 that does not arrive in order and to have submitted between user's the maximum message sequence number 4.4 be orderly formation 4.8, all arrive the message of local terminal according to the order of sequence and directly distribute upper layer application, the message buffering that the upper strata has little time to receive processing continues to distribute when waiting for the free time in this formation; The transmit leg minimum may not be identified the sequence number 4.5 of message and maximum message sequence number 4.2 differences that received are the sequence number of the message of maximized window, the recipient has received 4.2 identification message of maximum message sequence number, show that then transmit leg received that correctly comprising acknowledged sequence number is the affirmation message shown in 4.5, otherwise transmit leg can not to send sequence number be 4.2 message.
In addition, each logical process node configuration has one to switch buffer queue, is used for during the logical process node masterslave switchover of buffer memory opposite end to the main message that sends with processor in opposite end.
Table 1 shows the move into processing logic relation of the connection transmit leg of the embodiment of the invention in the connection status that receives various message.Table 2 shows the move into processing logic relation of recipient in the connection status that receives various message that connect." 0 " expression formation or window size in the table are zero, and " 1 " expression formation or window size are non-vanishing, and identical agreement is used in the similar description of back.
Table 1
| Connection status | The action of moving into | ||
| Re-transmit queue | Transmit queue | Send window | |
| ????0 | ????0 | ????0 | Start keep-alive |
| ????0 | ????0 | ????1 | Start keep-alive |
| ????0 | ????1 | ????0 | Starting window surveys |
| ????0 | ????1 | ????1 | 5.1 * |
| ????1 | ????0 | ????0 | Start and retransmit |
| ????1 | ????0 | ????1 | Start and retransmit |
| ????1 | ????1 | ????0 | Start and retransmit |
| ????1 | ????1 | ????1 | ??5.2 * |
Table 2
| Connection status | The action of moving into | ||
| Out of order formation | Sequential queue | Receive window | |
| ????0 | ????0 | ????0 | Remove link detecting timer expiry counting, reset the link detecting timer |
| ????0 | ????0 | ????1 | |
| ????0 | ????1 | ????0 | |
| ????0 | ????1 | ????1 | |
| ????1 | ????0 | ????0 | |
| ????1 | ????0 | ????1 | |
| ????1 | ????1 | ????0 | |
| ????1 | ????1 | ????1 | |
When maximum message sequence number that transmit leg can send and the maximum message sequence number that has sent equated, send window was with regard to stop motion (being " 0 "), and the maximum message sequence number that transmitting terminal can send is by control of recipient's receive window and adjustment.
When the sequence number of recipient's receivable maximum message and the maximum message sequence number that received overlapped, receive window just stopped operating (for " 0 "), and causes the send window stop motion.
Upper layer application is handled after the message, enlarges receive window by routine, and to opposite end send window update notification, drives the transmitting-receiving flow process again.
Receive and situation that send window does not move under, both sides can carry out re-transmission, window detection and these three kinds of actions of link keep-alive of message, will carry out the processing of moving into of state after Link State is moved into a new state.Describe the processing of moving into of every kind of state in the table in detail.
For in transmit leg, specifically, when in the non-vanishing and transmit queue of retransmission queue, send window, having one to be zero, start and retransmit; When retransmission queue and transmit queue are zero, start keep-alive, promptly send keep-alive message to the opposite end, carry the active and standby state of using of local terminal processor in the message; At re-transmit queue and send window is zero and transmit queue when non-vanishing, starting window surveys, transmit leg send window probe message is to carry out in the window updating message of in time not receiving the recipient, this is to lose unusually on network for fear of recipient's window updating message, causes can not upgrade in time window and carry out new round transmission flow of transmit leg; And transmit queue and send window are when all non-vanishing, be that link can send condition of information, in case because external condition causes the transmit status machine to enter this state, send continuously the message that is buffered on the transmit queue immediately, move to transmit queue and send window up to state machine and be not the state of " 1 " simultaneously.
The passive reception data of recipient, and reply, expand and reply according to the data character of receiving, receive and handle after a piece of news, receive window is enlarged, when 0 expanded certain limit to, the notice transmit leg carries out window to be upgraded at receive window.
The recipient has only the processing of moving into of a kind of state, removes the link detecting timer on the link exactly after receiving message each time, and removes the time-out count of this timer, resets this timer then, and keep-alive message also can cause aforementioned removing action.After the continuously certain number of times of link detecting timer is overtime, think that link ruptures, doing fracture and handling.
The maintenance of link life span activestandby state
In the establishment of connection process, pass through preceding method, two sides that participate in link establishment will can upgrade with state the active and standby of link according to the first time handshake or second handshake signal, and the message that realizes cushioning on the link is from the old main migration of using the new master of (new is standby) chain road direction with (old is standby) link.
In the present embodiment, if the fracture of in the link of partner's link each other (can judge) by timer, automatically the master with link switches on partner's link with state, and finishes the message that cushions on the link is used (old is standby) link with the new master of (new is standby) chain road direction from old master migration.
The partnership of link is at processor, in Fig. 2, processor 1: L1 and L2 are constituted partner's link relation, processor 2: L3 and L4 are constituted partnership, processor 3: L1 and L3 are constituted partnership, processor 4: L2 and L4 are constituted partnership.The fracture of L1 will influence reselecting of processor 1 and 4 pairs of active links of processor, and processor 1 is led with the mark that connects and is set to L2, and processor 4 is set to L3 main with the mark that connects.Can avoid processor to go off like the snuff of a candle like this or connect fracture the time not on the throne, cause main and standby relation to be difficult to correct.In the present embodiment, the processor that the fracture of link only influences both link ends is reselected active link, does not change the active and standby state of using of both link ends processor, and active and standby maintenance with state is finished by other module synthesis various factorss.
In the present embodiment, in the transmission course of message, also carry out the active and standby maintenance of link with state, the destination of having carried message in service message is main using or the information of backup processor, backup processor receives and send to when leading the message of using processor, to send active and standby renewal prompting to the source processor that comes of message, the informed source processor can be according to the active and standby state of use of the link that upgrades this logical process node, and finishes the migration of message on the link buffering.
For example, for example in the processing mode of link fracture, can cause obscuring of main and standby relation in a short time, if but backup processor can be notified opposite end main and standby relation mistake after receiving service message, the opposite end will be revised, and finally can correctly adapt to the configuration and the active and standby situation of processor.
In the present embodiment, between processor, exist to connect, and during link idle, both link ends sends keep-alive message to the opposite end mutually, carries the active and standby state of using of local terminal processor in the message; The opposite end processor obtains the active and standby with behind the state of local terminal, upgrades the active and standby state of using of the link of this logical process node according to it, and simultaneously, the message that realizes cushioning on the link is from the old main migration of using (old is standby) link with the new master of (new is standby) chain road direction.A side who receives keep-alive message does not need echo reply, only needs to carry out and the same process in opposite end, regularly sends keep-alive message to the opposite end in link idle.
Sending keep-alive message when temporarily not having service message mutual between processor, is to be diffused into the opposite end for as early as possible the processor main and standby relation with local terminal, avoids service message to lose.For example processor 1 has become standby; processor 2 has become the main time spent; if for a certain reason; all not having of processor 1 and processor 2 can successful notifying communication module; just simple modification the activestandby state mark of veneer; according to above-mentioned flow process; communication module has just lost the chance that once drives the flow process of opposite end active/standby link switchover; but rely on keep-alive and can in time correct this mistake; such design has increased system survivability; make that the switching of communication link is an adaptive process, do not rely on other modules.But there is not keep-alive message can in subsequent treatment, correct the active and standby state of using of link yet.
In the present embodiment, support the active of active and standby processor to switch and switch unusually.Initiatively switch and be meant that the master initiatively abandons the master with processor and uses attitude, backup processor enters operating state by flow process; Unusually switch be meant main with processor because fault can not work on, backup processor obtain main with after the unusual information (as, pass through hardware corridor), directly initiate to notify other processors to switch masters and arrive this processor with active link.The step of switching unusually is a part of the step of initiatively switching.In the present embodiment, the active and standby maintenance with state of processor is finished by other module synthesis various factorss, and communication just accepts to switch order, and trigger to oneself have the opposite end that is connected to send the order of link switchover respectively.
Shown in Figure 6 is the flow process that present embodiment is initiatively switched, and may further comprise the steps:
The master sends masterslave switchover to backup processor earlier with processor and begins to notify 6.1, consults the beginning masterslave switchover, step 200;
Backup processor is after receiving the masterslave switchover notice, begin to reply 7.2 to main with processor transmission masterslave switchover, if the master does not receive response message 7.2 within a certain period of time with processor, think that then backup processor is not on the throne, switch failure, carry out failure and handle routine, otherwise carry out next step, step 210;
Main have other processors that are connected to send link switchover order 7.3.1,7.3.2,7.3.3, step 220 with processor to all and oneself;
Other processors receive after the order, carry out the pre-switch of link, message is transferred on the buffer queue of masterslave switchover from original active link, and set masterslave switchover waiting timer, step 230;
Main with processor finishing self switch after the task handling, send main the commentaries on classics to current backup processor and be equipped with ready message 7.4, if backup processor is not received this message within a certain period of time, think and switch failure, carry out failure and handle routine, otherwise carry out next step, step 240;
Current backup processor receive main change be equipped with ready message 7.4 after, to the primary processor response main change to be equipped with readyly reply 7.5, step 250;
Current backup processor finish self switch processing after, switch with state to main, and there be other processors transmission link-recovery message 7.6.1,7.6.2, a 7.6.3 that is connected to all and oneself, other processors are received after this message, identification message comes source processor, in case of necessity with on the link that the active link mark switches to this processor links to each other, and shifts the transmission buffered message queue to the formation of this connection, the recovery normal messages is mutual, step 260.
If the masterslave switchover waiting timer that is provided with on other processors in step 230 is overtime, then overtime message is defaulted as one from link-recovery message of current reserve link, and carries out the processing of link-recovery message.The purpose that this timer is provided is active and standbyly to switch with state in order to finish to current backup processor time enough, avoids too early handoff links and sends that message buffered does not cause information drop-out because state matches on the link after the opposite end.
In conjunction with Fig. 2, at processor 1 in the masterslave switchover process of processor 2, processor 1 sends the masterslave switchover notice to 2, after receiving 2 reply, send the link switchover order respectively to processor 3,4, because the L1 fracture by L1, L2, processor 3 can not receive this message, do not carry out the pre-switch of link, processor 4 is backup processor, receives after this message the active link pre-switch to L4; Processor 1 is equipped with ready message to main commentaries on classics of 2 transmissions then, after processor 2 is replied, finishes alternation of hosts's status change fully, send link-recovery message by L3, L4 to processor 3,4 then, the master switches to L3 with active link after receiving this message with processor 3, recovers proper communication.
Present embodiment switch unusually flow process as in the square frame among Fig. 6 the part shown in, be a part of initiatively switching flow process.Backup processor obtain main with the unusual information of processor, finish inner to main with after the process of transition between states, other processors transmission link-recovery message that are connected to all and own existence; Other processors receive that identification message comes source processor after the link-recovery message, on the link that active link switches to this processor links to each other, and shift the transmission buffered message queue to the formation of this connection, begin normal interacting message.
For example, if processor 1 fault, its reversed process is: processor 2 is found processor 1 fault by hardware corridor, finish alternation of hosts's status change fully, send link-recovery message by L3, L4 to processor 3,4 then, the master switches to L3 with active link after receiving this message with processor 3, and shift to send and cushion the formation of message to this connection, recover proper communication.
If the order that occurs switching in the masterslave switchover process is lost, be equipped with ready message and link-recovery message as main the commentaries on classics, what perhaps other reason had caused the opposite end link is active and standby active and standby inconsistent with state with the state and the processor of main end, the embodiment of the invention can adopt following method to guarantee that the opposite end can be corrected and local terminal between the active and standby state of using of connection:
If do not connect between two processors, and it is mutual that service message does not take place, the mistake of so this main and standby relation can be tolerated, system not damaged;
If do not have connection between two processors, and service message has taken place mutual the time, then triggered the link setup flow process, obtained correct active state, made the active and standby of local terminal link be consistent with state with state and the active and standby of opposite end processor by the link setup flow process;
Connect if exist between two processors, and do not exist service message mutual on the link, then can start the link keep-alive, between the survival period of link, active and standby the using of correcting the link of the every other processor that existence is connected with this logical process node by the keep-alive message of link connects;
Connect if exist between two processors, and exist service message mutual on the link, the destination of carrying by service message is main using or the information of backup processor, can correct with this logical process node to exist active and standby the using of the link of the every other processor that is connected to connect;
For example, at above-mentioned processor 1 in the process of processor 2 masterslave switchovers, for a certain reason, processor 2 may think that processor 4 is main usefulness, is actually processor 3 and is only and main uses processor, is not having under the mutual situation of service message, this mistake can be tolerated, in case it is mutual that service message takes place processor 2 and 4, after backup processor 4 receives service message, can in time notify 2 main/slave links to select to make a mistake.If problem takes place and ruptures in the link L4 between 2 and 4, then can in time be switched to L3 according to link fracture treatment principle active link, corrected mistake equally.
Connectionless information sends the correction of purpose processor
Do not need the service message through above-mentioned controlling links when connectionless information is meant transmission, this message does not need reliable delivery, so need not to set up the connection that can guarantee reliable transmission for their transmission.When business can be born bigger packet loss, can adopt connectionless information to send data, be used in some media flow transmission, generally only mail to the main processor of using.With above-mentioned the same based on the message that is connected, the concurrent physical processor that has indicated the recipient in the connectionless information should be main with or standby.
When the master to a logical process node sends connectionless information with processor, if it is main unusual suddenly during sending with processor, because directly there is not connection in current processor, backup processor does not have the information of other processors at this moment, just can not correct the activestandby state record that the phase of other processors should the logical process node, it is main with after the processor causing the backup processor transition, and connectionless information can not arrive new master all the time and use processor.
The present invention has done specialized designs to connectionless information, can adapt to the variation of the activestandby state of other processors in order to guarantee when having only connectionless information mutual each processor in each system.As shown in Figure 8, when present embodiment is corrected the activestandby state that writes down on the processor that sends connectionless information, may further comprise the steps:
Connectionless information 7.1.1, the 7.1.2 of other unit, 7.1.3 mail to the master and use the unit, are used for transmitting data, step 600 between two processors;
The processor that sends connectionless information is counted respectively the connectionless information that oneself mails to each logic processor, when the connectionless information counting arrives a limit value on the logic processor, as 10, this processor sends main with processor probe message 72.1 to the current backup processor of purpose logical process node, main processor, the step 610 used of inquiry;
After a processor receives that the master is with processor probe message 7.2.1 in the logical process node, at first inquire about the activestandby state of oneself, find that this processor has become the main state of using, and then responds the change message 7.3.1 of a logical process node main and standby relation, step 620 to the opposite end;
The processor that sends connectionless information receives after the activestandby state change message, the switching of main/slave link on carry out this processor, the main processor (the former backup processor among the figure) of using after changing of changed course purpose logical process node sends connectionless information 7.1.4,7.1.5,7.1.6, finish the correction that sends the purpose processor, step 630;
When the connectionless information counting arrived a limit value, connectionless information sent processor and sends main with processor probe message 7.2.2, step 640 to the backup processor after changing (the former main processor of using among the figure) of purpose logical process node;
Connectionless information sends processor and sends after this detection, does not receive logical process node main and standby relation change message, continues to send connectionless information 7.1.7 etc., step 650 to the current main processor (the former backup processor among the figure) of using.
In the present embodiment, establishing the connectionless information limit value is 10, then main with processor unusual after, the connectionless information transmit leg may need to lose and just can send a message to new master after 10 connectionless informations and use processor.And connectionless information flow on the net can increase by 1/10, promptly 10%.
In addition, the recipient also can lead and use processor, also can be set at backup processor.If send connectionless information with processor to main, send so after 10 bags, just send 1 detective and survey message to backup processor, can be exactly connectionless information; If send connectionless information to standby processing, send so after 10 bags, just send 1 detective survey message with processor to main.Since indicated recipient's concurrent physical processor in the connectionless information and should be main with or standby, the recipient is after receiving connectionless information like this, judge the recipient's who indicates in the message active and standby the active and standby whether consistent of state and this processor of using: if then correctly receive with state; Otherwise,, allow the opposite end switch the sending direction of connectionless information to activestandby state updating message of opposite end response.
In sum, the invention provides a kind of link establishment, management method, and a kind of method for message transmission, the reliable transfer of data in the distributed system guaranteed.
In addition, by to the active and standby state correction flow process of using of the adaptive communication link of activestandby state transition in the standby system, when 1+1 backup node activestandby state changes in distributed system, make other processors can finish reselecting of link fast, and provide the reliability of transmission of messages in the state renewal process to guarantee.
In addition, it is a kind of when 1+1 backup node activestandby state changes that the present invention also provides, for other processors provide a kind of adaptive update method to the selection of the connectionless information transmission destination processor of this node.
In addition, the present invention realizes at link layer, thereby can adapt to the variation of the structure of product easily.
The present invention is applied in the 3G communication system, has obtained reasonable stability and laser propagation effect in the previous experiments.
Claims (13)
1, a kind of message based inter-processor communication method is applied to have in the logical process node in the distributed processing system(DPS) of active and standby processor, is included in link setup flow process, transmission of messages flow process, active state-maintenance flow process that link layer is realized, wherein:
Described link setup flow process is finished by three-way handshake, may further comprise the steps:
(a1) connect the initiator and initiate handshake for the first time, carry the physical address and the activestandby state of local terminal;
(a2) recipient returns the second handshake signal to the initiator, and carries local terminal physical address and activestandby state according to the active and standby state of using of the activestandby state determined link in the first time handshake;
(a3) after the initiator receives the second handshake signal, the activestandby state that connects link according to its activestandby state that carries decision, the related news that send in the buffer queue are transferred to main transmit queue with connection, send handshake for the third time to the recipient, and beginning sends data on connecting;
(a4) after the recipient received for the third time handshake, beginning is transceive data on connecting;
The sliding window of described transmission of messages flow process employing realizes that the transmitting-receiving of message is synchronous, reliability guarantees and flow control, may further comprise the steps:
(b1) initial bag sequence number and initial window size are determined in two ends when connection is set up, and the message sequence number that the initial maximum of transmit leg can send is that local terminal initial package sequence number adds the initial window size that the opposite end is given;
(b2) transmit leg detects the size of send window, retransmission queue, transmit queue in real time, and the recipient detects the size of receive window, formation in order, out of order formation in real time;
When (b3) transmit leg maximum message sequence number that can send and the maximum message sequence number that sent equate, the send window stop motion; When the sequence number of the receivable maximum message of recipient and the maximum message sequence number that has received overlapped, receive window just stopped operating, and makes the send window stop motion;
(b4) recipient enlarges receive window by routine after upper layer application is handled message, and to opposite end send window update notification, drives the transmitting-receiving flow process again;
Described active state-maintenance flow process comprises that the processor activestandby state initiatively switches and switch routine unusually, and the wherein said routine of initiatively switching may further comprise the steps:
(c1) master begins notice with processor to backup processor transmission masterslave switchover, after backup processor is received, begins to reply with processor transmission masterslave switchover to main;
(c2) main with processor finish self switch processing after, send main the commentaries on classics to backup processor and be equipped with ready message, after backup processor is received, be equipped with ready replying to main commentaries on classics of primary processor response;
(c3) backup processor finish self switch processing after, switch with state to main, and send link-recovery message to all connected other processors;
(c4) after other processors were received described link-recovery message, identification message came source processor, finish in case of necessity active link to the switching of connected link mutually of this processor, it is mutual to recover normal messages;
Described switch unusually routine backup processor know main with processor unusual after, (c3) and (c4) execution set by step.
2, communication means as claimed in claim 1, it is characterized in that, in the described link setup flow process, for the first time and for the third time the handshake processor that sends message by initiator's desire is dealt into the active and standby processor of target processing node, and described second handshake signal is sent to respectively by recipient's active and standby processor and sends the processor of handshake for the first time.
3, communication means as claimed in claim 1 is characterized in that, described step (a2) afterwards, whether recipient's active and standby processor is also judged to have with another processor of initiator and is connected, if do not have, connects with it by three-way handshake equally.
4, communication means as claimed in claim 1, it is characterized in that, described step (a2) before, the recipient judges at first whether local terminal exists the link setup flow process to the opposite end, if there is no, execution in step (a2), if exist, judge the size of two ends physical address again, according to the relation of the physical address size of setting with link setup, the link setup request of refusal opposite end, perhaps execution in step (a2).
5, communication means as claimed in claim 1 is characterized in that, in the described transmission of messages flow process, if step (b2) detects when having one to be zero in the non-vanishing and transmit queue of the retransmission queue of transmit leg, the send window, starts and retransmits; All non-vanishing with send window and when the link of connection was arranged, initiation message sent at transmit queue.
6, communication means as claimed in claim 1, it is characterized in that, in the described transmission of messages flow process, if step (b2) detects the retransmission queue of transmit leg and transmit queue when being zero, send keep-alive message to the opposite end, carry the active and standby state of using of local terminal processor in the message; The opposite end processor obtains the active and standby with behind the state of local terminal processor, upgrades the active and standby state of using of the link of this logical process node, realizes the migration to new active link of the message that cushions on the link simultaneously.
7, communication means as claimed in claim 1 is characterized in that, in the described transmission of messages flow process, is zero and transmit queue when non-vanishing starts window and surveys if step (b2) detects the re-transmit queue of transmit leg and send window.
8, communication means as claimed in claim 1 is characterized in that, described activestandby state is initiatively switched in the routine, and is in step (c2) with (c3), further comprising the steps of:
(c21) the main processor of using sends link switchover orders to all connected other processors;
(c22) after other processors receive described link switchover order, carry out the pre-switch of link, with message transfers to the buffer queue of masterslave switchover.
9, communication means as claimed in claim 8, it is characterized in that in the described step (c22), other processors also are provided with a masterslave switchover waiting timer, if this timer expiry then is defaulted as overtime message a link-recovery message from current reserve link.
10, communication means as claimed in claim 1, it is characterized in that, described active state-maintenance flow process also comprises link fracture routine: if a link fracture on processor, automatically the master with link switches on another link of this processor with state, and finishes the message that cushions on the link from the migration of old active link to new active link.
11, communication means as claimed in claim 1 is characterized in that, described active state-maintenance flow process also is included in carries out the routine that activestandby state is safeguarded in the message transmitting procedure, and this routine may further comprise the steps:
The destination of carrying message in the service message that sends is main using or the information of backup processor;
When backup processor receives the message that send to main usefulness processor, send active and standby renewal prompting to the informed source processor;
The informed source processor is pointed out according to described renewal, upgrades the active and standby state of using of the link of this logical process node, and finishes the migration that the link buffering goes up message.
12, communication means as claimed in claim 1 is characterized in that, also comprises the correction flow process that connectionless information is sent the purpose processor, may further comprise the steps:
(d1) processor that sends connectionless information is counted respectively the connectionless information that oneself mails to the current reception processing of each logical process node, when count value reaches set point, to another processor transmit status probe message of purpose logical process node;
(d2) another processor in the logical process node is received and is changed if inquire this processor state after the described probe message, then responds the change message of a main and standby relation to the opposite end;
(d3) after the processor that sends connectionless information receives described change message, carry out this processor on the switching of main/slave link, the processor that changed course sends described change message sends connectionless information, repeats above-mentioned flow process later on.
13, communication means as claimed in claim 12, it is characterized in that, described connectionless information is to mail to the main of logical process node to use processor, and described state detection message is the backup processor that mails to the opposite end, and the concurrent physical processor that comprises the recipient is main using or standby information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2004100007740A CN100407727C (en) | 2004-01-18 | 2004-01-18 | An inter-processor communication method based on message |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2004100007740A CN100407727C (en) | 2004-01-18 | 2004-01-18 | An inter-processor communication method based on message |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1558638A true CN1558638A (en) | 2004-12-29 |
| CN100407727C CN100407727C (en) | 2008-07-30 |
Family
ID=34350501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2004100007740A Expired - Fee Related CN100407727C (en) | 2004-01-18 | 2004-01-18 | An inter-processor communication method based on message |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100407727C (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007093087A1 (en) * | 2006-02-14 | 2007-08-23 | Hangzhou H3C Technologies Co., Ltd. | A synchronization method of connection status in data communications and the applied communication node |
| CN100420219C (en) * | 2005-10-27 | 2008-09-17 | 华为技术有限公司 | Method and system for realizing communication between nodes |
| CN100433679C (en) * | 2005-08-26 | 2008-11-12 | 上海贝尔阿尔卡特股份有限公司 | Seamless switching method and apparatus for communication network |
| CN100461803C (en) * | 2005-07-25 | 2009-02-11 | 中兴通讯股份有限公司 | Telephnoe traffic protection method based on E1 |
| CN101047700B (en) * | 2006-05-01 | 2010-05-12 | 华为技术有限公司 | Method and device for raising reliability of LMP control channel |
| CN101908931A (en) * | 2010-08-17 | 2010-12-08 | 中兴通讯股份有限公司 | Method and system of protection switching of optical transmission equipment |
| CN101232357B (en) * | 2008-02-27 | 2011-05-18 | 北京佳讯飞鸿电气股份有限公司 | Apparatus and method for realizing main and spare plate card rearrange |
| CN102142999A (en) * | 2011-03-28 | 2011-08-03 | 华为技术有限公司 | Method, equipment and system for recovering links |
| CN103905331A (en) * | 2012-12-26 | 2014-07-02 | 中国移动通信集团公司 | Real-time media data transmission method, device and system |
| CN106302243A (en) * | 2016-09-23 | 2017-01-04 | 杭州华三通信技术有限公司 | A kind of message transmitting method, CPU and the network equipment |
| CN107666423A (en) * | 2016-07-28 | 2018-02-06 | 操纵技术Ip控股公司 | Uninterrupted data availability during redundancy micro controller system failure |
| CN108075971A (en) * | 2017-12-25 | 2018-05-25 | 新华三技术有限公司 | A kind of main/standby switching method and device |
| CN112398744A (en) * | 2019-08-16 | 2021-02-23 | 阿里巴巴集团控股有限公司 | Network communication method and device and electronic equipment |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ZA988380B (en) * | 1997-09-16 | 1999-03-16 | Siemens Ag | Communications device for transmission of message signals |
| WO2002015526A1 (en) * | 2000-08-11 | 2002-02-21 | Netscape Communications Corporation | Server-side session management |
| CN1198406C (en) * | 2000-09-02 | 2005-04-20 | 中兴通讯股份有限公司 | Stand-by method and device of communication system |
| US6910148B1 (en) * | 2000-12-07 | 2005-06-21 | Nokia, Inc. | Router and routing protocol redundancy |
| CN1184744C (en) * | 2001-08-21 | 2005-01-12 | 华为技术有限公司 | Master-spare converting control circuit and method for realizing it |
| US7490161B2 (en) * | 2001-08-22 | 2009-02-10 | Nokia Inc. | Method and system for implementing OSPF redundancy |
| CN1251419C (en) * | 2002-05-30 | 2006-04-12 | 华为技术有限公司 | Method for realization of fast rearranging main spared device in communication devices |
-
2004
- 2004-01-18 CN CN2004100007740A patent/CN100407727C/en not_active Expired - Fee Related
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100461803C (en) * | 2005-07-25 | 2009-02-11 | 中兴通讯股份有限公司 | Telephnoe traffic protection method based on E1 |
| CN100433679C (en) * | 2005-08-26 | 2008-11-12 | 上海贝尔阿尔卡特股份有限公司 | Seamless switching method and apparatus for communication network |
| CN100420219C (en) * | 2005-10-27 | 2008-09-17 | 华为技术有限公司 | Method and system for realizing communication between nodes |
| WO2007093087A1 (en) * | 2006-02-14 | 2007-08-23 | Hangzhou H3C Technologies Co., Ltd. | A synchronization method of connection status in data communications and the applied communication node |
| AU2006338127B2 (en) * | 2006-02-14 | 2010-01-21 | Hewlett Packard Enterprise Development Lp | A synchronization method of connection status in data communications and the applied communication node |
| CN101022451B (en) * | 2006-02-14 | 2014-07-23 | 杭州华三通信技术有限公司 | Connection state synchronizing method in data communication and applied communication node thereof |
| US7860985B2 (en) | 2006-02-14 | 2010-12-28 | Hangzhou H3C Technologies Co., Ltd. | Method for synchronizing connection state in data communication, and communication node using the same |
| CN101047700B (en) * | 2006-05-01 | 2010-05-12 | 华为技术有限公司 | Method and device for raising reliability of LMP control channel |
| CN101232357B (en) * | 2008-02-27 | 2011-05-18 | 北京佳讯飞鸿电气股份有限公司 | Apparatus and method for realizing main and spare plate card rearrange |
| WO2012022147A1 (en) * | 2010-08-17 | 2012-02-23 | 中兴通讯股份有限公司 | Method and system for protection switching of optical transport device |
| CN101908931B (en) * | 2010-08-17 | 2015-08-12 | 中兴通讯股份有限公司 | The method and system of protection switching of optical transmission equipment |
| CN101908931A (en) * | 2010-08-17 | 2010-12-08 | 中兴通讯股份有限公司 | Method and system of protection switching of optical transmission equipment |
| CN102142999B (en) * | 2011-03-28 | 2013-08-07 | 华为技术有限公司 | Method, equipment and system for recovering links |
| CN102142999A (en) * | 2011-03-28 | 2011-08-03 | 华为技术有限公司 | Method, equipment and system for recovering links |
| CN103905331B (en) * | 2012-12-26 | 2017-05-03 | 中国移动通信集团公司 | Real-time media data transmission method, device and system |
| CN103905331A (en) * | 2012-12-26 | 2014-07-02 | 中国移动通信集团公司 | Real-time media data transmission method, device and system |
| CN107666423A (en) * | 2016-07-28 | 2018-02-06 | 操纵技术Ip控股公司 | Uninterrupted data availability during redundancy micro controller system failure |
| CN107666423B (en) * | 2016-07-28 | 2020-11-27 | 操纵技术Ip控股公司 | Processing unit for providing redundant processors and method for uninterrupted data communication |
| CN106302243A (en) * | 2016-09-23 | 2017-01-04 | 杭州华三通信技术有限公司 | A kind of message transmitting method, CPU and the network equipment |
| CN106302243B (en) * | 2016-09-23 | 2019-11-12 | 新华三技术有限公司 | A kind of message transmitting method, CPU and the network equipment |
| CN108075971A (en) * | 2017-12-25 | 2018-05-25 | 新华三技术有限公司 | A kind of main/standby switching method and device |
| CN108075971B (en) * | 2017-12-25 | 2021-01-26 | 新华三技术有限公司 | Main/standby switching method and device |
| CN112398744A (en) * | 2019-08-16 | 2021-02-23 | 阿里巴巴集团控股有限公司 | Network communication method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100407727C (en) | 2008-07-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1558638A (en) | An inter-processor communication method based on message | |
| TWI465051B (en) | System for efficient recovery of node b buffered data following serving high speed downlink shared channel cell change | |
| CN101361309B (en) | Data transmission method and data re-transmission method | |
| CN100438700C (en) | A method to process failing reconfiguration of wireless bearing service | |
| CN1889464A (en) | Method and apparatus for realizing communication take-over and control | |
| US20060013169A2 (en) | Reliable message distribution in an ad hoc mesh network | |
| CN1633647A (en) | System, method, and product for managing data transfers in a network | |
| TW200405691A (en) | System for efficient recovery of node-b buffered data following MAC layer reset | |
| CN1534947A (en) | Usability and flexibility of information transfer system improoed by transparent application procedure mode | |
| CN105450368A (en) | Method and device of HARQ (Hybrid Automatic Repeat Request) on license-free carrier in LTE-LAA (Long Term Evolution-Licensed Assisted Access) system | |
| CN101047643A (en) | Wireless communication apparatus | |
| CN102355343A (en) | Data transmission method and data re-transmission method | |
| CN1946058A (en) | Soft exchange device allopatric disaster recovery solution system and its method for software exchange network | |
| CN102781052B (en) | Wireless communication method and wireless communication apparatus | |
| CN103391215A (en) | Remote software downloading and updating method, device and system based on chain network | |
| CN1988533A (en) | Method for realizing IuUP/NBUP protocol process control function | |
| TWI352533B (en) | Method and apparatus for rlc protocol error handli | |
| CN1741505A (en) | Method for group transmitting static group backup and method of group transmitting message retransmission | |
| CN1423197A (en) | High usable system based on multi TCP linking map | |
| CN103516492A (en) | Data caching method and device | |
| CN1835444A (en) | Multi-processor static parameter synchronous system of radio internet controller and method thereof | |
| CN102082646B (en) | A kind of method and device of control with regard to the intervalometer of buffer status report | |
| CN101277174A (en) | Method and apparatus for data retransmission based on service quality as well as radio communication system | |
| CN107801200A (en) | The sending method and device of state report | |
| CN1956421A (en) | Reliable transmission method for web service message |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080730 Termination date: 20180118 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |