CN104348637A - Method for maintaining TCP connection in fault switching - Google Patents
Method for maintaining TCP connection in fault switching Download PDFInfo
- Publication number
- CN104348637A CN104348637A CN201310321034.6A CN201310321034A CN104348637A CN 104348637 A CN104348637 A CN 104348637A CN 201310321034 A CN201310321034 A CN 201310321034A CN 104348637 A CN104348637 A CN 104348637A
- Authority
- CN
- China
- Prior art keywords
- port
- tcp
- sequence number
- linkage record
- bag
- 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
Abstract
The invention relates to a method for maintaining TCP connection in fault switching. The following steps are performed simultaneously on a main scheduling server and a standby scheduling server: an IP packet is received and analyzed, the IP packet with no connection record from a client-terminal is preprocessed, and preprocessing comprises the steps that if the IP packet is a TCP data packet, one server is selected from a set of real servers, TCP connection is complementarily established for the address and the response port of the selected server, and the connection record is generated and stored; if the IP packet is not the TCP data packet, one sever is selected from a set of real servers and the connection record is generated and stored; and then the target address and the target port of the IP packet from the client-terminal are rewritten to be the address and the port of the selected real server, a source address and a source port are changed to be a local address and a local port with the complementarily established connection as for the IP packet with the complementarily established TCP connection, a TCP head sequence number and a confirmation sequence number are rewritten simultaneously, and then the rewritten IP packet is sent to the selected real server.
Description
Technical field
The present invention relates to communication technical field, particularly keep the method that TCP connects when failover.
Background technology
Failover (failover) refers to provide a kind of protection mechanism when breaking down, and when a station server lost efficacy, another station server can be taken over job at once, and failover realizes high availability by redundancy.
Linux Virtual Server(is called for short LVS) be based on Linux server cluster solution, as shown in Figure 1, for VS/NAT(Virtual Server via NAT) pattern, when client 100 is by Virtual IP Address (the IP address of Virtual Service) access network services, request message arrives dispatch server (being assumed to be master scheduling server 102), dispatch server selects a station server from one group of real server, such as real server 105, the destination address Virtual IP Address of message is rewritten into the address of selected server, the target port of message is rewritten into the corresponding port of selected server, finally amended message is sent to the server selected, dispatch server device records this connection in connection Hash table simultaneously, when the next message that this connects arrives, from connecting address and the port that can obtain former selected server Hash table, carry out same rewriting operation, and message is passed to former selected server, when the response message from real server is through dispatch server, dispatch server changes the source address of message and source port into Virtual IP Address and corresponding port, again message is issued client.
For the high availability that the active and standby dispatch server of LVS erection realizes, when master scheduling server 102 fault, standby dispatch server 103 to master scheduling server failure, then takes over the work of master scheduling server by heartbeat detection, and user only feels a virtual server.
Then there is following shortcoming in above-mentioned failure switching method of the prior art:
When failover, when from a certain connected belong to message that same TCP connects be balanced to different real server by standby dispatch server time, after this real server receives message, and do not know the connection mentioned in data message section, the principle of TCP process is that recipient is to reset as response, client will receive RST message, cause and connect by abnormal end.Therefore the TCP connected is connected, when failover, easily cause client TCP to connect abnormal end.
Summary of the invention
The object of the invention is to overcome failure switching method of the prior art easily causes client TCP to connect abortive defect, thus provides a kind of method keeping TCP to connect when failover.
To achieve these goals, the invention provides a kind of method keeping when failover TCP to connect, master scheduling server with standby dispatch server perform following steps simultaneously:
Receive and analyzing IP bag, preliminary treatment is done for the IP bag that there is not linkage record from client, described preliminary treatment comprises: if tcp data bag, a station server is selected from one group of real server, reconstruct TCP with the address of selected server and echo port to be connected, generate and preserve linkage record, if not tcp data bag, from one group of real server, select a station server, generate and preserve linkage record; Then address and the port of the real server selected will be rewritten as from the destination address of client ip bag and target port, for belonging to the IP bag reconstructed TCP and connect, also source address and source port to be changed into the local address and local port of reconstructing connection, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, then the real server selected described in revised IP bag is sent to.
In technique scheme, also comprise: the IP from real server is wrapped, if there is linkage record, the source address of wrap described IP and source port change virtual ip address and corresponding port into, for belonging to the IP bag reconstructed TCP and connect, also destination address and target port to be rewritten as client ip and client port, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, more revised IP bag is sent to client; If there is not linkage record, abandon described IP and wrap.
In technique scheme, the method specifically comprises:
Step 1), master scheduling server or standby dispatch server receive and analyzing IP bag, judge the sender that IP wraps;
Step 2) if described IP wraps from client, perform step 3), if described IP wraps from real server, perform step 9);
Step 3), inspection linkage record table, judge whether described IP bag exists linkage record, if exist, perform step 8), otherwise perform step 4); Wherein, each record in described linkage record table at least comprises: client ip, client port, reconstruct connect local IP, reconstruct connect local port, real server IP, real server port, sequence number is poor, acknowledged sequence number is poor;
Step 4), the IP head analyzing described IP bag and transport layer header, judge IP Packet type;
If the described IP bag of step 5) is tcp data bag, perform step 6), otherwise, perform step 7);
Step 6), from one group of real server, select a station server, then using master scheduling server or standby dispatch server as the address of client and server selected in this step with corresponding port is supplementary sets up TCP and be connected, generate a new linkage record, and newly-generated linkage record is saved in linkage record table, then perform step 8);
Step 7), from one group of real server, select a station server, then generate a new linkage record, and newly-generated linkage record is saved in linkage record table, then perform step 8);
Step 8), the destination address of being wrapped by described IP and target port are rewritten as in described linkage record address and the port of the real server selected comprised; If described IP bag belongs to the IP bag reconstructed TCP and connect, then also need source address and source port to change into and reconstruct the local address and local port that TCP is connected, simultaneously the sequence number of rewriting TCP head and acknowledged sequence number; The real server selected described in finally being sent to by above-mentioned revised IP bag, performs step 12);
Step 9), inspection linkage record table, judge whether the IP bag from real server exists linkage record, if exist, perform step 10), otherwise, perform step 11);
Step 10), the source address of being wrapped by described IP and source port change virtual ip address and corresponding port into, and judge whether described IP bag belongs to the IP bag reconstructed TCP and connect further, if belong to, destination address and target port are rewritten as client ip and client port, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, then revised IP bag is issued client, perform step 12) again, otherwise, directly revised IP bag is issued client, then performs step 12);
Step 11), abandon described IP and wrap;
Step 12), flow process terminate.
In technique scheme, in described step 1), judge that the sender that IP wraps comprises: if IP destination address is virtual ip address, then described IP bag is from client, if source address is real server, then described IP bag is from real server.
In technique scheme, in described step 3), judge whether described IP bag exists linkage record and comprise: when the source IP that the described IP from client wrap, source port and client ip in linkage record table, client port are equal, represent that described IP bag correspondence exists record in linkage record table.
In technique scheme, in described step 6) or step 7), the method selecting a station server from one group of real server comprise following any one: Round-Robin Algorithm, weighted round-robin algorithm, destination address hash algorithm, source address hash algorithm, Smallest connection algorithm.
In technique scheme, in described step 6), generate a new linkage record to comprise: the source IP wrapped by described IP and source port are as the client ip in linkage record and client port, the local IP reconstructing connection in this step and local port are connected local IP as reconstructing in described linkage record and reconstruct connection local port, using the address of the real server selected in this step and port as the real server IP in described linkage record and real server port, it is poor as the sequence number in described linkage record that the SEQ field of the TCP head wrapped by described IP deducts the SEQ reconstructing last confirmation bag of three-way handshake in connection establishment process, the ACK field of the TCP head wrapped by described IP deducts to be reconstructed in connection establishment process three-way handshake last confirms that bag is poor as the ACK acknowledged sequence number in described linkage record.
In technique scheme, in described step 7), generate a new linkage record to comprise: the source IP wrapped by described IP and source port are as the client ip in described linkage record and client port, reconstructing in described linkage record is connected local IP be connected local port be set to sky with reconstructing, using the address of the real server selected in this step and port as the real server IP in described linkage record and real server port, sequence number difference in described linkage record is set to zero, the acknowledged sequence number difference in described linkage record is set to zero.
In technique scheme, in described step 8), the sequence number of rewriting TCP head comprises: the sequence number of TCP head is changed into the sequence number that former TCP head sequence number subtracts in linkage record poor; The acknowledged sequence number rewriting TCP head comprises: the acknowledged sequence number of TCP head is changed into the acknowledged sequence number that former TCP head acknowledged sequence number subtracts in linkage record poor.
In technique scheme, in described step 10), the sequence number of rewriting TCP head comprises: the sequence number of TCP head is changed into former TCP head sequence number and add that the acknowledged sequence number in linkage record is poor; The acknowledged sequence number rewriting TCP head comprises: the acknowledged sequence number of TCP head is changed into former TCP head acknowledged sequence number and add that the sequence number in linkage record is poor.
In technique scheme, in described step 9), judge whether the IP bag from real server exists linkage record and comprise: when the source IP that described IP wraps, source port, object IP, destination interface and real server IP in linkage record table, real server port, client ip, client port are equal, represent that described IP wraps in linkage record table and there is record; When the source IP that described IP wraps, source port, object IP, destination interface and real server IP, real server port in linkage record table, reconstruct connect local IP, reconstruct connect local port equal time, represent that described IP wraps in linkage record table and there is record, and this connection connects for reconstructing TCP.
The invention has the advantages that:
The method of the present invention successful TCP that can keep when failover having connected connects, and avoids client TCP to connect abnormal end, and in whole failover process, is transparent to client.
Accompanying drawing explanation
Fig. 1 is the schematic diagram of Linux server cluster;
Fig. 2 is the flow chart of the method keeping TCP to connect when failover of the present invention;
Fig. 3 is the schematic diagram of linkage record table.
Embodiment
Now the invention will be further described by reference to the accompanying drawings.
In one embodiment, still for the Linux server cluster shown in Fig. 1, the method keeping TCP to connect when failover of the present invention is described.
Can know from the description of background technology part: after master scheduling server 102 lost efficacy, standby dispatch server 103 was lost efficacy to master scheduling server 102 by HeartBeat heartbeat inspecting, the task of then adapter master scheduling server 102, again takes over the task of standby dispatch server 103 again after master scheduling server 102 recovers.Method of the present invention needs to implement on master scheduling server 102 and standby dispatch server 103 simultaneously, thus when standby dispatch server 103 adapter master scheduling server 102 task and master scheduling server 102 recover rear tube for dispatch server 103 task, keep the TCP set up to connect.
With reference to figure 2, below method of the present invention is elaborated.
Step 2001), main (or standby) dispatch server utilizes netfilter to receive IP bag, the IP header parameter of analyzing IP bag, the sender that IP wraps is judged: if IP destination address is Virtual IP Address according to analysis result, then described IP bag is from client, if source address is real server, then described IP bag is from real server;
Step 2002), according to IP bag source, perform different operations: if IP bag from client, go to step 2003), if IP bag from real server, go to step 2009);
Step 2003), check linkage record table, judge whether this IP bag exists linkage record, goes to step 2008 if exist), otherwise go to step 2004);
In this step, as described in Figure 3, every bar record comprises and is not limited to described linkage record table: client ip, client port, reconstruct connect local IP, reconstruct connect local port, real server IP, real server port, sequence number is poor, acknowledged sequence number is poor.
In this step, when the source IP that the described IP from client wraps, source port and client ip in linkage record table, client port are equal, represent that described IP wraps and correspondingly in linkage record table, there is record.
Step 2004), analyze described IP wrap IP head and transport layer header, judge IP Packet type: if transport layer be TCP and TCP load not as sky, be then tcp data bag;
Step 2005), according to IP bag type perform different operations: if tcp data bag, go to step 2006), otherwise, go to step 2007);
Step 2006), main (or standby) dispatch server selects a station server from one group of real server, then lead (or standby) dispatch server to supplement as the address of client and selected server and corresponding port and set up TCP and be connected, generate a new linkage record, and newly-generated linkage record is saved in linkage record table, then perform step 2008).
In this step, the selection algorithm that described master (or standby) dispatch server selects a station server from one group of real server can have multiple, includes but not limited to Round-Robin Algorithm, weighted round-robin algorithm, destination address hash algorithm, source address hash algorithm, Smallest connection algorithm.
In this step, linkage record includes but are not limited to: client ip, client port, reconstruct and connect local IP, reconstruct and connect that local port, real server IP, real server port, sequence number are poor, acknowledged sequence number.Specifically, the source IP of client ip wherein and client port described IP bag before being and source port, reconstruct and connect local IP and reconstruct that to be connected local port be local IP and the local port of reconstructing connection in this step, real server IP and real server port are address and the port of the real server selected, sequence number difference is that the SEQ field of the TCP head of described IP bag deducts the SEQ reconstructing last confirmation bag of three-way handshake in connection establishment process, acknowledged sequence number difference is that the ACK field of the TCP head of described IP bag deducts the ACK reconstructing last confirmation bag of three-way handshake in connection establishment process.
Step 2007), main (or standby) dispatch server selects a station server from one group of real server, then generate a new linkage record, and newly-generated linkage record be saved in linkage record table.
In this step, the selection algorithm that described master (or standby) dispatch server selects a station server from one group of real server can have multiple, includes but not limited to Round-Robin Algorithm, weighted round-robin algorithm, destination address hash algorithm, source address hash algorithm, Smallest connection algorithm etc.
Parameter value in described newly-generated linkage record is as follows: client ip and client port are source IP and the source port of described IP bag, reconstruct the local IP of connection and be connected local port for empty with reconstructing, real server IP and real server port are address and the port of the real server selected, sequence number difference is zero, and acknowledged sequence number difference is zero.
Step 2008), the destination address of being wrapped by described IP and target port be rewritten as in described linkage record address and the port of the real server selected comprised; If described IP bag belongs to the IP bag reconstructed TCP and connect, then also need source address and source port to change into and reconstruct the local address and local port that TCP is connected, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, the real server selected described in finally being sent to by above-mentioned revised IP bag, goes to step 2012);
In this step, the sequence number of rewriting TCP head comprises: the sequence number of TCP head is changed into the sequence number that former TCP head sequence number subtracts in linkage record poor; The acknowledged sequence number rewriting TCP head comprises: the acknowledged sequence number of TCP head is changed into the acknowledged sequence number that former TCP head acknowledged sequence number subtracts in linkage record poor.
Step 2009), check linkage record table, judge whether the IP bag from real server exists linkage record, if exist, perform step 2010), otherwise, perform step 2011);
In this step, when the source IP that described IP wraps, source port, object IP, destination interface and real server IP in linkage record table, real server port, client ip, client port are equal, represent that described IP wraps in linkage record table and there is record; When the source IP that described IP wraps, source port, object IP, destination interface and real server IP, real server port in linkage record table, reconstruct connect local IP, reconstruct connect local port equal time, represent that described IP wraps in linkage record table and there is record, and this connection connects for reconstructing TCP.
Step 2010), the source address of being wrapped by described IP and source port change Virtual IP Address(virtual ip address into) and corresponding port, and judge whether described IP bag belongs to the IP bag reconstructed TCP and connect further, if belong to, destination address and target port are rewritten as client ip and client port, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, then revised IP bag is issued client, perform step 2012 again), otherwise, directly revised IP bag is issued client, then perform step 2012);
In this step, the sequence number of rewriting TCP head comprises: the sequence number of TCP head is changed into former TCP head sequence number and add that the acknowledged sequence number in linkage record is poor; The acknowledged sequence number of TCP head is changed into former TCP head acknowledged sequence number and add that the sequence number in linkage record is poor.
Step 2011), abandon described IP and wrap;
Step 2012), flow process terminates.
Foregoing description, for Linux server cluster, is described method of the present invention.In other embodiments, method of the present invention is also applicable to the server cluster of other types.
It should be noted last that, above embodiment is only in order to illustrate technical scheme of the present invention and unrestricted.Although with reference to embodiment to invention has been detailed description, those of ordinary skill in the art is to be understood that, modify to technical scheme of the present invention or equivalent replacement, do not depart from the spirit and scope of technical solution of the present invention, it all should be encompassed in the middle of right of the present invention.
Claims (11)
1. keep the method that TCP connects when failover, master scheduling server with standby dispatch server perform following steps simultaneously:
Receive and analyzing IP bag, preliminary treatment is done for the IP bag that there is not linkage record from client, described preliminary treatment comprises: if tcp data bag, a station server is selected from one group of real server, reconstruct TCP with the address of selected server and echo port to be connected, generate and preserve linkage record, if not tcp data bag, from one group of real server, select a station server, generate and preserve linkage record; Then address and the port of the real server selected will be rewritten as from the destination address of client ip bag and target port, for belonging to the IP bag reconstructed TCP and connect, also source address and source port to be changed into the local address and local port of reconstructing connection, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, then the real server selected described in revised IP bag is sent to.
2. the method keeping TCP to connect when failover according to claim 1, it is characterized in that, also comprise: the IP from real server is wrapped, if there is linkage record, the source address of wrap described IP and source port change virtual ip address and corresponding port into, for belonging to the IP bag reconstructed TCP and connect, also destination address and target port to be rewritten as client ip and client port, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, more revised IP bag is sent to client; If there is not linkage record, abandon described IP and wrap.
3. the method keeping TCP to connect when failover according to claim 2, it is characterized in that, the method specifically comprises:
Step 1), master scheduling server or standby dispatch server receive and analyzing IP bag, judge the sender that IP wraps;
Step 2) if described IP wraps from client, perform step 3), if described IP wraps from real server, perform step 9);
Step 3), inspection linkage record table, judge whether described IP bag exists linkage record, if exist, perform step 8), otherwise perform step 4); Wherein, each record in described linkage record table at least comprises: client ip, client port, reconstruct connect local IP, reconstruct connect local port, real server IP, real server port, sequence number is poor, acknowledged sequence number is poor;
Step 4), the IP head analyzing described IP bag and transport layer header, judge IP Packet type;
If the described IP bag of step 5) is tcp data bag, perform step 6), otherwise, perform step 7);
Step 6), from one group of real server, select a station server, then using master scheduling server or standby dispatch server as the address of client and server selected in this step with corresponding port is supplementary sets up TCP and be connected, generate a new linkage record, and newly-generated linkage record is saved in linkage record table, then perform step 8);
Step 7), from one group of real server, select a station server, then generate a new linkage record, and newly-generated linkage record is saved in linkage record table, then perform step 8);
Step 8), the destination address of being wrapped by described IP and target port are rewritten as in described linkage record address and the port of the real server selected comprised; If described IP bag belongs to the IP bag reconstructed TCP and connect, then also need source address and source port to change into and reconstruct the local address and local port that TCP is connected, simultaneously the sequence number of rewriting TCP head and acknowledged sequence number; The real server selected described in finally being sent to by above-mentioned revised IP bag, performs step 12);
Step 9), inspection linkage record table, judge whether the IP bag from real server exists linkage record, if exist, perform step 10), otherwise, perform step 11);
Step 10), the source address of being wrapped by described IP and source port change virtual ip address and corresponding port into, and judge whether described IP bag belongs to the IP bag reconstructed TCP and connect further, if belong to, destination address and target port are rewritten as client ip and client port, rewrite sequence number and the acknowledged sequence number of TCP head simultaneously, then revised IP bag is issued client, perform step 12) again, otherwise, directly revised IP bag is issued client, then performs step 12);
Step 11), abandon described IP and wrap;
Step 12), flow process terminate.
4. the method keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 1), judge that the sender that IP wraps comprises: if IP destination address is virtual ip address, then described IP bag is from client, if source address is real server, then described IP bag is from real server.
5. the method keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 3), judge whether described IP bag exists linkage record and comprise: when the source IP that the described IP from client wrap, source port and client ip in linkage record table, client port are equal, represent that described IP bag correspondence exists record in linkage record table.
6. the method keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 6) or step 7), the method selecting a station server from one group of real server comprise following any one: Round-Robin Algorithm, weighted round-robin algorithm, destination address hash algorithm, source address hash algorithm, Smallest connection algorithm.
7. the method keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 6), generate a new linkage record to comprise: the source IP wrapped by described IP and source port are as the client ip in linkage record and client port, the local IP reconstructing connection in this step and local port are connected local IP as reconstructing in described linkage record and reconstruct connection local port, using the address of the real server selected in this step and port as the real server IP in described linkage record and real server port, it is poor as the sequence number in described linkage record that the SEQ field of the TCP head wrapped by described IP deducts the SEQ reconstructing last confirmation bag of three-way handshake in connection establishment process, the ACK field of the TCP head wrapped by described IP deducts to be reconstructed in connection establishment process three-way handshake last confirms that bag is poor as the ACK acknowledged sequence number in described linkage record.
8. the method keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 7), generate a new linkage record to comprise: the source IP wrapped by described IP and source port are as the client ip in described linkage record and client port, reconstructing in described linkage record is connected local IP be connected local port be set to sky with reconstructing, using the address of the real server selected in this step and port as the real server IP in described linkage record and real server port, sequence number difference in described linkage record is set to zero, acknowledged sequence number difference in described linkage record is set to zero.
9. the method keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 8), the sequence number of rewriting TCP head comprises: the sequence number of TCP head is changed into the sequence number that former TCP head sequence number subtracts in linkage record poor; The acknowledged sequence number rewriting TCP head comprises: the acknowledged sequence number of TCP head is changed into the acknowledged sequence number that former TCP head acknowledged sequence number subtracts in linkage record poor.
10. the method keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 10), the sequence number of rewriting TCP head comprises: the sequence number of TCP head is changed into former TCP head sequence number and add that the acknowledged sequence number in linkage record is poor; The acknowledged sequence number rewriting TCP head comprises: the acknowledged sequence number of TCP head is changed into former TCP head acknowledged sequence number and add that the sequence number in linkage record is poor.
11. methods keeping TCP to connect when failover according to claim 3, it is characterized in that, in described step 9), judge whether the IP bag from real server exists linkage record and comprise: when the source IP that described IP wraps, source port, object IP, destination interface and real server IP in linkage record table, real server port, client ip, client port are equal, represent that described IP wraps in linkage record table and there is record; When the source IP that described IP wraps, source port, object IP, destination interface and real server IP, real server port in linkage record table, reconstruct connect local IP, reconstruct connect local port equal time, represent that described IP wraps in linkage record table and there is record, and this connection connects for reconstructing TCP.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310321034.6A CN104348637B (en) | 2013-07-26 | 2013-07-26 | A kind of method that TCP connection is kept in failover |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310321034.6A CN104348637B (en) | 2013-07-26 | 2013-07-26 | A kind of method that TCP connection is kept in failover |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104348637A true CN104348637A (en) | 2015-02-11 |
CN104348637B CN104348637B (en) | 2018-07-03 |
Family
ID=52503516
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310321034.6A Active CN104348637B (en) | 2013-07-26 | 2013-07-26 | A kind of method that TCP connection is kept in failover |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104348637B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106878052A (en) * | 2016-12-21 | 2017-06-20 | 新华三技术有限公司 | A kind of customer shift method and device |
CN107306264A (en) * | 2016-04-25 | 2017-10-31 | 腾讯科技(深圳)有限公司 | Network security monitoring method and apparatus |
CN107438012A (en) * | 2016-05-27 | 2017-12-05 | 北京金山云网络技术有限公司 | A kind of load balancing service retransmission method, system, balancer and host |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1391378A (en) * | 2001-06-11 | 2003-01-15 | 华为技术有限公司 | Connection switching system at front end of server cluster and its switching method |
CN1567919A (en) * | 2003-07-02 | 2005-01-19 | 西安大唐电信有限公司 | A method for primary and standby machine to take turns on TCP connection |
CN101252591A (en) * | 2008-04-03 | 2008-08-27 | 中国科学技术大学 | Apparatus and method for realizing uplink and downlink data separation |
US20080235382A1 (en) * | 2007-01-22 | 2008-09-25 | The Regents Of The University Of Colorado | Fault tolerant tcp splice systems and methods |
CN103077242A (en) * | 2013-01-11 | 2013-05-01 | 北京佳讯飞鸿电气股份有限公司 | Method for hot standby of dual database servers |
-
2013
- 2013-07-26 CN CN201310321034.6A patent/CN104348637B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1391378A (en) * | 2001-06-11 | 2003-01-15 | 华为技术有限公司 | Connection switching system at front end of server cluster and its switching method |
CN1567919A (en) * | 2003-07-02 | 2005-01-19 | 西安大唐电信有限公司 | A method for primary and standby machine to take turns on TCP connection |
US20080235382A1 (en) * | 2007-01-22 | 2008-09-25 | The Regents Of The University Of Colorado | Fault tolerant tcp splice systems and methods |
CN101252591A (en) * | 2008-04-03 | 2008-08-27 | 中国科学技术大学 | Apparatus and method for realizing uplink and downlink data separation |
CN103077242A (en) * | 2013-01-11 | 2013-05-01 | 北京佳讯飞鸿电气股份有限公司 | Method for hot standby of dual database servers |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107306264A (en) * | 2016-04-25 | 2017-10-31 | 腾讯科技(深圳)有限公司 | Network security monitoring method and apparatus |
CN107438012A (en) * | 2016-05-27 | 2017-12-05 | 北京金山云网络技术有限公司 | A kind of load balancing service retransmission method, system, balancer and host |
CN107438012B (en) * | 2016-05-27 | 2020-07-28 | 北京金山云网络技术有限公司 | Load balancing service forwarding method, system, balancing device and host machine |
CN106878052A (en) * | 2016-12-21 | 2017-06-20 | 新华三技术有限公司 | A kind of customer shift method and device |
CN106878052B (en) * | 2016-12-21 | 2020-04-03 | 新华三技术有限公司 | User migration method and device |
Also Published As
Publication number | Publication date |
---|---|
CN104348637B (en) | 2018-07-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105308930B (en) | Connection publication in distributed load balancer | |
CN103580880B (en) | Method, equipment and system abnormal a kind of fast notification CGN | |
CN105308931B (en) | Distributed load balancer system and balancing method of loads | |
CN105264865B (en) | Method and distributed load balancer system for distributed load balance | |
CN110198226A (en) | Access request processing method, system and readable medium in a kind of load balancing cluster | |
CN110166568A (en) | Distributed load balancer | |
CN102148767A (en) | Network address translation (NAT)-based data routing method and device | |
CN112217843B (en) | Service unit switching method, system and equipment | |
Shi et al. | NDNLP: A link protocol for NDN | |
NO331320B1 (en) | Balancing network load using host machine status information | |
US11303737B2 (en) | Method and device for data transmission | |
CN107025222A (en) | A kind of distributed information log acquisition method and device | |
CN104038376A (en) | Method and device for managing real servers and LVS clustering system | |
US10693753B2 (en) | Network device snapshots | |
US20150113339A1 (en) | Relay node, control method of relay node and network system | |
CN104503801B (en) | Software upgrade instruction distribution method, the communication server and terminal device | |
CN104348637A (en) | Method for maintaining TCP connection in fault switching | |
EP2939401A1 (en) | Method for guaranteeing service continuity in a telecommunication network and system thereof | |
CN104767655A (en) | Analog result detection method and device | |
US9313291B2 (en) | Systems and methods for transparent communication with bandwidth conservation and HTTP caching | |
US8572245B1 (en) | Using the TCP window size for identifying packets and debugging | |
CN109039959A (en) | A kind of the consistency judgment method and relevant apparatus of SDN network rule | |
CN105897929B (en) | A kind of method and device of video monitoring data backup | |
CN111193813A (en) | Test request processing method and device for determining NAT type and computer equipment | |
CN107943617A (en) | Restorative procedure, device and the server cluster of data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |