CN100362507C - A server load equalizing method - Google Patents
A server load equalizing method Download PDFInfo
- Publication number
- CN100362507C CN100362507C CNB031329802A CN03132980A CN100362507C CN 100362507 C CN100362507 C CN 100362507C CN B031329802 A CNB031329802 A CN B031329802A CN 03132980 A CN03132980 A CN 03132980A CN 100362507 C CN100362507 C CN 100362507C
- Authority
- CN
- China
- Prior art keywords
- server
- interval
- range
- tree
- distributed area
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention discloses a server load equalizing method which is based on the principles of a server weight and probability statistics so as to realize load equalization. The present invention has the following technical scheme that the corresponding relationship between each server of a server group and distribution intervals in accordance with a first predetermined rule is determined in advance; when a new service requests for access, a random number is generated to determine one of the distribution intervals where the random number is in; a server scheduled at the time is determined and provided according to one determined interval of the distribution intervals and the corresponding relationship so as to forward the access request to the server. High load equalization scheduling speed can be ensured by the present invention, and a server in a high weight statistically provides access service by a large number of times so as to realize network forwarding of high performance.
Description
Technical field
The present invention relates to the load-balancing technique of multiserver, be specifically related to a kind of server load balancing method, more specifically, relate to a kind of server load balancing method based on the server weights.
Background technology
Along with developing rapidly of Internet technology, the Internet user sharply increases, and the range of application of internet also enlarges day by day.Simultaneously the ICP satisfies the higher level requirement of Internet user to internet content and form, utilizes various technology and means, and multimedia technology especially makes the easier attraction of the visual effect user of Web content.The increase of the increase of user capture quantity and user's single reference flow is had higher requirement to the ability that server bears concurrent visit.Because the load capacity of separate unit server is always limited, in the face of the user capture quantity and the flowing of access that increase day by day, central processing unit of separate unit server (CPU) and I/O (I/O) can become bottleneck very soon.And this problem is not just soluble by the performance that improves server hardware merely.Common solution is to adopt the group of server that is made of multiple servers and group of server is implemented load-balancing technique, thus the effectively choose reasonable and the traffic sharing problem of settlement server.The common system of selection to server has Hash method, the minimum linking number method of weighting, rotation therapy, weighted round robin method or the like in the present load-balancing technique.These methods or pay attention to the speed that scheduling rates is promptly selected server, as the Hash method, or the balanced intensity of paying attention to scheduling but ignored speed, as weighted round robin method and the minimum connection method of weighting, lacks a kind of dispatching method fast and effectively.
The weighted round robin method is the present representative load-balancing method based on the weights equilibrium.Fig. 1 is a weighted round robin method implementation synoptic diagram.Wherein, S representative server, S1 are represented first station server 110, and Sn represents n station server 1n0, and W represents the weights of each server.At first wheel changes the selection server in the load balancing process, if the weights of selecteed server are 0, then skip this server and select next, if the weights of selecteed server are not 0, then choose this server to provide access service also the weights of this server to be subtracted 1 as the server that is scheduled, commentaries on classics is taken turns in follow-up being chosen on the basis of last time selecting, after the weights of Servers-all all are kept to 0, again weights are reverted to initial value.Being respectively 1,2,3 three station servers with weights is example, selects the first station server S1 for the first time, and its weights are kept to 0, for the second time select the second station server S2, weights are kept to 1, select the 3rd station server S3 for the third time, the 4th next round forwards the first station server S1 to, but its weights are 0, continue wheel and change, and select the first station server S2, its weights are not 0, so the server that is scheduled for the 4th time is the second station server S2, the weights of the second station server S2 are kept to 0 simultaneously ...Whole one order of preference of taking turns is successively: S1, S2, S3, S2, S3, S3.Like this, the number of times that the server that weights are high obtains response just is directly proportional with weights, has guaranteed server is selected to distribute according to weights on the number of times.
But the weighted round robin load-balancing method needs to revise the weighted value of each server in the load balancing process, and network data in high speed processor with the hardware multithreading parallel processing, must lock and release is handled to public variable (as the weights of server etc.), this has just had a strong impact on the processing speed of network data.In addition, the weighted round robin load-balancing method is based on that wheel changes, if the weights of current server are 0, just must skip this server, and this has just reduced the chance of direct hit, thereby has influenced the processing speed of network data.
Summary of the invention
The objective of the invention is to overcome the above-mentioned shortcoming of prior art, a kind of load equilibration scheduling method based on statistics is provided.Utilize the present invention, do not need to revise the data such as weighted value of server, avoided the locking and de-locking of public variable is handled problems, avoided the gap wheel to change phenomenon simultaneously.Both guaranteed load balance scheduling speed faster, and on the statistics angle, also made the high server of weights that more access service number of times is provided, thereby realize high performance forwarded.
The invention provides a kind of server load balancing method, may further comprise the steps:
Pre-determine each station server and the corresponding relation that meets the distributed area of first pre-defined rule in the group of server; Described first pre-defined rule comprises: the interval range of the distributed area of server and the proportional rule of the weights of server;
When new services request inserts,
Determine described one of them distributed area according to second pre-defined rule; Wherein, described second pre-defined rule is meant and generates a random number that is positioned at the total interval range that is made of all distributed areas, determines the rule of one of them distributed area that described random number is positioned at;
Determine to provide this server that is scheduled and will insert and ask to be transmitted to this server according to determined described one of them distributed area and described corresponding relation.
Preferably, the described step that pre-determines described corresponding relation comprises the step of the weighted value that pre-determines described each server, and according to determined weighted value, the step of the distributed area that is directly proportional with its weights for interval range of described each server-assignment; Wherein, described first pre-defined rule is meant according to determined weighted value, the rule that is directly proportional with its weights for the interval range of the distributed area of described each server-assignment.
Alternatively, describedly determine according to second pre-defined rule that the step of distributed area comprises and generate a random number that is positioned at the total interval range that constitutes by all distributed areas, determine the step of one of them distributed area that described random number is positioned at.
Alternatively, the described step that pre-determines described corresponding relation comprises the step of the corresponding relation of the tree type distributed area of determining described each station server and meeting first pre-defined rule, the described step of determining the server that is scheduled comprises according to second pre-defined rule searches described tree type distributed area, determines the step of the server that is scheduled.
Alternatively, the described step that pre-determines described corresponding relation comprises the step of the corresponding relation of the binary tree type distributed area of determining described each station server and meeting first pre-defined rule, the described step of determining the server that is scheduled comprises according to second pre-defined rule searches described binary tree type distributed area, determines the step of the server that is scheduled.
Preferably, the described step that pre-determines described corresponding relation comprises the step of the corresponding relation of the binary sort tree type distributed area of determining described each station server and meeting first pre-defined rule, the described step of determining the server that is scheduled comprises according to second pre-defined rule searches described binary sort tree type distributed area, determines the step of the server that is scheduled.
Preferably, described method comprises the step that pre-determines the weighted value of each server in the described group of server, and the step of the corresponding relation of described definite binary sort tree type distributed area comprises:
Weighted value according to described each server, an interval order of continuous numerical value is the sub-range that described each station server distributes this interval left margin value representation of usefulness, the right margin in this sub-range is represented with the left margin in the next son interval in this sub-range, in other words, the right margin in this sub-range equals the left margin in the next son interval in this sub-range;
Described server is divided into some subgroups in order, the pairing sub-range of server that wherein constitutes each subgroup constitutes a continuous interval, this interval left side dividing value equals to constitute the minimum left side dividing value in all sub-ranges in this interval, and the right dividing value in this interval equals the corresponding interval left side dividing value in next subgroup;
This node is represented an interval with the binary sort tree of the corresponding interval left side dividing value in described subgroup as node to construct one, and this left side, interval dividing value is a nodal value, and the right dividing value is the nodal value of next node.
Preferably, the step of described structure binary sort tree comprises that structure balanced binary tree or its left side/right subtree are the step of the binary sort tree of balanced binary tree.
Alternatively, the step of described definite one of them distributed area comprises:
Generate a random number that is positioned at the total interval range that constitutes by all described sub-ranges;
With described binary sort tree as the current binary sort tree of being searched;
Judge whether described random number is positioned at the interval of the root node of the described current binary sort tree of being searched,
If,
In the pairing subgroup of the root node of the described current binary sort tree of being searched, determine the sub-range of one of them described server correspondence that described random number is positioned at;
According to the corresponding relation of described described sub-range and described server, determine the server that is scheduled;
If not,
Judge described random number whether less than the left side dividing value in the pairing interval of root node of the described current binary sort tree of being searched,
If,
With the left subtree of the described current binary sort tree of being searched as the current binary sort tree of being searched;
Return and judge whether described random number is positioned at the interval step of the root node of the described current binary sort tree of being searched;
If not,
With the right subtree of the described current binary sort tree of being searched as the current binary sort tree of being searched;
Return and judge whether described random number is positioned at the interval step of the root node of the described current binary sort tree of being searched.
Alternatively, the described step that pre-determines described corresponding relation comprises the step of the distribution table of the static state of determining described each station server and meeting first pre-defined rule.
As can be seen from the above technical solutions, the present invention adopts the method that processing separates, search in the interval, the distributed area table of queries static when load balance scheduling, and not to the public variable operation that makes an amendment, avoided the lock of multithreading to handle problems, once selected when selecting server simultaneously, avoided selection gap phenomenon.Adopt the static distribution interval table of binary sort tree form, the inquiry velocity in the time that load dispatch can being accelerated.Be the distributed area scope that server-assignment is directly proportional with its weights, guaranteed the weighting of server is covered.Adopt the connection of random number mechanism assigns access server, reduced bump server, to a certain extent better protection server.Thereby on the angle of statistics, see, guaranteed the uniqueness and the high efficiency of server scheduling.
Description of drawings
Fig. 1 represents existing weighted round robin method synoptic diagram based on the weights equilibrium;
Fig. 2 represents a function distribution plan of realizing system of the present invention;
Fig. 3 represents the load balance scheduling process flow diagram of the embodiment of the invention;
Fig. 4 represents server-weights in the embodiments of the invention-border graph of a relation;
Fig. 5 is illustrated in the FB(flow block) of determining the sub-range at random number place in the binary sort tree;
Fig. 6 represents a kind of binary sort tree structural drawing of distributed area;
Fig. 7 represents the binary sort tree structural drawing of another kind of distributed area;
Embodiment
Below in conjunction with accompanying drawing and embodiment the present invention is described in further detail.
Fig. 2 represents the distribution schematic diagram of function in the present embodiment.The user can carry out function setting to master control borad by administration interface, comprise weighted value that load balancing strategy and each server are set etc., distributed area used when generating server load balancing by master control borad according to predetermined rule is inquired about list item, and the inquiry list item that will represent the corresponding relation of the distributed area that generates and server directly is configured to the network data processing interface board, network processing unit in scheduling process on the interface board is absorbed in the interval and is searched and dispatch processing, thereby realizes the quick forwarding of the network equipment.
Fig. 3 is the process flow diagram of load balance scheduling in the expression embodiment of the invention.In step 310, be that every station server is selected a weighted value according to the load capacity of server in advance, the load capacity of server is big more, for the weighted value of its selection also big more.So that 256 servers to be arranged in the load-balanced server group is example, and these servers are respectively with S1, S2 ..., S256 represents that pairing weighted value is respectively W1, W2 ..., W256 for the sake of simplicity, supposes that the numerical value of weighted value is followed successively by 1,2 ..., 256.
In step 320, for every station server distributes a sub-range that does not overlap between 0～65535, total interval range that these sub-ranges constitute is 0～65535.For the step in server-assignment sub-range be at first with server by the descending series arrangement of weighted value, ask for the summation of above-mentioned all weighted values simultaneously, the principle that is directly proportional with weighted value according to the interval range in sub-range goes out the sub-range of every station server according to the weights order computation of every station server then.With 3 station servers is example, supposes that the weights of three station servers are respectively 3,2,1, and the summation of all weighted values is: 3+2+1=6, the sub-range of such 3 servers is followed successively by [0,65535 * 3/6), [65535 * 3/6,65535 * (3+2)/6), [65535 * (3+2)/6,65535].Be according to the method described above 256 station servers orderings and successively (after the server orders order change into S256, S255 ..., S2, S1) distribute the sub-range, be followed successively by: [0,65535 * 256/32896), [65535 * 256/32896,65535 * (256+255)/32896) ..., [65535 * (256+...+2)/32896,65535), promptly [0,510), [510,1018) ..., [65533,65535].In processor, each sub-range is represented with a value, promptly represents with the left side dividing value in this sub-range, and the right margin in this sub-range is represented with the left margin in the next son interval in this sub-range, in other words, the right dividing value in this sub-range equals the left side dividing value in the next son interval in this sub-range.So above-mentioned 256 sub-ranges can be expressed as 0,510,1018 ..., 65533.
In step 330, order is done a subgroup with per 16 branches of above-mentioned ordering 256 station servers, be divided into be 16 subgroup: SG1, SG2 ... SG16, the pairing sub-range of server that constitutes each subgroup constitutes a continuous interval, this interval left margin is the minimum left margin in all sub-ranges that constitute this interval, this interval right margin is the maximum right margin in all sub-ranges that constitute this interval, the corresponding interval left side dividing value in promptly next subgroup.Pairing interval, 16 subgroups is followed successively by: [0,7920), [7920,15331) ..., [65264,65535), the expression in processor is followed successively by: 0,7920 ..., 65264.In the internal memory of processor, with the block information of a subgroup correspondence of 32 bytes store, this block information comprises the sub-range information of 16 station servers of forming this subgroup again, is node with such memory block, and totally 16 nodes make up binary sort tree.Fig. 4 is that described server-weights-border concerns synoptic diagram, the sub-range of n station server correspondence after wherein En represents to sort, and Node0 represents first node of binary sort tree.
A kind of scheme of possible structure binary sort tree is as described below, but is not limited to this scheme.Select subgroup node SGj placed in the middle in 16 tactic subgroup nodes root node as binary sort tree, wherein node SGj in subgroup satisfies: j=INT ((n+m+1)/2), here INT represents rounding operation, m represents the start node numbering, n represents the terminal note numbering, in this example for the first time m be 1, the n value is 16, so with the root node of subgroup node SG9 as binary sort tree, the subgroup node before the SG9, also be the left subtree that subgroup node SG1～SG8 continues to set up according to the method described above binary sort tree; The subgroup node after the SG9, also be the right subtree that subgroup node SG10～SG16 continues to set up according to the method described above binary sort tree.For the subgroup node SG1～SG8 before the SG9, calculating is as the subgroup node SGk of left subtree root node, k=INT ((8+1+1)/2)=5 wherein, being SG5 is divided into two parts as the root node of left subtree again with the subgroup node before the SG9, and laying respectively at SG5 is in the left and right sides subtree of binary sort tree of root node.According to the method described above, the final binary sort tree of setting up as shown in Figure 6.Because subgroup node SG1 interval border value is minimum, binary sort tree deformable shown in Figure 6 is a form shown in Figure 7.It will be understood by those skilled in the art that this binary sort tree also can be enumerated no longer one by one by the distortion of various ways here.Preferably set up complete binary sort tree shown in Figure 6, occupation space can be less when with the continuous cell stores of array form or other one group.
When new services request inserted, promptly being judged as when sure of step 3 50 changed step 360 over to and carries out the task of generating random number.If being judged as of step 350 is negative, whether system continues monitoring new services request.In step 360, random number between processor generates one 0～65535, in the binary sort tree of being built, search the subgroup node that this random number is positioned at according to step 370 then, and then in the node of 16 servers forming this subgroup node, search the sub-range of this random digit in any station server correspondence, can adopt sequential search, also can adopt other modes to search, in step 380 according to random digit in the sub-range and the corresponding relation of aforementioned predetermined sub-range and server determine the server that is scheduled, and services request is transmitted to this server.
Fig. 5 is illustrated in the process flow diagram of searching the sub-range that random number is positioned in the binary sort tree of being built.For example, when new services request inserted, processor had at first generated a random number, supposes that this random number is 28734, searches the sub-range of this random number correspondence in the binary sort tree of being built.In step 510, at first with the binary sort tree built as the current binary sort tree of being searched, judge in step 520 whether random number 28734 is positioned at the root node scope of this binary sort tree, the corresponding interval of the root node of the current binary sort tree of being searched is: [49087,52928), judged result is for negating, change step 530 over to and judge that random number 28734 is whether less than the left side dividing value 49280 of the root node of this binary sort tree, judged result is for affirming, step 540 with the left subtree of the current binary sort tree of being searched be root node be the binary sort tree of SG5 as the current binary sort tree of being searched next time, change step 520 over to and continue said process.If the judged result of step 530 is for negating, then in step 550 with the right subtree of the current binary sort tree of being searched as the current binary sort tree of being searched next time, change step 520 continuation said process over to.The pairing interval of SG5 is [28623,34504), sure when entering for the second time the judgement of step 520, promptly the random number 28734 of Chan Shenging is positioned at the interval of the 5th subgroup correspondence, change step 560 over to, the sub-range that with first sub-range of subgroup node SG5 is server S 192 correspondences is as the sub-range of being searched, judge in step 570 whether random number 28734 is positioned at this sub-range, judged result is for affirming, be the sub-range of random digit in server S 192 correspondences, change step 580 over to and determine that according to the corresponding relation in server and sub-range the server that is scheduled is S192, query script finishes.If the judged result of step 570 is then moved a sub-range for negating after step 590, be about to next sub-range as the sub-range of being searched, change step 570 over to and continue said process.
Though described the present invention by embodiment, those of ordinary skills know, the present invention has many distortion and variation and do not break away from spirit of the present invention, wish that appended claim comprises these distortion and variation and is not limited to the content that embodiment describes.
Claims (11)
1. a server load balancing method comprises the steps:
Pre-determine each station server and the corresponding relation that meets the distributed area of first pre-defined rule in the group of server; Described first pre-defined rule comprises: the interval range of the distributed area of server and the proportional rule of the weights of server;
When new services request inserts,
Determine one of them of described distributed area according to second pre-defined rule; Wherein, described second pre-defined rule is meant and generates a random number that is positioned at the total interval range that is made of all distributed areas, determines the rule of one of them distributed area that described random number is positioned at;
Determine to provide this server that is scheduled and will insert and ask to be transmitted to this server according to determined described one of them distributed area and described corresponding relation.
2. server load balancing method as claimed in claim 1, it is characterized in that the described step that pre-determines described corresponding relation comprises the step of the weighted value that pre-determines described each server, and according to determined weighted value, the step of the distributed area that is directly proportional with its weights for interval range of described each server-assignment; Wherein, described first pre-defined rule is meant according to determined weighted value, the rule that is directly proportional with its weights for the interval range of the distributed area of described each server-assignment.
3. server load balancing method as claimed in claim 1, it is characterized in that describedly determining according to second pre-defined rule that the step of distributed area comprises and generating a random number that is positioned at the total interval range that constitutes by all distributed areas, determine the step of one of them distributed area that described random number is positioned at.
4. server load balancing method as claimed in claim 1, it is characterized in that the described step that pre-determines described corresponding relation comprises the step of the corresponding relation of the tree type distributed area of determining described each station server and meeting first pre-defined rule, the described step of determining the server that is scheduled comprises according to second pre-defined rule searches described tree type distributed area, determines the step of the server that is scheduled.
5. server load balancing method as claimed in claim 1, it is characterized in that the described step that pre-determines described corresponding relation comprises the step of the corresponding relation of the binary tree type distributed area of determining described each station server and meeting first pre-defined rule, the described step of determining the server that is scheduled comprises according to second pre-defined rule searches described binary tree type distributed area, determines the step of the server that is scheduled.
6. server load balancing method as claimed in claim 1, it is characterized in that the described step that pre-determines described corresponding relation comprises the step of the corresponding relation of the binary sort tree type distributed area of determining described each station server and meeting first pre-defined rule, the described step of determining the server that is scheduled comprises according to second pre-defined rule searches described binary sort tree type distributed area, determines the step of the server that is scheduled.
7. server load balancing method as claimed in claim 6 is characterized in that described method comprises the step that pre-determines the weighted value of each server in the described group of server, and the step of the corresponding relation of described definite binary sort tree type distributed area comprises:
Weighted value according to described each server, an interval order of continuous numerical value is the sub-range that described each station server distributes this interval left margin value representation of usefulness, the right margin in this sub-range is represented with the left margin in the next son interval in this sub-range, in other words, the right margin in this sub-range equals the left margin in the next son interval in this sub-range;
Described server is divided into some subgroups in order, the pairing sub-range of server that wherein constitutes each subgroup constitutes a continuous interval, this interval left side dividing value equals to constitute the minimum left side dividing value in all sub-ranges in this interval, and the right dividing value in this interval equals the corresponding interval left side dividing value in next subgroup;
This node is represented an interval with the binary sort tree of the corresponding interval left side dividing value in described subgroup as node to construct one, and this left side, interval dividing value is a nodal value, and the right dividing value is the nodal value of next node.
8. server load balancing method as claimed in claim 7 is characterized in that the step of described structure binary sort tree comprises that structure balanced binary tree or its left side/right subtree are the step of the binary sort tree of balanced binary tree.
9. server load balancing method as claimed in claim 6 is characterized in that the step of described structure binary sort tree comprises the step of structure right subtree [djs312874] for the binary sort tree of balanced binary tree.
10. server load balancing method as claimed in claim 7 is characterized in that the step of described definite one of them distributed area comprises:
Generate a random number that is positioned at the total interval range that constitutes by all described sub-ranges;
With described binary sort tree as the current binary sort tree of being searched;
Judge whether described random number is positioned at the interval of the root node of the described current binary sort tree of being searched,
If,
In the pairing subgroup of the root node of the described current binary sort tree of being searched
In determine the sub-range of one of them described server correspondence that described random number is positioned at;
According to the corresponding relation of described described sub-range and described server, determine the server that is scheduled;
If not,
Judge described random number whether less than the left side dividing value in the pairing interval of root node of the described current binary sort tree of being searched,
If,
With the left subtree of the described current binary sort tree of being searched as the current binary sort tree of being searched;
Return and judge whether described random number is positioned at the interval step of the root node of the described current binary sort tree of being searched;
If not,
With the right subtree of the described current binary sort tree of being searched as the current binary sort tree of being searched;
Return and judge whether described random number is positioned at the interval step of the root node of the described current binary sort tree of being searched.
11. server load balancing method as claimed in claim 1 is characterized in that the described step that pre-determines described corresponding relation comprises the step of the distribution table of the static state of determining described each station server and meeting first pre-defined rule.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031329802A CN100362507C (en) | 2003-07-23 | 2003-07-23 | A server load equalizing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031329802A CN100362507C (en) | 2003-07-23 | 2003-07-23 | A server load equalizing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1570903A CN1570903A (en) | 2005-01-26 |
| CN100362507C true CN100362507C (en) | 2008-01-16 |
Family
ID=34469991
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB031329802A Expired - Fee Related CN100362507C (en) | 2003-07-23 | 2003-07-23 | A server load equalizing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100362507C (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7941165B2 (en) | 2005-03-02 | 2011-05-10 | Cisco Technology, Inc. | System and method for providing a proxy in a short message service (SMS) environment |
| US7640023B2 (en) * | 2006-05-03 | 2009-12-29 | Cisco Technology, Inc. | System and method for server farm resource allocation |
| CN101945041B (en) * | 2010-09-02 | 2012-11-28 | 中国科学院计算机网络信息中心 | Authoritative server load balancing method, equipment and system |
| CN102932389B (en) * | 2011-08-11 | 2016-06-22 | 阿里巴巴集团控股有限公司 | A kind of request processing method, device and server system |
| CN102664753B (en) * | 2012-04-17 | 2015-12-02 | 杭州华三通信技术有限公司 | A kind of forwarding tree generation method and device thereof |
| CN103580988B (en) * | 2012-07-31 | 2017-10-17 | 阿里巴巴集团控股有限公司 | Message sink, push, method, device, server group and the system of transmission |
| CN103905530A (en) * | 2014-03-11 | 2014-07-02 | 浪潮集团山东通用软件有限公司 | High-performance global load balance distributed database data routing method |
| US9923794B2 (en) | 2014-04-28 | 2018-03-20 | Huawei Technologies Co., Ltd. | Method, apparatus, and system for identifying abnormal IP data stream |
| CN105099732B (en) * | 2014-04-28 | 2018-11-20 | 华为技术有限公司 | A kind of methods, devices and systems identifying abnormal IP traffic |
| CN104283963B (en) * | 2014-10-21 | 2018-02-09 | 无锡云捷科技有限公司 | A kind of CDN load-balancing methods of Distributed Cooperative formula |
| CN104468823B (en) * | 2014-12-24 | 2018-02-06 | 浪潮(北京)电子信息产业有限公司 | A kind of cluster load balance dispatching method and system |
| CN107656980B (en) * | 2017-09-07 | 2020-09-22 | 北京神州绿盟信息安全科技股份有限公司 | Method applied to distributed database system and distributed database system |
| CN108769101A (en) * | 2018-04-03 | 2018-11-06 | 北京奇艺世纪科技有限公司 | A kind of information processing method, client and system |
| CN113900731B (en) * | 2021-09-30 | 2023-08-15 | 百度在线网络技术（北京）有限公司 | Request processing method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5872930A (en) * | 1996-07-11 | 1999-02-16 | Microsoft Corporation | Load balancing between E-mail servers within a local area network |
| CN1267439A (en) * | 1997-06-18 | 2000-09-20 | 诺基亚流动电话有限公司 | Method for identifying base station of time division cellular network in mobile station and mobile station |
| CN1403934A (en) * | 2001-09-06 | 2003-03-19 | 华为技术有限公司 | Load balancing method and equipment for convective medium server |
-
2003
- 2003-07-23 CN CNB031329802A patent/CN100362507C/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5872930A (en) * | 1996-07-11 | 1999-02-16 | Microsoft Corporation | Load balancing between E-mail servers within a local area network |
| CN1267439A (en) * | 1997-06-18 | 2000-09-20 | 诺基亚流动电话有限公司 | Method for identifying base station of time division cellular network in mobile station and mobile station |
| CN1403934A (en) * | 2001-09-06 | 2003-03-19 | 华为技术有限公司 | Load balancing method and equipment for convective medium server |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1570903A (en) | 2005-01-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100362507C (en) | A server load equalizing method | |
| CN1098488C (en) | Method and apparatus for dynamic interval-based load balancing | |
| Karger et al. | Simple efficient load balancing algorithms for peer-to-peer systems | |
| Noordhuis et al. | Mining twitter in the cloud: A case study | |
| Karger et al. | Simple efficient load-balancing algorithms for peer-to-peer systems | |
| CN107800768A (en) | Open platform control method and system | |
| CN102137128A (en) | Method and device for balancing load of cluster service | |
| CN105577806B (en) | A kind of distributed caching method and system | |
| Ibrahim et al. | Intelligent data placement mechanism for replicas distribution in cloud storage systems | |
| CN104283963B (en) | A kind of CDN load-balancing methods of Distributed Cooperative formula | |
| KR100718907B1 (en) | Load balancing system based on fuzzy grouping and the load balancing method | |
| Khodar et al. | New scheduling approach for virtual machine resources in cloud computing based on genetic algorithm | |
| Shen et al. | A lightweight and cooperative multifactor considered file replication method in structured P2P systems | |
| Giakkoupis et al. | A scheme for load balancing in heterogenous distributed hash tables | |
| CN105120008B (en) | A kind of distributed cloud computing center balancing method of loads based on layering | |
| Soltani et al. | A dynamic popularity-aware load balancing algorithm for structured p2p systems | |
| CN101442466B (en) | Superpose network and implementing method | |
| Datta et al. | On de Bruijn routing in distributed hash tables: there and back again | |
| CN107528884B (en) | User request processing method and device of aggregation server | |
| Nakazato et al. | Data allocation method considering server performance and data access frequency with consistent hashing | |
| Kumar et al. | BMAQR: balanced multi attribute QoS aware replication in HDFS | |
| CN105959427A (en) | DNS server automatic expansion method | |
| Xia et al. | Heterogeneity and load balance in structured P2P system | |
| CN116455756B (en) | Bandwidth equalization method, device and medium based on market push and inquiry | |
| Berenbrink et al. | Randomized and adversarial load balancing |
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: 20080116 Termination date: 20150723 |
|
| EXPY | Termination of patent right or utility model |