A kind of load balancing apparatus and the method realizing load balancing thereofTechnical field
The present invention relates to the network communications technology, particularly relate to a kind of load balancing dress improving server service qualityThe method put and realize load balancing.
Background technology
Along with the development of the Internet (Internet) technology, the access number faced by the webserver is greatly increasedAnd it is more unpredictable.In a network, if only using a server to provide service to client, then, serverDisposal ability and I/O capability will necessarily become the bottleneck providing service.
It addition, growing along with Internet technology, the service content that the Internet is provided also gets more and more, for oneFor a little service entrys or service content, it usually needs multiple server is set service is provided simultaneously, to carry out load balancing(Load Balance), it is to avoid single server cannot provide the situation of service in time simultaneously for multiple users, improves serviceQuality, load balancing here, refer to load (task) is balanced dividing on multiple operating unit execution, thusJointly complete task.
Equipment connection schematic diagram when Fig. 1 is to use load-balancing technique in prior art.Refer to Fig. 1, loadBalancing technique is exactly that to arrange at network side can be client multiple servers of providing service, and arranges one in server sideLoad-balancing device, when client needs to conduct interviews, only need to access this load-balancing device, and load-balancing device receives clientAfter the access request of end, from multiple servers, select one according to certain rule, by selected server response clientEnd, provides service for client.
Visible, load-balancing technique can utilize multiple servers to provide for a large number of users simultaneously and concurrently access service, because ofThis, it is possible to access speed and the throughput capacity of network are provided significantly.But due in prior art SiteServer LBS mostly with wheelAsking or Weighted Round Robin is main, these algorithms have certain effect really for general case, but, in complexity all the moreIn network of network situation, it is inadequate for depending merely on this algorithm, such as, when the system failure of certain station server, common poll is calculatedMethod can not recognize fault, and it still can be pass by assignment of traffic, but server can't respond, and this has resulted in bandWide waste, and when server does not responds to when, load-balancing device is also possible to will be considered that this station server is not taken turnsInquiry is arrived, and therefore may proceed to sending flow rate to it the most again less than response, will result in network and interrupts.
Certainly, along with the improvement of load-balancing technique, existing technical scheme also occurs in that the negative of comprehensive various algorithmCarry equalization function, such as: inside some load equalizer, be integrated with healthy requestor, may be used for authentication server whether faultOr whether can use, but this inheritance method is also that comparison is unilateral, if server just begins to use, then be this integratedMethod is feasible, but when performing multiple application on server, when providing the user multiple service, this integrated approach showsSo do not meet reality.
Summary of the invention
For the deficiency overcoming above-mentioned prior art to exist, the purpose of the present invention be to provide a kind of load balancing apparatus andThe method realizing load balancing, it can make server enjoy more preferable load-balancing function.
For reaching above and other purpose, the present invention proposes a kind of load balancing apparatus, including:
Rear end equipment, for the use information of each server of timing acquiring, and sets the transmission of collection data timing to front endStandby；
Headend equipment is after receiving collection data, for gathering data distribution weights, according to the collection data after distribution weightsCalculate the server of comprehensive maximum weight, and the user received request is directed to corresponding server.
Further, described rear end equipment installs a harvester in each server, represents that each server makes with timing acquiringBy the use information of situation.
Further, described headend equipment includes:
Receive unit, for receiving the collection data of each harvester of described rear end equipment；
Weights allocation unit, data distribute weights each of in by the data that gathered；
Computing unit, for calculating the comprehensive weights of each server according to the collection data of each server, and obtainsThe server of comprehensive maximum weight；
Request pilot unit, for being directed to the server of comprehensive maximum weight by the user received request.
Further, described rear end equipment includes: multiple harvesters and multiple handover request unit, each harvester andHandover request unit is arranged at each server, and described harvester is for the use information of each server of timing acquiring, and timing connectsConnect described headend equipment, the data gathered are transferred to described headend equipment；Described handover request unit, for according to describedThe collection data of harvester, it may be judged whether send path handover request to described headend equipment, and when judged result is for being, toHeadend equipment sends path handover request.
Further, described handover request unit is according to the collection data of described harvester, it is judged that making of current serverWhether reach SC service ceiling by rate, if current server has reached SC service ceiling, then send path switching to described headend equipmentRequest.
Further, described headend equipment includes:
Receive unit, for receiving the collection data of each harvester of described rear end equipment；
Weights allocation unit, data distribute weights each of in by the data that gathered；
Computing unit, for calculating the comprehensive weights of each server according to the collection data of each server, and obtainsThe server of comprehensive maximum weight；
Request pilot unit, for being directed to the server of comprehensive maximum weight by the user received request；
Path switching unit, in time receiving path handover request, carries out flow and issues path switching.
Further, described path switching unit flow is issued path switch to comprehensive weights be only second to current path clothesThe path corresponding to server of business device.
For reaching above-mentioned purpose, the present invention also provides for a kind of load-balancing method, comprises the steps:
Step one, utilize the use information of each server of harvester timing acquiring, and data timing transmission will be gathered to the most frontEnd equipment；
Step 2, headend equipment, after receiving collection data, distributes weights for gathering in data, after distribution weightsGather data and calculate the server of comprehensive maximum weight, and the user received request is directed to corresponding server.
Further, step 2 farther includes:
Receive the collection data of each harvester；
Weights are distributed by data each of in the data that gathered；
Collection data according to each server calculate the comprehensive weights of each server, and obtain comprehensive maximum weightServer；
The user received request is directed to the server of comprehensive maximum weight.
Further, described method also includes:
Collection data according to harvester, it may be judged whether forward end equipment sends path handover request, and in judged resultDuring for being, forward end equipment sends path handover request；
Headend equipment, in time receiving path handover request, carries out flow and issues path switching, flow is issued path and cutsIt is changed to comprehensive weights and is only second to the path that the server of current path server is corresponding.
Compared with prior art, one load balancing apparatus of the present invention and load-balancing method thereof are by utilizing rear end equipmentHarvester gather each server service condition, and by gather data send to headend equipment, utilize headend equipment for adoptingCollection data distribution weights, and the comprehensive weights of each server are calculated according to collection data, ask user to guide comprehensive weightsMaximum server, by the invention it is possible to make server enjoy more preferable load-balancing function.
Accompanying drawing explanation
Equipment connection schematic diagram when Fig. 1 is to use load-balancing technique in prior art；
Fig. 2 is the system architecture schematic diagram of a kind of load balancing apparatus of the present invention；
Fig. 3 is the detail structure chart of headend equipment 21 in present pre-ferred embodiments；
Fig. 4 is the thin portion structural representation of rear end equipment 20 in another preferred embodiment of the present invention；
Fig. 5 is the detail structure chart of headend equipment 21 in another preferred embodiment of the present invention；
Fig. 6 is the load balancing process schematic of the specific embodiment of a kind of load balancing apparatus of the present invention；
Fig. 7 is the flow chart of steps of a kind of load-balancing method of the present invention；
Fig. 8 is the thin portion flow chart of step 702 in present pre-ferred embodiments.
Detailed description of the invention
Below by way of specific instantiation accompanying drawings embodiments of the present invention, those skilled in the art canFurther advantage and effect of the present invention is understood easily by content disclosed in the present specification.The present invention also can be different by otherInstantiation implemented or applied, the every details in this specification also can based on different viewpoints and application, without departing substantially fromVarious modification and change is carried out under the spirit of the present invention.
Fig. 2 is the system architecture schematic diagram of one of a kind of load balancing apparatus of present invention preferred embodiment.As in figure 2 it is shown,One load-balancing device of the present invention, including: rear end equipment 20 and headend equipment 21.
Wherein, rear end equipment 20 is for the use information of each server of timing acquiring, and will gather data timing transmission extremelyHeadend equipment 21, in present pre-ferred embodiments, rear end equipment 20 installs a harvester in each server, with timing acquiring tableShow the use information of each server service condition, such as: CPU usage, memory usage, the bandwidth usage of port andAccess the range of information such as interactive information of application, each harvester timing can actively connect headend equipment 21, being gatheredData pass to headend equipment 21；Headend equipment 21, after receiving collection data, distributes weights for gathering, calculates comprehensive in dataClose the server of maximum weight, and the user received request is directed to corresponding server, specifically, headend equipment 21After receiving collection data, weights can be distributed for each of gathering in data data, such as, gather packet and use containing CPURate, memory usage, the bandwidth usage etc. of port, headend equipment 21 then distributes a weights A for CPU usage, and internal memory makesDistributing a weights B by rate, the bandwidth usage of port distributes a weights C ..., the then collection data meter to each serverCalculate comprehensive weights, it is thus achieved that the server of comprehensive maximum weight, and the user received request is directed to comprehensive maximum weightServer.
Fig. 3 is the detail structure chart of headend equipment 21 in present pre-ferred embodiments.In present pre-ferred embodiments, frontEnd equipment 21 farther includes: receive unit 210, weights allocation unit 211, computing unit 212 and request pilot unit213, wherein receive unit 210, for receiving the collection data of each harvester of rear end equipment 20；Weights allocation unit 211, usesData distribute weights each of in the data by being gathered, and the data gathered include cpu, internal memory, wide mouth state etc.Utilization rate, weights allocation unit 211 is every kind of data one weights of distribution, such as CPU usage distribution weights A, and internal memory usesRate distribution weights B, wide mouth state utilization rate distribution weights C, concrete weights are depending on environment, such as: the internal memory of server 1 makesWith reaching 50%, the memory usage of server 2 is 30%, it is possible to by the weights shunting table given, carry out flowShunting, allows the server that more flux and flow direction utilization rate is low.Computing unit 212, for the collection number according to each serverAccording to calculating the comprehensive weights of each server, and obtain the server of comprehensive maximum weight；Request pilot unit 213, being used for willThe user's request received is directed to the server of comprehensive maximum weight.
Fig. 4 is the thin portion structural representation of rear end equipment 20 in another preferred embodiment of the present invention.As shown in Figure 4, rear endEquipment 20 includes multiple harvester 201 and multiple handover request unit 202, wherein, each harvester 201 and handover request listUnit 202 is arranged at each server, and harvester 201 is for the use information of each server of timing acquiring, and timing connects front end and setsStandby 21, the data gathered are transferred to headend equipment 21；Handover request unit 202, for the collection number according to harvester 201According to, it may be judged whether forward end equipment 21 sends path handover request, and when judged result is for being, forward end equipment sends pathHandover request, specifically, handover request unit 202 can be according to the collection data of harvester 201, it is judged that the use of current serverWhether rate reaches SC service ceiling, such as, judge to gather whether data (such as utilization rates such as cpu, internal memory, wide mouth states) reach defaultThreshold value, if all reaching default threshold value, then it is assumed that current server has reached SC service ceiling, then forward end equipment 21 sends roadFootpath handover request, issues path to ask headend equipment 21 to select flow else.
Fig. 5 is the detail structure chart of headend equipment 21 in another preferred embodiment of the present invention.In the present embodiment, except comprisingReceiving outside unit 210, weights allocation unit 211, computing unit 212 and request pilot unit 213, headend equipment 21 also includesPath switching unit 214, in time receiving path handover request, carries out flow and issues path switching, flow is issued path and cutsIt is changed to comprehensive weights and is only second to the path that the server of current path server is corresponding.Such as, present flow rate issues path is comprehensiveClosing the path of maximum weight, then switch to the passage that weights ranked second, the rest may be inferred.
Fig. 6 is the load balancing process schematic of the specific embodiment of a kind of load balancing apparatus of the present invention.In the present inventionIn specific embodiment, load balancing process is as follows: the harvester timing acquiring the forward end equipment that 1, are arranged on server returnThe data of back production collection；2, headend equipment is for gathering data distribution weights, and calculates the comprehensive weights of each server, according to combiningClose weights distribution data；3, the server sending flow rate of comprehensive maximum weight is selected；4, harvester timing acquiring data, according to adoptingCollection data judge the comprehensive utilization rate of server, when the comprehensive utilization rate of server reaches to a certain degree (such as cpu, internal memory, widthThe utilization rates such as mouthful state are above a threshold value preset) time return request to headend equipment with toggle path.
Fig. 7 is the flow chart of steps of a kind of load-balancing method of the present invention.As it is shown in fig. 7, one load balancing of the present inventionMethod, comprises the steps:
Step 701, utilizes the use information of each server of harvester timing acquiring, and will gather data timing transmission to the most frontEnd equipment.In the present invention, each server installs a harvester, associates the use letter of each server service condition with timing acquiringBreath, such as: the interactive information that CPU usage, memory usage, the bandwidth usage of port are applied with access etc. are a series ofInformation, each harvester timing can actively connect headend equipment, so that the data gathered are passed to headend equipment.
Step 702, headend equipment, after receiving collection data, distributes weights for gathering, calculates comprehensive weights in dataMaximum server, and the user received request is directed to corresponding server.Specifically, headend equipment is adopted in receivingAfter collection data, weights can be distributed for each of gathering in data data, such as, gather packet containing CPU usage, internal memoryUtilization rate, the bandwidth usage etc. of port, headend equipment then distributes a weights A, memory usage distribution one for CPU usageWeights B, the bandwidth usage of port distributes a weights C ..., it is then each server according to the collection data of each serverCalculate comprehensive weights, it is thus achieved that the server of comprehensive maximum weight, and the user received request is directed to comprehensive weightsBig server.
Fig. 8 is the thin portion flow chart of step 702 in present pre-ferred embodiments.Specifically, step 702 farther includes:
Step S1, receives the collection data of each harvester；
Step S2, distributes weights by data each of in the data that gathered, and the data gathered include cpu, interiorDeposit, the utilization rate such as wide mouth state, step S2 is every kind of data one weights of distribution, such as CPU usage distribution weights A, internal memoryUtilization rate distribution weights B, wide mouth state utilization rate distribution weights C, concrete weights are depending on environment；
Step S3, calculates the comprehensive weights of each server, and obtains synthetic weights according to the collection data of each serverThe server that value is maximum；
Step S4, is directed to the server of comprehensive maximum weight by the user received request.
It is preferred that one load-balancing method of the present invention, also comprise the steps:
Collection data according to harvester, it may be judged whether forward end equipment sends path handover request, and in judged resultDuring for being, forward end equipment sends path handover request.Specifically, according to the collection data of harvester, it is judged that current serverUtilization rate whether reach SC service ceiling, such as judge to gather whether data (such as utilization rates such as cpu, internal memory, wide mouth states) reachTo the threshold value preset, if all reaching default threshold value, then it is assumed that current server has reached SC service ceiling, then forward end equipment is sent outOutbound path handover request, to ask headend equipment alternative flow to issue path；
Headend equipment, in time receiving path handover request, carries out flow and issues path switching, flow is issued path and cutsIt is changed to comprehensive weights and is only second to the path that the server of current path server is corresponding.Such as, present flow rate issues path is comprehensiveClosing the path of maximum weight, then switch to the passage that weights ranked second, the rest may be inferred.
In sum, a kind of load balancing apparatus of the present invention and load-balancing method thereof are by utilizing the collection of rear end equipmentDevice gathers the service condition of each server, and sends the data gathered to headend equipment, utilizes headend equipment for gathering dataDistribution weights, and the comprehensive weights of each server are calculated according to collection data, ask user to guide comprehensive maximum weightServer, by the invention it is possible to make server enjoy more preferable load-balancing function.
The principle of above-described embodiment only illustrative present invention and effect thereof, not for limiting the present invention.AnyAbove-described embodiment all can be modified under the spirit and the scope of the present invention and change by skilled person.Therefore,The scope of the present invention, should be as listed by claims.