CN115529316A - Micro-service deployment method based on cloud computing center network architecture - Google Patents
Micro-service deployment method based on cloud computing center network architecture Download PDFInfo
- Publication number
- CN115529316A CN115529316A CN202211206589.1A CN202211206589A CN115529316A CN 115529316 A CN115529316 A CN 115529316A CN 202211206589 A CN202211206589 A CN 202211206589A CN 115529316 A CN115529316 A CN 115529316A
- Authority
- CN
- China
- Prior art keywords
- micro
- service
- server
- deployment
- microservice
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000005457 optimization Methods 0.000 claims abstract description 18
- 238000004891 communication Methods 0.000 claims abstract description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 45
- 230000006870 function Effects 0.000 claims description 44
- 238000010845 search algorithm Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 13
- 239000000126 substance Substances 0.000 claims description 10
- 230000002068 genetic effect Effects 0.000 claims description 7
- 230000005540 biological transmission Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000001934 delay Effects 0.000 claims description 4
- 238000012546 transfer Methods 0.000 claims description 3
- FGUUSXIOTUKUDN-IBGZPJMESA-N C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 Chemical compound C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 FGUUSXIOTUKUDN-IBGZPJMESA-N 0.000 claims 1
- 101150050759 outI gene Proteins 0.000 claims 1
- 230000004044 response Effects 0.000 abstract description 10
- 238000010586 diagram Methods 0.000 description 13
- 230000008569 process Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000011156 evaluation Methods 0.000 description 7
- 230000008878 coupling Effects 0.000 description 5
- 238000010168 coupling process Methods 0.000 description 5
- 238000005859 coupling reaction Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000003860 storage Methods 0.000 description 3
- 239000002243 precursor Substances 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
Abstract
The invention provides a micro-service deployment method based on a cloud computing center network architecture, which adopts a joint optimization mode to simultaneously solve the micro-service deployment problem and the user request routing problem. The system can support different types of services under certain constraint and can simultaneously respond to massive mobile user requests. In addition, the invention considers the interdependency among different micro services, effectively reduces the response time delay of the system to the user request by fully considering the communication dependency among the micro services, and improves the user application experience.
Description
Technical Field
The invention relates to the field of cloud computing, in particular to a micro-service deployment method based on a cloud computing center network architecture.
Background
The mobile internet industry has been developed rapidly, and the number of global users is approximately 44 hundred million. People are using myriad mobile applications such as WeChat, QQ, paibao, taobao, etc. every day. With the prosperity of mobile applications and the dramatic increase of user data volume, enterprise architecture of the mobile internet is overwhelmed, and thus, traditional monolithic, distributed, layered and Service Oriented Architecture (SOA) is shifted to the popular micro-service architecture in recent years.
With the increasing business forms, the requirements of users on service performance are higher and higher, and how to optimize the user experience becomes the most concerned problem of each large business on the premise of ensuring the positive increase of the revenue of service providers. The rapid popularization of the mobile internet and the explosive growth of various services bring great challenges to enterprises while bringing development opportunities. Under the scene of high-concurrency massive user requests, how to ensure the user service quality, and the micro-service architecture provides a good solution for the problems. The micro-services are modules with specific functions, one micro-service only focuses on one single function, and the micro-services are interactively cooperated to jointly complete user requests. The characteristics of high cohesion and low coupling of the micro-service are very easy to expand and maintain. Clearly, the micro-service architecture has irreplaceable advantages in service scenarios with a surge in traffic.
At present, no related work is available for detailed research on joint optimization of micro-service instance deployment and request routing of the cloud computing center. Much work has focused on the research of microservice deployment and load balancing, and the strong coupling of the microservice deployment and the load balancing is not fully utilized to optimize the time delay. Some of the research efforts have considered coarse-grained resource allocation, such as with servers as the smallest units of computing resources, which can result in a significant amount of wasted resources. In addition, research is also carried out on neglecting communication delay among micro-services, and only time consumption brought by computing resources is considered, but actually, the micro-services with dependency relationship are deployed on the same server, so that the network congestion probability can be greatly reduced, and the user service experience is greatly improved.
Disclosure of Invention
In order to solve the problem that a high-concurrency mass of user requests can reach a cloud computing center, the invention provides a micro-service deployment method in the cloud computing center, which integrates a micro-service technical architecture in the cloud computing, establishes a performance model with resource leasing cost as constraint, performs joint optimization on micro-service instance deployment and request routing, obtains an optimized micro-service deployment and user request routing scheme based on an improved greedy heuristic algorithm, a genetic algorithm, a local search algorithm and the like, and aims to minimize user request response time delay under the condition of meeting ASP cost constraint.
The cloud computing center network architecture comprises a plurality of switches and a plurality of servers, each switch is connected with the plurality of servers, a plurality of micro services are deployed on the servers of the cloud computing center, the combination of different micro services forms micro service chains with different functions, each micro service has a plurality of instances, and each type of user request corresponds to one micro service chain;
and obtaining an optimal micro-service deployment scheme based on the micro-service deployment strategy and the route request strategy combined optimization, wherein the micro-service deployment strategy comprises the specific number of each micro-service instance deployed on the server of the cloud computing center and the deployment position of each micro-service instance on the server of the cloud computing center, and the request route strategy comprises the specific route path of each user request between the servers of the cloud computing center.
The invention provides a cloud computing center network architecture, and provides a method for micro-service instance deployment and routing request in a network based on the network architecture, wherein a cloud computing center network model consisting of a switch and a server is designed, the switch is responsible for data transmission, the server is responsible for request processing, multiple micro-service instances are deployed on the server of the cloud computing center, and different combinations among the micro-services form micro-service chains with different functions; and with the resource leasing cost as a constraint, establishing a performance model of the total time delay of the system according to the queuing network; based on the model, a microservice instance deployment algorithm supporting various microservice chains is designed, the algorithm is improved based on a hybrid inheritance and local search algorithm, a plurality of instances of various microservices are dynamically deployed in a network, and a request routing strategy is correspondingly adjusted according to a deployment result; the request routing algorithm designed for the request from the user aims to find an effective path of the microservice in the network for the request, so that the system can more efficiently complete the processing of the request and return a corresponding result. The invention adopts a joint optimization mode to simultaneously solve the problem of micro-service deployment and the problem of user request routing, and particularly, the invention takes a micro-service instance deployment scheme as a precondition and takes response time delay obtained by calculating a request routing result as an evaluation standard to evaluate the instance deployment scheme, thereby fully utilizing the strong coupling relationship of the micro-service instance deployment scheme and the request routing result. The system can support different types of services such as WeChat, buffalo, youkou, paibao and the like under certain constraint, and can simultaneously respond to massive mobile user requests. In addition, the invention considers the interdependency among different micro services, effectively reduces the response time delay of the system to the user request by fully considering the communication dependency among the micro services, and improves the user application experience.
Drawings
Fig. 1 is a schematic structural diagram of a cloud computing center network architecture provided in the present invention;
FIG. 2 is a schematic diagram of a microservice and its deployment on a server;
FIG. 3 is a flowchart of a micro-service deployment and request routing joint optimization algorithm to obtain an optimal micro-service deployment scenario;
FIG. 4 is a schematic diagram of request routing;
fig. 5 is a flow chart of a request routing algorithm.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention. In addition, technical features of various embodiments or individual embodiments provided by the present invention may be arbitrarily combined with each other to form a feasible technical solution, and such combination is not limited by the sequence of steps and/or the structural composition mode, but must be realized by a person skilled in the art, and when the technical solution combination is contradictory or cannot be realized, such a technical solution combination should not be considered to exist and is not within the protection scope of the present invention.
Referring to fig. 1, a cloud computing center network architecture is provided, which is composed of a large cloud computing center including a plurality of switches and a plurality of servers. The switches are responsible for data forwarding, the servers are responsible for request processing and routing, the switches are connected with each other to form a connected graph, the switches are connected with each other through optical fibers, each switch is connected with a plurality of servers, one switch and the server connected with the switch form a small network, the servers in the small network are communicated with the servers in other small networks through the data forwarding of the switches, and the communication time delay among the servers in the same small network can be ignored. The difference between the micro service and the traditional service is that the micro service divides an application program into a plurality of modules with specific functions, and the modules cooperate with each other to jointly complete a user request. In order to guarantee the service quality, each micro service has a plurality of instances, and the user request can be selected from the plurality of instances to complete the self requirement. A plurality of micro services are combined into a linear chain according to a certain sequence, namely the micro service chain, and one micro service chain corresponds to one user request. After the user request reaches the system, the exchanger and the server cooperate to sequentially process and route the request according to the micro-service sequence on the corresponding service chain so as to complete the user requirement. The specific process is as follows: when a user request reaches an acceptance module of the cloud computing center, the acceptance module obtains a first micro-service instance deployment position of the request through table lookup, selects an instance with low prediction delay for the request according to the current state of the instance, routes the request to the selected instance, and when the first micro-service function is completed, the server selects a second micro-service instance for the request according to a routing algorithm. And by analogy, when each micro service in the chain is executed, the result is finally returned to the user.
The method comprises the steps that based on a constructed network architecture of the cloud computing center, an optimal micro-service deployment scheme is obtained based on a micro-service deployment strategy and a route request strategy combined optimization, wherein the micro-service deployment strategy comprises the specific number of each micro-service instance deployed on a server of the cloud computing center and the deployment position of each micro-service instance on the server of the cloud computing center, and the request route strategy comprises the specific route path of each user request between the servers of the cloud computing center.
It is understood that a delay model constrained by service provider (ASP) resource lease costs is built based on the network architecture; each micro service has a plurality of instances (mirrors), the micro service instance deployment strategy includes determining a specific number of each micro service instance and a deployment position of each micro service instance on a server of the cloud computing center, and as can be seen in fig. 2, a schematic diagram of deployment of a micro service on a server is shown; the user request routing strategy comprises the steps of designing a specific routing path of the request between the servers; the combined optimization strategy of micro-service instance deployment and request routing makes full use of the coupling relationship between the two, namely, the instance deployment is used as a precondition of the request routing strategy, and the time delay obtained after the request routing is used as a standard for measuring the quality of the deployment strategy. The algorithm optimizes the two problems simultaneously in the continuous iterative updating process to finally obtain an optimal micro-service deployment scheme and a request routing scheme, and finally the network architecture can process user requests under high-concurrency massive scenes.
As an embodiment, the microservice deployment policy includes: generating an initial deployment scheme of the micro-service instance based on a greedy algorithm; and obtaining an optimal micro-service deployment scheme based on iterative optimization of a hybrid genetic algorithm and a local search algorithm.
It is understood that fig. 3 is a flow chart of a method for deploying micro services, and referring to fig. 3, for the micro services, the deployment problem, the present invention adopts an improved hybrid genetic and local search algorithm to obtain a corresponding solution. In each iteration, the individual invokes a request routing algorithm to minimize request response latency under the resource lease cost constraint. The micro-service instance deployment algorithm flow is as follows:
algorithm 1: micro-service instance deployment algorithm:
input iteration number
Micro-service set M, server set I and initial population number
The total core number C of the server;
and outputting the individual S with the maximum fitness function.
Calculating the minimum core number of each micro service M in the set M
for 
do
Randomly selecting r micro services from M to obtain minimum core number
Extension to 2
for 
do
According to
And
selecting a deployment location for each instance of microservice m
end for
end for// initial population P 0 Has generated
for 
do
for 
do 
Get round upwards
From P 0 Two parent solutions P1 and P2 are selected
Obtaining offspring solutions C1 and C2 by applying cross algorithm
Performing local search on C1 and C2 to obtain optimized solutions S1 and S2
Putting S1 and S2 into a population P 0
end for
Calling request routing algorithm and calculating evaluation function of each individual
2P is added 0 The solutions are sorted according to the fitness function descending order
end for
Returning the solution S with the maximum fitness function value
The flow of the algorithm is as follows:
a. calculating the minimum core number required by each micro service in the micro service set, and randomly selecting r micro services from the micro service set to expand the required core number to be twice of the minimum core number;
b. based on the degree of dependence of any micro service on other micro services and the degree of dependence of any micro service on each server, selecting a deployment position for the instance of any micro service, and generating an initial population P0 of the micro service instance, wherein P0 is the number of individuals in the initial population;
c. selecting two solutions P1 and P2 from an initial population P0, respectively selecting one server from P1 and P2, exchanging micro-service instances deployed on the two servers to obtain solutions C1 and C2, executing a local search algorithm on C1 and C2 to obtain solutions S1 and S2, and putting the solutions S1 and S2 back into the initial population P0;
d. c is repeatedly executed until the number of individuals in the population is 2P0;
e. executing a request routing algorithm on 2P0 individuals in the population, and calculating a fitness function of each individual;
f. sorting the 2P0 individuals in a descending order according to a fitness function, and forming a new population by the P0 individuals which are sorted in the front as an initial population of the next iteration;
g. c-f is repeatedly executed until the iteration times reach the maximum iteration times
；
h. And selecting the individual S with the highest fitness function and meeting the leasing cost constraint from the last iteration result as a final micro-service deployment scheme.
It is understood that the number of cores occupied by the micro-service can be estimated from the queue stability condition. The request arrival rate of the microservice k is
Then the request arrival rate of all microservices on the service chain k is
. If there is only one instance per microservice on k, then for any microservice m on k, the queue stability condition can be derived
Then the minimum value of the initial core number of the micro service m can be obtained
。
The calculating the minimum core number required by each micro service in the micro service set comprises the following steps: calculating the minimum core number required by each micro service m according to the relation between the request arrival rate on the micro service chain and the core processing rate of the server
：
Wherein the content of the first and second substances,
is the arrival rate of requests containing a microservice m, the request arrival rate of a microservice chain n is
Then the request arrival rate of all the microservices on the microservice chain n is
，
Is the processing rate of the server core to the micro-service m.
As an embodiment, the selecting a deployment location for an instance of any one microservice based on the degree of dependency of the any one microservice on other microservices and the degree of dependency of the any one microservice on each server includes: calculating the degree of dependence of any micro service on each server according to the calculated degree of dependence of any micro service on each server, and deploying the instance of any micro service on the server with the highest degree of dependence; wherein the degree of dependency between two microservices is characterized based on data traffic between the two microservices and is expressed as:
the dependency degree of the micro service m1 on the server i is defined as the sum of data traffic between the micro service m1 and all micro services deployed on the server i, that is:
where m' is the microservice on server i.
It can be understood that, when the micro-service is deployed on the servers of the cloud computing center, the server on which the micro-service to be deployed is determined according to the degree of dependency between the micro-service to be deployed and the micro-service already deployed on each server. Specifically, instance deployment is performed according to the dependency degree between the micro services, that is, the micro service instances with high dependency degree are deployed on the same server, so as to reduce the request response delay.
The micro-service deployment scheme of the invention fully considers the dependency relationship between the micro-services and uses
The data traffic between the micro-services m1 and m2 is expressed as a criterion for evaluating the degree of dependency between the micro-services, and the larger the data traffic, the higher the degree of dependency between the two micro-services. On the basis, the degree of dependence of the microservice m on the server i is defined
. When deploying the micro-service instance, the micro-service m to be deployed is preferentially deployed to the degree of dependence
Thus reducing latency at higher servers.
When the deployment scheme of the micro-service is carried out, an iterative optimization method is adopted, firstly, an initial population P0 of a micro-service instance is generated, and P0 is the number of individuals in the initial population. For the generated initial population P0, a crossing method is adopted to expand a search space so as to avoid the search from being trapped in local optimization, a crossing algorithm has certain randomness and jump performance, and two individuals with higher evaluation functions are selected to be crossed, namely, example sets deployed on any two servers on the two individuals are exchanged. This operation may have a large influence on the genetic evaluation function of an individual, which may result in the evaluation function becoming better or worse, but it is this randomness that expands the search range of the algorithm and the algorithm can jump out of local optima. Specifically, two individuals P1 and P2 are selected from the initial population P0 of the generated micro-service instance, and the micro-service instances deployed on P1 and P2 are exchanged to obtain C1 and C2.
The cross method can prevent the algorithm from falling into local optimum, but because the jump variability is too strong, the global optimum solution can be missed, so the local search algorithm is adopted to search the solution space carefully after the cross method, and the problem of slow algorithm convergence speed caused by the jump characteristic of the cross method is avoided. In the invention, the local search algorithm is defined as adding or subtracting one operation to the number of instance cores on the basis of the instance deployment of an original individual, and finally, the operation which reduces the time delay to the maximum is selected as the result of local search. The neighborhood of an individual is defined as follows:
wherein the content of the first and second substances,
is a mini-suitService M occupies the number of cores on server i, M is the micro-service set,
is a collection of servers with microservice m deployed.
The specific operation of the local search algorithm is that, for the above C1 and C2, one microservice instance is added to C1 or one microservice instance is reduced to obtain S1, then C2 is unchanged, and S2= C2; or adding a micro-service instance to C2 or subtracting a micro-service instance to obtain S2, then C1 is unchanged, and S1= C1. And (3) after the C1 and the C2 are subjected to local search algorithm processing, S1 and S2 are obtained, and the S1 and the S2 are added into the initial population P0. And for the initial population, repeatedly adopting a cross method and a local search algorithm to expand the number of individuals of the initial population until the number of individuals of the initial population reaches 2P0.
And then, calling a request routing algorithm for 2P0 individuals in the population, and calculating a fitness function of each individual. In the iteration process of the micro-service instance deployment algorithm, a request routing algorithm is called to obtain a fitness function of a current individual, and the advantages and disadvantages of a current deployment and routing scheme are evaluated according to the fitness function, wherein the flow of the request routing algorithm is as follows:
and 2, algorithm: request routing algorithm:
inputting a micro-service instance deployment scheme P and a user request n
Routing scheme Rn for outgoing requests n
for 
do// Ln is the set of microservices on the request chain n
if micro service m successor service is null
break
else
The successor service of the micro-service m is m s
Remember that the instance of executing microservice m selected for request n by the last iteration is m j
for 
do // J m s Is a micro-service m s Example set of,// m s q Is m s Q example of (1)
Computing
And
computing
end for
To generate [0,1]Random number in between
for 
do
if 
end if
end for
end else
end for
The request routing algorithm comprises the following steps:
the method comprises the following steps: and for the current micro service m of the micro service chain n, judging whether the micro service m is the last service on the request chain n, if so, exiting the routing algorithm and outputting a routing result.
Step two: subsequent microservice if microservice m exists
Then to
Each example of (a)
The following operations are performed. Recording the execution instance selected for the micro service m by the last iteration as the
。
Step three: calculating out
To
Routing probability based on distance between servers where the two servers are located
And calculating
To
Routing probability based on the number of cores on server occupied by both
。
Step four: according to
Computing
Is routed to
Final probability of
。
Step five: to generate [0,1]Random number in between
。
Step six: according to the generated random number
And the final probability obtained by calculation
For micro-service
Selecting a specific instance to be executed, and adding the selected instance and the corresponding probability to the routing result
In (1).
Wherein, for example, the subservices of microservices m
There are three examples, calculation
Obtaining three corresponding probabilities according to the final probabilities among the three examples, and respectively connecting the three probabilities with the generated random numbers
A comparison is made and an instance of the selected route is determined based on the comparison. For example, the three probabilities are 0, respectively.2. 0.3 and 0.5, then if a random number is generated
Then routing to the first instance; if generated random number
Then routing to the third instance; if generated random number
Then the route is selected to the second instance.
Step seven: and repeating the first step to the sixth step to obtain the routing paths of all the micro services of the micro service chain n.
It can be appreciated that in finding a routing path for the micro-service chain n, a request routing algorithm based on a weighted sum of server distance and core number includes routing user requests on a specific deployment scenario. And sequentially calculating the probability of all the examples of each micro service route to the subsequent service thereof according to the sequence of the micro services on the request chain, namely obtaining the possible route path of each service chain request. The routing probability mainly comprises a probability based on the distance between the servers and a probability based on the number of example cores, and a weighting factor is used
Indicating the degree of influence of both on the final probability.
And for each individual, after a request algorithm is called, calculating a fitness function of each individual, evaluating the quality of the individuals in the population through the fitness function, reserving the individual with a large fitness value in the iterative optimization process, and discarding the individual with a small fitness value. The fitness function is expressed as the inverse of the optimization objective, i.e.:
Wherein the total latency of a chain of microservices n in the system comprises the linger latency at the microservices on the chain
And communication delay of data in network
。
Stay time delay of microservice chain
The sum of the linger delays at each microservice on the chain is calculated as:
wherein, the first and the second end of the pipe are connected with each other,
representing the queuing delay at the instance of microservice m on server i,
representing the computation latency at the instance of microservice m on server i,
whether the micro service m on the service chain n and the precursor node thereof are deployed on the same server or not is shown, and if yes, the micro service m and the precursor node thereof are deployed on the same server
Is 0, otherwise is 1.
Representing a collection of micro-services on a service chain n,
representing a set of servers on the n routing path of the request chain.
Communication delay of microservice chain n in network
By data forwarding delay on the switch
And data transmission delay between the switches, namely:
wherein the content of the first and second substances,
representing the data forwarding delay at the switch s,
representing the total amount of data to be transmitted on switch s,
indicating the rate of data transfer between the switches,
representing a set of switches on the microservice chain n-route path.
Total time delay of system
I.e., the sum of the time delays to all microservice chains within the cloud computing center for a period of time T, i.e.:
after the fitness functions of 2P0 individuals are calculated, 2P0 individuals are selectedSorting the volumes in descending order according to fitness function, and sorting the volumes in the front
And forming a new population by the individuals to serve as an initial population for the next iteration. Repeatedly executing iterative processing until the iterative times reach the maximum iterative times
. And selecting the individual S with the highest fitness function and meeting the leasing cost constraint from the last iteration result as a final micro-service deployment scheme.
It should be noted that the present invention is mainly divided into two parts, namely micro service instance deployment and request routing. First, an application is divided into a plurality of microservices, each microservice having a plurality of instances. The micro-services with larger data traffic are deployed in the same server or the same small network according to the dependency relationship among the micro-services, so that the communication delay can be greatly reduced, and the user experience is improved. After the micro-service instance is deployed, the request reaches the cloud computing center, and a path needs to be planned for the request to complete the user request with low time delay. Microservice deployment and request routing flow diagrams are shown in fig. 2 and 3. The invention establishes a performance model aiming at minimizing the time delay with the ASP resource leasing cost as a constraint, and evaluates the advantages and disadvantages of deployment and routing schemes by calculating the queuing time delay, the calculation time delay, the data forwarding time delay and the data transmission time delay requested in a system.
As an embodiment, the method for deploying the micro service includes the steps of selecting an individual S with the highest fitness function and meeting the leasing cost constraint from the last iteration result as a final micro service deployment scheme, and includes: calculating a fitness function of each individual and the number of server cores required by the micro-service deployment scheme of each individual in the last iteration result, and determining a required server and a required switch according to the required number of the server cores; calculating the resource lease cost Y corresponding to the micro-service deployment scheme of each individual according to the required server and the required switch:
wherein, the first and the second end of the pipe are connected with each other,
is the overhead of leasing one server,
in order to lease the overhead of one switch,
for the number of servers to be leased,
number of switches leased.
And (3) limiting the resource leasing cost, namely leasing cost constraint:
wherein the content of the first and second substances,
maximum value of resource lease cost for obtaining forward profit.
And selecting the individual S with the highest fitness function and meeting the leasing cost constraint as a final micro-service deployment scheme based on the fitness function and the resource leasing cost of each individual in the last iteration result.
It can be understood that the last iteration result includes a plurality of individuals, the fitness function and the resource lease cost of each individual are calculated, and the individual with the highest fitness and the resource lease cost meeting the lease cost constraint condition is taken as the final micro-service deployment scheme.
The embodiment of the invention provides a micro-service deployment method based on a cloud computing center, which mainly comprises the following steps:
(1) A network architecture of a cloud computing center supporting micro-services is designed, and the micro-services are deployed on servers of the cloud computing center. Different micro services form a micro service chain with specific functions according to a certain sequence. When the user request reaches the data center, the system selects proper micro-service examples in the system to sequentially complete the user request according to the composition of the micro-service chain corresponding to the user request.
(2) The research content is divided into two parts of micro-service instance deployment and request routing. The micro-service instance deployment comprises the step of determining the number of the micro-service instances and the instance deployment positions, and the request routing part comprises the step of determining the specific routing strategy of the request in the network, namely determining the transmission path of the user request in the network. And accordingly, establishing a total time delay model of the system.
(3) And performing joint optimization on the micro-service instance deployment and the request routing, and evaluating the current deployment scheme by taking the response time delay obtained by calculation after the request routing as an evaluation standard in each algorithm iteration process, wherein different deployment schemes have different routing strategies and response time delays.
(4) The method solves the problem of micro-service instance deployment based on an improved genetic algorithm and a local search combined algorithm, and obtains a deployment scheme of multiple instances of multiple micro-services on a server of a cloud computing center.
(5) The request routing problem is solved by an algorithm based on distance and computing resource weighted transfer probability, and a routing path of a user request on a server of the cloud computing center is obtained.
The invention adopts a joint optimization mode to simultaneously solve the problem of micro-service deployment and the problem of user request routing, and particularly, the invention takes a micro-service instance deployment scheme as a precondition and takes response time delay obtained by calculating a request routing result as an evaluation standard to evaluate the instance deployment scheme, thereby fully utilizing the strong coupling relationship of the micro-service instance deployment scheme and the request routing result. The system can support different types of services under certain constraint and can simultaneously respond to massive mobile user requests. In addition, the invention considers the interdependency among different micro services, effectively reduces the response time delay of the system to the user request by fully considering the communication dependency among the micro services, and improves the user application experience.
It should be noted that, in the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to relevant descriptions of other embodiments for parts that are not described in detail in a certain embodiment.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A micro-service deployment method based on a cloud computing center network architecture is characterized in that the cloud computing center network architecture comprises a plurality of switches and a plurality of servers, each switch is connected with the plurality of servers, a plurality of micro-services are deployed on the servers of a cloud computing center, the combination of different micro-services forms a micro-service chain with different functions, each micro-service has a plurality of instances, and each type of user request corresponds to one micro-service chain;
and obtaining an optimal micro-service deployment scheme based on the micro-service deployment strategy and the route request strategy combined optimization, wherein the micro-service deployment strategy comprises the specific number of each micro-service instance deployed on the server of the cloud computing center and the deployment position of each micro-service instance on the server of the cloud computing center, and the request route strategy comprises the specific route path of each user request between the servers of the cloud computing center.
2. The method of claim 1, wherein the microservice deployment policy comprises:
generating an initial deployment scheme of the micro-service instance based on a greedy algorithm;
and obtaining an optimal micro-service deployment scheme based on iterative optimization of a hybrid genetic algorithm and a local search algorithm.
3. The method of micro-service deployment as claimed in claim 2, wherein the greedy-based algorithm generating an initial deployment scenario for micro-service instances comprises:
calculating the minimum core number required by each micro service in the micro service set, and randomly selecting r micro services from the micro service set to expand the required core number to be twice of the minimum core number;
based on the degree of dependence of any micro service on other micro services and the degree of dependence of any micro service on each server, selecting a deployment position for the instance of any micro service, and generating an initial population P0 of the instance of the micro service, wherein P0 is the number of individuals in the initial population;
the optimal micro-service deployment scheme is obtained based on iterative optimization of the hybrid genetic algorithm and the local search algorithm, and comprises the following steps:
a. selecting two solutions P1 and P2 from an initial population P0, respectively selecting one server from P1 and P2, exchanging micro-service instances deployed on the two servers to obtain solutions C1 and C2, executing a local search algorithm on C1 and C2 to obtain solutions S1 and S2, and putting the solutions S1 and S2 back into the initial population P0;
b. repeating the step a until the number of individuals in the population is 2P0;
c. executing a request routing algorithm on 2P0 individuals in the population, and calculating a fitness function of each individual;
d. sorting the 2P0 individuals in a descending order according to a fitness function, and sorting the individuals in the top order
Forming a new population by the individuals as an initial population of the next iteration;
e. repeatedly executing a-d until the iteration times reach the maximum iteration times
；
f. And selecting the individual S with the highest fitness function and meeting the leasing cost constraint from the last iteration result as a final micro-service deployment scheme.
4. The method of claim 3, wherein the calculating the minimum number of cores required for each micro-service in the set of micro-services comprises:
calculating the minimum core number required by each micro service m according to the relation between the request arrival rate on the micro service chain and the core processing rate of the server
：
Wherein the content of the first and second substances,
is the arrival rate of requests containing a microservice m, the request arrival rate of a microservice chain n is
The request arrival rate of all the microservices on the microservice chain n is
，
Is the processing rate of the server core to the micro-service m.
5. The method of micro-service deployment according to claim 3, wherein said selecting a deployment location for an instance of any one of the micro-services based on the degree of dependence of said any one micro-service on other micro-services and the degree of dependence of said any one micro-service on each server comprises:
calculating the degree of dependence of any micro service on each server according to the calculated degree of dependence of any micro service on each server, and deploying the instance of any micro service on the server with the highest degree of dependence;
wherein the degree of dependency between two microservices is characterized based on the data traffic between the two microservices and is expressed as:
the degree of dependence of the micro-service m1 on the server i is defined as the sum of data traffic between the micro-service m1 and all micro-services deployed on the server i, i.e.:
where m' is the microservice on server i.
6. The method of claim 3, wherein the local search algorithm is defined such that the neighborhood of solution S is a space obtained by adding or subtracting one to or from the number of cores of the instance of the micro-service deployed in S, and the neighborhood of solution S N (S) can be expressed as:
wherein the content of the first and second substances,
is the number of cores on server I occupied by microservice M, M being the microservice set, I m Is deployed with microservicesm, server set.
7. The method of claim 3, wherein the request routing algorithm is:
the method comprises the following steps: judging whether the current micro service m is the last service on the request chain n or not according to each micro service on the micro service chain corresponding to the user request, if so, exiting the routing algorithm, and outputting a routing result;
step two: if the micro-service m has a subsequent service
Then to
Each example of (a)
Executing the third step to the sixth step, and recording the execution instance selected for the microservice m by the last iteration as an execution instance
；
Step three: calculating out
To
Routing probability based on distance between servers where the two servers are located
And calculating
To
Routing probability based on the number of cores on server occupied by both
；
Step four: according to
Calculating out
Is routed to
Final probability of
；
Step five: to generate [0,1]Random number in between
；
Step six: according to the generated random number
And the final probability obtained by calculation
For micro-service
Selecting instances of specific executions
And will select the instance
And corresponding final probability is added to the routing result
Performing the following steps;
step seven: and repeating the first step to the sixth step until each micro service on the micro service chain corresponding to the user request selects a specific example of the route, and obtaining a final route result of the user request.
8. The method of micro-service deployment according to claim 3, wherein the fitness function is:
wherein, T dc The total system latency is the sum of the latencies to all the microservice chains in the cloud computing center over a period of time T.
9. The method of micro-service deployment according to claim 8, characterized in that the system total time delay T is calculated by dc ：
Lingering delay of microservice chain n
The sum of the linger time delay of each micro service on the micro service chain n is calculated to obtain:
wherein the content of the first and second substances,
representing the queuing delay at the instance of microservice m on server i,
representing the computation latency at the instance of microservice m on server i,
whether the micro-service m on the service chain n is deployed on the same server as the predecessor micro-service m on the service chain n or not is represented, and if yes, the micro-service m and the predecessor micro-service are deployed on the same server
Is 0, otherwise is 1,
representing a collection of micro-services on a service chain n,
representing a server set on a request chain n routing path;
communication delay of micro service chain n in network
Data forwarding delay from switch
And data transmission delay between the switches, namely:
wherein the content of the first and second substances,
representing the data forwarding delay at the switch s,
representing the total amount of data to be transmitted on switch s,
indicating the data transfer rate between the switches,
representing a switch set on a routing path of the microservice chain n;
the total system delay is the sum of the delays of all microservice chains in the cloud computing center within a period of time T, that is:
10. the method of claim 3, wherein the step f of selecting the individual S with the highest fitness function and satisfying the lease cost constraint from the last iteration result as the final microservice deployment scenario includes:
calculating a fitness function of each individual and the number of server cores required by the micro-service deployment scheme of each individual in the last iteration result, and determining a required server and a required switch according to the required number of the server cores;
calculating the resource lease cost Y corresponding to the micro-service deployment scheme of each individual according to the required server and the required switch:
wherein the content of the first and second substances,
is the overhead of leasing one server,
in order to lease the overhead of one switch,
for the number of servers to be leased,
number of switches leased;
and (3) limiting the resource leasing cost, namely leasing cost constraint:
wherein the content of the first and second substances,
maximum value of resource leasing cost for obtaining forward income;
and selecting the individual S with the highest fitness function and meeting the leasing cost constraint as a final micro-service deployment scheme based on the fitness function and the resource leasing cost of each individual in the last iteration result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211206589.1A CN115529316A (en) | 2022-09-30 | 2022-09-30 | Micro-service deployment method based on cloud computing center network architecture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211206589.1A CN115529316A (en) | 2022-09-30 | 2022-09-30 | Micro-service deployment method based on cloud computing center network architecture |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115529316A true CN115529316A (en) | 2022-12-27 |
Family
ID=84700543
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211206589.1A Pending CN115529316A (en) | 2022-09-30 | 2022-09-30 | Micro-service deployment method based on cloud computing center network architecture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115529316A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116233262A (en) * | 2023-05-07 | 2023-06-06 | 湖北省楚天云有限公司 | Micro-service deployment and request routing method and system based on edge network architecture |
| CN116627660A (en) * | 2023-07-24 | 2023-08-22 | 湖北省楚天云有限公司 | Micro-service resource allocation method based on cloud data center |
| CN117201319A (en) * | 2023-11-06 | 2023-12-08 | 三峡高科信息技术有限责任公司 | Micro-service deployment method and system based on edge calculation |
| CN117573379A (en) * | 2024-01-16 | 2024-02-20 | 国网湖北省电力有限公司信息通信公司 | Micro-service deployment method based on symmetrical scaling merging |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022007212A1 (en) * | 2020-07-10 | 2022-01-13 | 南京邮电大学 | Software microservice combination optimization method for edge cloud system |
| CN114338504A (en) * | 2022-03-15 | 2022-04-12 | 武汉烽火凯卓科技有限公司 | Micro-service deployment and routing method based on network edge system |
| CN114615338A (en) * | 2022-04-11 | 2022-06-10 | 河海大学 | Micro-service deployment method and device based on layer sharing in edge environment |
-
2022
- 2022-09-30 CN CN202211206589.1A patent/CN115529316A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022007212A1 (en) * | 2020-07-10 | 2022-01-13 | 南京邮电大学 | Software microservice combination optimization method for edge cloud system |
| CN114338504A (en) * | 2022-03-15 | 2022-04-12 | 武汉烽火凯卓科技有限公司 | Micro-service deployment and routing method based on network edge system |
| CN114615338A (en) * | 2022-04-11 | 2022-06-10 | 河海大学 | Micro-service deployment method and device based on layer sharing in edge environment |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116233262A (en) * | 2023-05-07 | 2023-06-06 | 湖北省楚天云有限公司 | Micro-service deployment and request routing method and system based on edge network architecture |
| CN116627660A (en) * | 2023-07-24 | 2023-08-22 | 湖北省楚天云有限公司 | Micro-service resource allocation method based on cloud data center |
| CN117201319A (en) * | 2023-11-06 | 2023-12-08 | 三峡高科信息技术有限责任公司 | Micro-service deployment method and system based on edge calculation |
| CN117201319B (en) * | 2023-11-06 | 2024-01-09 | 三峡高科信息技术有限责任公司 | Micro-service deployment method and system based on edge calculation |
| CN117573379A (en) * | 2024-01-16 | 2024-02-20 | 国网湖北省电力有限公司信息通信公司 | Micro-service deployment method based on symmetrical scaling merging |
| CN117573379B (en) * | 2024-01-16 | 2024-03-29 | 国网湖北省电力有限公司信息通信公司 | Micro-service deployment method based on symmetrical scaling merging |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115529316A (en) | Micro-service deployment method based on cloud computing center network architecture | |
| Sun et al. | TIDE: Time-relevant deep reinforcement learning for routing optimization | |
| CN108260169B (en) | QoS guarantee-based dynamic service function chain deployment method | |
| Liu et al. | DRL-R: Deep reinforcement learning approach for intelligent routing in software-defined data-center networks | |
| CN114338504B (en) | Micro-service deployment and routing method based on network edge system | |
| Wang et al. | Scheduling with machine-learning-based flow detection for packet-switched optical data center networks | |
| Wang et al. | Towards network-aware service composition in the cloud | |
| Quang et al. | Multi-domain non-cooperative VNF-FG embedding: A deep reinforcement learning approach | |
| CN113708972B (en) | Service function chain deployment method and device, electronic equipment and storage medium | |
| CN109358971B (en) | Rapid and load-balancing service function chain deployment method in dynamic network environment | |
| Wang et al. | Presto: Towards efficient online virtual network embedding in virtualized cloud data centers | |
| He et al. | QoE-based cooperative task offloading with deep reinforcement learning in mobile edge networks | |
| Liu | Intelligent routing based on deep reinforcement learning in software-defined data-center networks | |
| CN109614215A (en) | Stream scheduling method, device, equipment and medium based on deeply study | |
| CN109379230B (en) | Service function chain deployment method based on breadth-first search | |
| Nguyen et al. | Toward adaptive joint node and link mapping algorithms for embedding virtual networks: A conciliation strategy | |
| CN113472671A (en) | Method and device for determining multicast route and computer readable storage medium | |
| CN114466407A (en) | Network slice arranging algorithm based on particle swarm heredity | |
| Zhao et al. | Flow aggregation through dynamic routing overlaps in software defined networks | |
| Chen et al. | QMORA: A Q-learning based multi-objective resource allocation scheme for NFV orchestration | |
| Guler et al. | Genetic algorithm enabled virtual multicast tree embedding in Software-Defined Networks | |
| CN112995032B (en) | Segment routing traffic engineering method and device based on limited widest path | |
| Pham et al. | Multi-domain non-cooperative VNF-FG embedding: A deep reinforcement learning approach | |
| Liu et al. | Balanced off-chain payment channel network routing strategy based on weight calculation | |
| Yang et al. | Replica placement in content delivery networks with stochastic demands and M/M/1 servers |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |