TWI729613B - Method of virtual network function deployment in service function chain - Google Patents

Abstract

The present invention is a method of virtual network function deployment in a service function chain. The method is used to calculate the deployment position of the virtual network function in the service function chain. When the Internet Service Provider provides user-customized network services through the service function chain, multiple virtual network functions need to be deployed in the physical server of the cloud environment at the same time, and the packets are sequentially routed to different virtual network functions. The method described in the present invention can get a better deployment location, so that when the service is actually running, the calculation load of the physical server is reduced, and the network delay when the user accesses the service is reduced.

Description

Virtual network function deployment decision method of service function chain

The present invention relates to a virtual network function deployment mechanism, in particular to a virtual network function deployment decision method of a service function chain.

The current cloud environment uses a large number of virtualization technologies, and network device function virtualization is also one of them. Through the virtual network function technology, the establishment of network services can be more flexible and the physical server resources can be fully utilized. In the existing technology, the virtual network function is usually deployed in a way that is evenly distributed to different physical servers, so that the resources of each physical server can be evenly distributed. However, in some cases, this will Increasing the CPU computing capacity of the physical server and causing additional network delay time when users access these services, and this situation becomes more serious after the service function chain technology starts to be used.

In addition, the service function chain technology connects one or several virtualized network functions through the network to provide corresponding network functions. When deploying a service function chain in the prior art, each virtualized network function of a series of service function chains is evenly distributed to different physical servers. This makes users need to go through a lot when accessing this series of service function chains. Different physical servers, this will cause additional network delay time.

In addition, since the connection between the physical servers uses Generic Routing Encapsulation Tunnel (GRE Tunnel) technology, the Generic Routing Encapsulation Tunnel (GRE Tunnel) technology needs to process the packet header (Header) at both the transmitting end and the receiving end. , If it passes through many physical servers, this will also increase the time for users to access the service.

It can be seen from the above that there is a need to find a way to solve the above problems in the existing cloud virtualization environment, especially for the network delay caused by the average distribution of each virtualized network function to different physical servers or increase the access Service time, and propose measures to improve the current technology, which will become an urgent goal pursued by those skilled in the art.

The purpose of the present invention is to propose a virtual network function deployment location decision method for the network environment established by the virtual network function of the service function chain, so that the CPU operation of the physical server can be reduced when the virtual network function is deployed. It also reduces the network latency of users accessing services. In addition to saving the available resources of the physical server, it can also provide users with better service quality.

To achieve the above-mentioned object of the invention or other objects, the present invention discloses a virtual network function deployment decision method for a service function chain, which includes: setting a network node as the starting point; setting the network connection between nodes in the network The initial value of the weight of is the same; search for multiple computing nodes connected to the starting point, find the computing node with the largest number of virtual network functions that can be deployed, and define it as the selected computing node; determine the resources of the selected computing node Whether it is sufficient to deploy the virtual network function to determine the deployment of the virtual network function on the selected computing node when the resources of the selected computing node are sufficient; increase a weight of the same value to pass from the starting point to deployed The network connection of the computing node of the virtual network function; and determine whether there are other virtual network functions that have not been deployed, so as to complete the deployment of all virtual network functions when there are no other virtual network functions that have not been deployed, or When there are other virtual network functions that have not yet been deployed, take the computing node of the deployed virtual network function as a new starting point, and repeat the above steps of searching, judging and adding weights until all virtual network functions are completed deploy.

In the above method, if there are multiple operation nodes with the most available resources at the same time, find the operation node with the shortest path among the plurality of operation nodes with the most available resources, and select the operation node with the shortest path as the selected operation node.

In the above method, the computing node with the shortest path refers to the one with the smallest total weight of the network connection from the starting point to the selected computing node.

In the above method, if there is only one computing node with the most available resources, it is selected as the selected computing node.

In the above-mentioned method, if it is determined that the resources of the selected computing node are insufficient to deploy the virtual network function, then the subordination of the virtual network function is terminated.

In the above method, the network node is a physical server, which is used to implement deployment decisions of all virtual network functions.

In the above method, the multiple computing nodes are all physical servers for the deployment of various virtual network functions.

In the above method, a number of virtual network functions are cascaded and combined to form a service function chain.

In the above method, the network connection between each node has no directionality.

In the above method, the virtual network function or the other virtual network function includes a firewall function, a load balancing function, a gateway function, or a forwarding function.

In summary, the virtual network function deployment decision method of the service function chain disclosed in the present invention provides a decision method to calculate the deployment position of the virtual network function in the service function chain. After a network node is selected, Search for the next suitable computing node, that is, the one that can deploy the largest number of virtual network functions and has sufficient resources. Regarding the problem of traffic distribution, whenever there is a virtual network function's traffic passing, the weight of the network connection will be increased. It is used to record the flow of this network connection, so that when the virtual network function is deployed, the path with a smaller weight is selected, that is, the path where the virtual network function's traffic passes fewer times, so that the network traffic can be Evenly distribute, reduce the occurrence of network congestion, through the decision-making method of the present invention, a better deployment location can be calculated, and this can reduce the CPU operation of the physical server when the service is actually running, and reduce the user's presence The network delay when fetching the service.

100‧‧‧Network Node

201, 202, 203‧‧‧operation node

300, 300’‧‧‧Service Function Chain

3100‧‧‧Firewall function

3200‧‧‧Load balancing function

3300‧‧‧Gateway function

3400‧‧‧Redirecting function

401, 402, 403‧‧‧Web server

500, 501‧‧‧User

601~611‧‧‧Process

S11~S16‧‧‧Step

Please refer to the detailed description of the present invention and its accompanying drawings to further understand the technical content of the present invention and its objectives and effects.

Figure 1 is a step diagram of the virtual network function deployment decision method of the service function chain of the present invention.

Figures 2A-2C are schematic diagrams of the operation of a user accessing a web server using the method of the present invention.

Figure 3 is a flowchart of the practical application of the virtual network function deployment decision method of the service function chain of the present invention.

The following describes the technical content of the present invention with specific specific embodiments. Those familiar with the art can easily understand the advantages and effects of the present invention from the content disclosed in this specification. However, the present invention can also be implemented or applied by other different specific embodiments.

The virtual network function deployment decision method of the service function chain proposed in the present invention is based on the concept of placing the virtual network functions of the same service function chain on the same physical server as much as possible, thereby reducing users’ access to the virtual network. The number of physical servers that the function passes through. Furthermore, the different service function chains are evenly placed on different physical servers, so that the resource usage of each physical server can be evenly distributed.

Figure 1 is a step diagram of the virtual network function deployment decision method of the service function chain of the present invention, and illustrates how the present invention performs the subordination of the virtual network function to the service function chain. In step S11, a network node is set as a starting point. This step is a network node when a user (user) wants to obtain a network service when establishing a network service connection. This network node is a physical server used to implement deployment decisions for all virtual network functions.

In step S12, the initial value of the weight of the network connection between each node in the network is set to be the same. The present invention deals with the problem of traffic distribution through a weighting method, that is, when the traffic with virtual network function passes, the network connection can be weighted, so as to record the traffic passing by the network connection, so that in the deployment of virtual For network functions, choosing a path with a smaller weight will help evenly distribute the traffic. In this case, the initial value of the weight of the network connection between all nodes will of course be the same at the beginning, and then there will be virtual networks. The weight value needs to be increased only when the flow of the function passes.

In step S13, search for multiple operation nodes connected to the starting point to find The computing node with the largest number of virtual network functions that can be deployed is defined as the selected computing node. This step is to find the next node connected to the starting point, which is called a computing node here, and the search rule is that the computing node with the largest number of virtual network functions is the first choice. After finding the computing node, the computing node is called These are the selected computing nodes. Specifically, the multiple computing nodes are physical servers for the deployment of various virtual network functions.

In one embodiment, if there is only one computing node with the most available resources, of course, the computing node is selected as the selected computing node. Conversely, if there are multiple computing nodes with the most available resources at the same time, the multiple available resources Find the operation node with the shortest path among the most operation nodes, and select the operation node with the shortest path as the selected operation node. The aforementioned arithmetic node with the shortest path refers to the one with the smallest total weight of the network connection from the starting point to the selected arithmetic node. It can be seen that when there are multiple computing nodes with the most available resources, the computing node with the shortest path is selected as the selected computing node, and the definition of the shortest path is the network connection from the starting point to the computing node After the weights are added up, the total value is the smallest.

In step S14, it is determined whether the resources of the selected computing node are sufficient to deploy the virtual network function, so as to determine to deploy the virtual network function on the selected computing node when the resources of the selected computing node are sufficient. This step is to judge whether the resources of the selected computing node are sufficient to deploy the virtual network function. Since the selected computing node is the component selected to perform the virtual network function, it is necessary to first confirm whether the selected computing node is the resource Abundant, when resources are abundant, virtual network functions can be deployed.

In one embodiment, if it is determined that the resources of the selected computing node are insufficient to deploy the virtual network function, then the subordinated virtual network function is terminated.

In step S15, add a weight of the same value to the starting point The network connection of the computing node where the virtual network function has been deployed. This step is to increase the weight of the network connection passing through after the selected computing node is deployed with the virtual network function, that is, for this deployment of the virtual network function, the starting point and the selected computing node Increase the weight of the network connection so that when you continue to execute the deployment, you know that the network connection has a higher weight, and avoid choosing this node as the computing node for the next deployment. That is, the higher the weight value, the path selection The lower the chance of being selected.

In step S16, it is determined whether there are other virtual network functions that have not yet been deployed, so as to complete the deployment of all virtual network functions when there are no other virtual network functions that have not been deployed, or if there are other virtual networks that have not yet been deployed When functioning, take the computing node that has deployed the virtual network function as a new starting point, and repeat the above steps of searching, judging, and adding weight until all virtual network functions are deployed. This step is to determine whether there are other virtual network functions to be deployed. If not, it means that all virtual network functions are deployed. On the contrary, if there are other virtual network functions to be deployed, steps S13-S15 are executed again until All network virtual network functions are deployed. The virtual network function deployment decision method of the service function chain of the present invention is an entity that needs to simultaneously deploy multiple virtual network functions in a cloud environment when ISP operators provide users with customized network services through the service function chain. In the server, the packets are routed to different virtual network functions in sequence.

It can be seen from the above that several virtual network functions will become a service function chain after being cascaded and combined. The present invention places the virtual network functions of the same service function chain on the same physical server as much as possible, thus reducing user presence. Take the number of physical servers that the virtual network function passes through. This can avoid the network delay time problem caused by many different physical servers when accessing the service function chain. Moreover, the different service function chains are evenly placed on Different physical servers, so that every The resource usage of two physical servers can be evenly distributed, so that the average service function chain can be distributed to different physical servers, thereby reducing the computing load of the physical servers.

The virtual network function described in the present invention includes a firewall function, a load balancing function, a gateway function or a forwarding function, but is not limited to this.

Figures 2A-2C illustrate a schematic diagram of the operation of a user accessing a web server using the method of the present invention, that is, an example of the composition of related components when the present invention executes network function deployment. This embodiment illustrates the user accessing a web server The process of obtaining network services and the relationship between components.

First of all, when the method of the present invention is applied in practice, it can be explained with Graph Theory, that is, using graph theory to convert network topology into graph and physical server into vertex, the present invention The physical server includes a network node (Network Node) and a computing node (Compute Node). The network connection between the two physical servers can be called an edge. The edge has no directionality, and each Each side has the weight (Weight). Whenever the traffic with the virtual network function passes, we can increase the weight of the side to record the flow of this side. In this way, when deploying the virtual network function, choose the smaller weight Path, that is, the path that the virtual network function's traffic passes less frequently, so that the network traffic can be evenly distributed, reducing the occurrence of network congestion.

As shown in Figure 2A, the physical server of the present invention includes a network node 100 and computing nodes 201, 202, and 203. The network node 100 is the starting position of the decision-making method. The users 500 and 501 are accessing the virtual network. When functioning, it needs to pass through this network node 100 first, and the computing nodes 201, 202, and 203 are the actual locations where virtual network functions are deployed. Web servers 401, 402, and 403 are the users in this embodiment. 500, 501 access service's final station point.

When users 500 and 501 access the web servers 401, 402, and 403, they will pass through the network node 100 and one or more of the computing nodes 201, 202, and 203. Each computing node 201, 202, and 203 can execute one or Multiple virtual network functions, after one or several virtual network functions are connected in series, will become a service function chain. As shown in Figure 2B, it is the service function chain 300 when the user 500 executes the network service, and as shown in Figure 2C, it is the service function chain 300 when the user 501 executes the network service. ', virtual network functions may include but are not limited to the following functions: Firewall function 3100, Load Balancer function 3200, Gateway function 3300, Network Address Translation (NAT) function 3400, etc. .

Take user 500 as an example. Its service function chain 300 includes forwarding function 3400 and load balancing function 3200. Network node 100 is the starting position of the decision-making method and executes the decision. First, find out the number of virtual network functions that can be deployed The most computing node, that is, after the computing node 202 is regarded as the selected computing node, it is then determined whether the computing node 202 has sufficient resources to deploy the virtual network function. If it is sufficient, the virtual network function is executed. In this embodiment , The virtual network function of the computing node 202 is the forwarding function 3400.

Next, determine whether there are other virtual network functions that need to be deployed. In this embodiment, there is a load balancing function 3200 to be deployed. Therefore, the computing node 202 is regarded as the starting point to find the next virtual network function that can be deployed with the largest number In this embodiment, the computing node 201 is the computing node, and then it is judged whether the computing node 201 has enough resources to deploy the virtual network function. Similarly, if it is sufficient, it will execute the subordinate of the virtual network function. In this embodiment, the computing node The virtual network function deployed by 201 is load balancing function 3200.

Finally, when it is determined that no other virtual network functions need to be deployed, the network connection can be connected to the web server 402 that provides network services, and this entire path is the service function chain 300.

Figure 3 is a flowchart of the actual application of the virtual network function deployment decision method of the service function chain of the present invention. Please refer to Figure 2 together. Assuming that several virtual network functions are expected to be deployed, in the processes 601 and 602, first Taking the network node 100 as the starting point, among the computing nodes 201, 202, and 203 connected to the starting point, find the computing node that can deploy the largest number of virtual network functions, that is, the computing node with the most available resources It is regarded as the selected operation node.

In the process 603, it means that if multiple computing nodes meet the conditions at the same time, then the process 604 is entered to further filter the computing node with the shortest path, that is, to find the weight of the edge from the starting point to the computing node plus the total value The smallest computing node, if the user 500 in FIG. 2 is taken as an example, the selected computing node is the computing node 202. In addition, if only one computing node meets the condition in the process 603, then the process 605 is entered.

In the process 605, if the resources of the computing node 202 found based on the above conditions are insufficient to deploy the virtual network function, the process 606 is entered, that is, the manager is reported to be insufficient in resources. Conversely, if the resources of the found computing node 202 are sufficient to deploy the virtual network function When the network function is used, the process 607 is entered, that is, the virtual network function is deployed on this computing node 202. Then enter the process 608, increase the weight of the edge passed by the computing node 202 from the starting point, which means that the virtual network function deployed this time passes through this path, and the number of connections of the path increases. The priority is then Decrease due to increased weight.

In the process 609, confirm whether there are any virtual network functions that need to be deployed. If so, enter the process 610, start with the computing node 202 found this time, and return to the process 602 to find the computing node with the most available resources. As an example, the user 500 in Figure 2 is a computing node 201. Then, repeat the above process until process 611, that is, all virtual network functions are deployed.

In summary, the current virtual network function technology is still new technology, many considerations are not yet mature, and there is much room for improvement when used in the cloud environment. The present invention proposes a virtual network function deployment decision method for the service function chain, which can solve Compared with the prior art, the virtual network function deployment location decision problem has the following advantages: (1) The present invention calculates the virtual network function deployment location, so that packets can reduce unnecessary processing or forwarding, thereby reducing cloud The overall CPU computing capacity in the environment relatively increases the available resources of the physical machine; (2) The present invention reduces the number of packets passing through the physical server by calculating the virtual network function deployment location, thereby allowing external users to access services At the same time, it can have lower network delay, reduce network congestion, and make the service have better quality.

The above-mentioned embodiments only exemplarily illustrate the principles and effects of the present invention, and are not intended to limit the present invention. Anyone who is familiar with this technique can modify and change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the rights of the present invention should be listed in the scope of patent application described later.

S11~S16‧‧‧Step

Claims (10)

A decision-making method for virtual network function deployment of a service function chain includes: Set a network node as the starting point; Set the initial value of the weight of the network connection between each node in the network to be the same; Search multiple computing nodes connected to the starting point, find the computing node with the largest number of virtual network functions that can be deployed, and define it as the selected computing node; Determine whether the resources of the selected computing node are sufficient to deploy the virtual network function, so as to determine the deployment of the virtual network function on the selected computing node when the selected computing node resources are abundant; Add a weight of the same value to the network connection from the starting point to the computing node where the virtual network function has been deployed; and Determine whether there are other virtual network functions that have not been deployed, so as to complete the deployment of all virtual network functions when there are no other virtual network functions that have not been deployed, or when there are other virtual network functions that have not yet been deployed, the Deploy the computing node of the virtual network function as a new starting point. Repeat the steps of searching, judging, and increasing the weight until all virtual network functions are deployed. For example, the virtual network function deployment decision method of the service function chain described in the scope of patent application, wherein, if there are multiple computing nodes with the most available resources at the same time, find out among the plurality of computing nodes with the most available resources For the operation node with the shortest path, the operation node with the shortest path is selected as the selected operation node. For example, the virtual network function deployment decision method of the service function chain described in the scope of the patent application, wherein the computing node with the shortest path refers to the network connection from the starting point to the selected computing node The one with the smallest total weight. For example, the virtual network function deployment decision method of the service function chain described in the scope of the patent application, wherein if there is only one computing node with the most available resources, it is selected as the selected computing node. For example, the virtual network function deployment decision method of the service function chain described in the scope of patent application, wherein if it is determined that the resources of the selected computing node are insufficient to deploy the virtual network function, the virtual network function is terminated The subordinate. For example, the virtual network function deployment decision method of the service function chain described in the scope of the patent application, wherein the network node is a physical server for performing deployment decisions of all virtual network functions. For example, the virtual network function deployment decision method of the service function chain described in the scope of the patent application, wherein the multiple computing nodes are all physical servers for the deployment of various virtual network functions. For example, the virtual network function deployment decision method of the service function chain described in item 1 of the scope of patent application, in which several virtual network functions are combined in series to form a service function chain. For example, the virtual network function deployment decision method of the service function chain described in item 1 of the scope of patent application, in which the network connection between each node is not directional. For example, the virtual network function deployment decision method of the service function chain described in the scope of patent application, wherein the virtual network function or the other virtual network function includes firewall function, load balancing function, gateway function or forwarding Features.

Images