CN109714223B - System and method for realizing network service access dynamic load sharing function under NFV architecture - Google Patents
System and method for realizing network service access dynamic load sharing function under NFV architecture Download PDFInfo
- Publication number
- CN109714223B CN109714223B CN201811079677.3A CN201811079677A CN109714223B CN 109714223 B CN109714223 B CN 109714223B CN 201811079677 A CN201811079677 A CN 201811079677A CN 109714223 B CN109714223 B CN 109714223B
- Authority
- CN
- China
- Prior art keywords
- node
- access
- load
- access node
- edge node
- 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.)
- Active
Links
Images
Abstract
The invention relates to a system and a method for realizing a network service access dynamic load sharing function under an NFV architecture, wherein the system comprises a plurality of access nodes and a plurality of edge nodes, the access nodes and the edge nodes are VNF nodes, each VNF node is connected with an internal network of the system, the edge nodes are connected with an external remote user module and periodically multicast the IP address of the edge nodes to the access nodes, the edge nodes calculate the load index of the edge nodes according to the received load state of the access nodes and select proper access nodes for the external remote user module to access; the access node is connected with each edge node, and is used for periodically unicast-sending the real-time load state of the access node to the edge node through a protocol message, and recording the IP address of the edge node in the current network. The adoption of the invention can avoid the realization of the load sharing of the user access under the NFV architecture through a special load balancer, and simultaneously can realize the dynamic expansion and contraction capacity of VNF resources.
Description
Technical Field
The invention relates to the field of NFV, in particular to the technical field of NFV architecture systems, and specifically relates to a system and a method for realizing a network service access dynamic load sharing function under an NFV architecture.
Background
In the NFV architecture system, a network administrator may create and delete a virtual container in a virtual resource pool to carry user access traffic, which may be referred to as a VNF. The common method is as follows: and a certain amount of VNFs are created in advance, and the VNFs are selected in sequence for online when a user accesses the VNFs. When the VNF reaches the upper limit of the user access capacity, the next VNF is selected again or a new VNF is created. When the user accesses a certain number, users are inevitably offline, and the users may be accessed on any VNF, some VNFs may run at full load, and some VNFs may be in an idle state. If a new user accesses subsequently, it is likely that the user accesses the VNF close to the full load according to the original access method. Therefore, the load of the entire NFV system is unbalanced, and a fully loaded VNF may be off-line for a user due to factors such as busy CPU, insufficient memory, and the like, or even a VNF container may be in a false state.
Disclosure of Invention
The present invention is directed to overcome the above drawbacks of the prior art, and provides a system and a method for implementing a network service access dynamic load sharing function under an NFV architecture, which enables a VNF system to operate in a relatively balanced and idle load state and solves the problem of unbalanced user access load in the NFV architecture.
In order to achieve the above object, the system and method for realizing the network service access dynamic load sharing function under the NFV architecture of the present invention are as follows:
the system comprises a plurality of access nodes and a plurality of edge nodes, wherein the access nodes and the edge nodes are VNF nodes, and each VNF node is connected with an internal network of the system, wherein:
the edge node is connected with an external remote user module and periodically multicasts the IP address of the edge node to an access node, calculates the load index of the access node according to the received load state of the access node and selects a proper access node for the external remote user module;
the access node is connected with each edge node, and is used for periodically unicast-sending the real-time load state of the access node to the edge node through a protocol message, and recording the IP address of the edge node in the current network.
Preferably, the system further comprises a MANO module for creating and reclaiming the VNF node module.
Preferably, the system internal network is a two-layer OVS network.
The method for realizing network service access dynamic load sharing control under the NFV architecture based on the system comprises the following steps:
(1-1) the edge node periodically multicast and sends own IP address information message to each VNF node;
(1-2) the access node receiving the IP address information message periodically unicast and sends a load state information message of the access node to the corresponding VNF node;
(1-3) the edge node which receives the load state information message dynamically calculates the load index of the access node according to the load state information of the access node;
(1-4) the edge node performs access node allocation operation for the external remote user module with online demand according to the load index information, and returns to the step (1-3).
Preferably, the step (1-1) specifically comprises the following steps:
(1-1.1) the edge node starts a first period timer and sets a first period, and periodically multicast and sends an IP address information message to each VNF node;
(1-1.2) the VNF node receives the IP address information message and judges whether the VNF node is an access node, if so, the step (1-1.3) is continued, otherwise, the message is directly discarded;
(1-1.3) the access node updates the received IP address information to a local edge node IP address management table, starts a first aging timer and sets a second period;
(1-1.4) the access node judges whether the IP address message of the edge node is received again in the second period, if so, the step (1-2) is continued, otherwise, the IP address of the edge node is deleted in the local IP address management table of the edge node.
Preferably, the first period is a default 10 seconds (configurable).
Preferably, the second period is 3 times the length of the first period.
Preferably, the step (1-2) specifically comprises the following steps:
(1-2.1) the access node starts a second period timer and sets a third period, and periodically unicast and sends a load state information message of the access node to each VNF node;
(1-2.2) the VNF node receives the load status information packet and determines whether it is an edge node, if so, continues with step (1-2.3), otherwise, directly discards the packet.
(1-2.3) the edge node searches whether the access node exists in the load management table or not according to the VNF ID, and if so, the access node in the load management table is updated; otherwise, establishing a new load management table access node element, starting a second aging timer and setting a fourth period;
(1-2.4) the edge node judges whether the load management table is updated again in the fourth period, if yes, the step (1-3) is continued, otherwise, the access node is deleted in the load management table, and then the step (1-1) is continued.
Preferably, the third period defaults to 10 seconds (configurable).
Preferably, the fourth period is 3 times the length of the third period.
Preferably, the load status information message includes a field of a current accessible number of the user, and the step (1-3) specifically includes the following steps:
(1-3.1) judging whether the value of the field of the current accessible number of the user in the load state information message field is 0, if so, arranging the access node at the tail part of the load management table, then continuing the step (1-1), otherwise, continuing the step (1-3.2)
(1-3.2) the edge node calculates the load index of the access node according to the load state information of the access node, and sorts the load management table according to the load index.
Preferably, the load index of the access node calculated according to the load status information of the access node is calculated by the following formula:
preferably, the edge node ranks the load management table with a low priority load index, and if the load indexes are the same, the priority ranking user can access the access node with a large number.
Preferably, the steps (1-4) specifically include the following steps:
(1-4.1) the edge node judging whether the external remote user module with the online requirement is a new online user module;
(1-4.2) if so, the edge node distributes the access node with the lowest load index to the user module, creates a corresponding table of the user module and the corresponding access node, and then continues to the step (1-3);
(1-4.3) if not, the edge node allocates a corresponding access node to the user module according to the existing correspondence table of the user module and the corresponding access node, and then continues to the step (1-3).
Preferably, the method further comprises an operation process of dynamic capacity expansion, specifically comprising the following steps:
(2-1) the edge node judges whether the load indexes of all the access nodes are not less than a preset high-level early warning threshold value of the system, if so, the MANO module newly establishes a VNF node, and then the step (1-1) is continued; otherwise, continuing the step (2-2);
(2-2) the edge node judges whether the load indexes of all the access nodes are not more than a low early warning threshold value preset by the system, if so, the MANO module recovers the access nodes with the user accessible number of 0, and then the step (1-1) is continued; otherwise, continuing (1-3.2).
Preferably, the method further comprises an operation process of user priority scheduling, comprising the following steps:
(3-1) the edge node judges whether the current NFV architecture resource is tense, if yes, the step (3-2) is continued, otherwise, the step (1-1) is continued;
(3-2) the edge node judging whether the user module which is newly on line belongs to a high-priority user;
(3-3) if yes, the edge node directly allocates available access nodes for the user module, and then continues to the step (1-1);
(3-4) if not, the edge node delays the online time for the user module, and then continues to the step (1-1).
Preferably, the step (3-4) further comprises the following steps:
(3-4.1) the edge node judges whether the load indexes of the access nodes which are not less than 1 are lower than a preset high early warning threshold value of the system, if so, the access nodes are distributed to the user module, and then the step (1-1) is continued; otherwise, recording the first online request time point t1 of the user module, discarding the online request message, and continuing the step (1-1);
(3-4.2) if the edge node receives the on-line request message of the user module again, recording the time point t2 of the on-line request of the user module again;
(3-4.3) the edge node judges whether the load indexes of the access nodes which are not less than 1 are lower than a preset high early warning threshold value of the system, if so, the access nodes are distributed to the user module, and then the step (1-1) is continued; otherwise, when the time length of the user module trying to get on line reaches the designated time length, allocating an access node for the user module, and then continuing the step (1-1).
Preferably, the time period for the user module to try to go online in the step (3-4.3) is t2-t 1.
Preferably, the fields in the load status information message include the current accessible number of the user and the total access capacity, and the method further includes an operation process of the access node actively quitting, specifically including the following steps:
(4-1) the edge node judges whether the current NFV architecture resource is tense, if yes, the step (4-2) is continued, otherwise, the step (1-1) is continued;
(4-2) the access node actively unicast transmits a load index information message;
(4-3) the edge node receives the load index information message sent by the access node, judges whether the values of the field of the current accessible number of the user and the field of the total access capacity are both 0, and deletes the access node in the load management table if the values are both 0; if not, continue with step (1-3).
Preferably, the determination of whether the current NFV architecture resource is in shortage is to determine whether the value of the field of the current accessible number of the user in the received load status information message of the access node is lower than a preset threshold value of the system.
If yes, returning a judgment result of the resource shortage of the NFV architecture; otherwise, returning the judgment result of the resource tension of the NFV architecture.
Preferably, the load status information message includes the following fields:
type, Length, VNF ID, current accessible number of users, total access capacity, memory usage rate and CPU usage rate.
The system and the method for realizing the network service access dynamic load sharing function under the NFV architecture can avoid realizing the load sharing of user access under the NFV architecture through a special load balancer and can realize the dynamic expansion and contraction of VNF resources at the same time. In the technical scheme, the VNF nodes directly pass through an internal two-layer network, the access nodes send the load states, and the edge nodes collect the load states of the VNFs, so that the networking requirements are simplified, the convergence speed is improved, the condition that the user access load in the NFV framework is unbalanced is solved, the phenomenon of false death of VNF containers is avoided, and the VNFs in the whole system are continuously operated in a relatively balanced and idle load state.
Drawings
Fig. 1 is a schematic structural diagram of a system for sharing network service access dynamic load under the NFV architecture of the present invention.
Fig. 2 is a flowchart of a method for sharing network service access dynamic load under the NFV architecture of the present invention.
Detailed Description
In order to more clearly describe the technical contents of the present invention, the following further description is given in conjunction with specific embodiments.
The system for realizing the network service access dynamic load sharing function under the NFV architecture comprises a plurality of access nodes and a plurality of edge nodes, wherein the access nodes and the edge nodes are VNF nodes, and each VNF node is connected with an internal network of the system, wherein:
the edge node is connected with an external remote user module and periodically multicasts the IP address of the edge node to the access node, calculates the load index of the access node according to the received load state of the access node and selects a corresponding access node for the external remote user module;
the access node is connected with each edge node, and is used for periodically sending the real-time load state of the access node to the edge node in a unicast mode through a protocol message, and recording the IP address of the edge node in the current network.
The system further comprises a MANO module for creating and reclaiming said VNF node modules.
The system internal network is a two-layer OVS network.
The method for realizing network service access dynamic load sharing control under the NFV architecture based on the system comprises the following steps:
(1-1) the edge node periodically multicast and sends own IP address information message to each VNF node;
(1-1.1) the edge node starts a first period timer and sets a first period, the first period is defaulted to 10 seconds (other time lengths can be configured according to needs), and periodically multicast and send an IP address information message to each VNF node;
(1-1.2) the VNF node receives the IP address information message and judges whether the VNF node is an access node, if so, the step (1-1.3) is continued, and if not, the message is directly discarded;
(1-1.3) the access node updates the received IP address information to a local edge node IP address management table, starts a first aging timer and sets a second period, wherein the second period is 3 times of the first period, and is default 30 seconds;
(1-1.4) the access node judges whether the IP address message of the edge node is received again in the second period, if so, the step (1-2) is continued, otherwise, the IP address of the edge node is deleted in a local IP address management table of the edge node. (ii) a
(1-2) the access node receiving the IP address information message periodically unicast and sends a load state information message of the access node to the corresponding VNF node;
(1-2.1) the access node starts a second period timer and sets a third period, the third period defaults to 10 seconds (the third period can be configured to other time lengths according to needs), and periodically unicast and sends a load state information message of the access node to each VNF node;
(1-2.2) the VNF node receives the load state information message and judges whether the VNF node is an edge node, if so, the step (1-2.3) is continued, and if not, the message is directly discarded;
(1-2.3) the edge node searching whether the access node exists in a load management table according to the VNF ID, and if so, updating the access node in the load management table; otherwise, establishing an access node element of the load management table, starting a second aging timer and setting a fourth period, wherein the fourth period is 3 times of the length of the third period, namely the default time is 30 seconds;
(1-2.4) the edge node judges whether the load management table is updated again in the fourth period, if so, the step (1-3) is continued, otherwise, the access node is deleted from the load management table, and then the step (1-1) is continued;
(1-3) the edge node which receives the load state information message dynamically calculates the load index of the access node according to the load state information of the access node;
(1-3.1) judging whether the value of the field of the current accessible number of the user in the field of the load state information message is 0, if so, arranging the access node at the tail part of the load management table, and then continuing the step (1-1); otherwise, continuing the step (1-3.2);
(1-3.2) the edge node calculates the load index of the access node according to the load state information of the access node, and sorts the load management table according to the load index;
(1-4) the edge node performs access node allocation operation for the external remote user module with online demand according to the load index information, and returns to the step (1-3);
(1-4.1) the edge node judging whether the external remote user module with the online requirement is a new online user module;
(1-4.2) if yes, the edge node distributes the access node with the lowest load index for the user module, creates a corresponding table of the user module and the corresponding access node, and then continues to the step (1-3);
(1-4.3) if not, the edge node allocates corresponding access nodes for the user modules according to the existing corresponding table of the user modules and the corresponding access nodes, and then the step (1-3) is continued. The load index of the access node calculated according to the load state information of the access node is calculated by the following formula:
when the edge node sorts the load management table, the access node with low priority load index is selected, if the load indexes are the same, the access node with large number can be accessed by the priority sorting user.
In a preferred embodiment of the present invention, the method further includes an operation process of dynamically scaling the capacity, specifically including the following steps:
(2-1) the edge node judges whether the load indexes of all the access nodes are not less than a preset high-level early warning threshold value of the system, if so, the MANO module newly establishes a VNF node, and then the step (1-1) is continued; otherwise, continuing the step (2-2);
(2-2) the edge node judges whether the load indexes of all the access nodes are not more than a low early warning threshold value preset by the system, if so, the MANO module recovers the access nodes with the user accessible number of 0, and then the step (1-1) is continued; otherwise, continuing (1-3.2).
In a preferred embodiment of the present invention, the method further includes an operation process of user priority scheduling, including the following steps:
(3-1) the edge node judges whether the current NFV architecture resource is tense, if yes, the step (3-2) is continued, otherwise, the step (1-1) is continued;
(3-2) the edge node judges whether the user module which is newly on line belongs to a high-priority user;
(3-3) if yes, the edge node directly allocates available access nodes for the user module, and then the step (1-1) is continued;
(3-4) if not, delaying the online time for the user module by the edge node, and then continuing the step (1-1).
(3-4.1) the edge node judges whether the load index of at least 1 access node is lower than a preset high early warning threshold value of the system, if so, the edge node allocates the access node for the user module, and then the step (1-1) is continued; otherwise, recording the first online request time point t1 of the user module, discarding the online request message, and continuing the step (1-1);
(3-4.2) if the edge node receives the on-line request message of the user module again, recording a time point t2 of the on-line request of the user module again;
(3-4.3) the edge node judges whether the load index of at least 1 access node is lower than a preset high early warning threshold value of the system, if so, the edge node allocates the access node for the user module, and then the step (1-1) is continued; otherwise, when the time length of the user module trying to get on line reaches the designated time length (t 2-t 1), allocating an access node for the user module, and then continuing the step (1-1).
In a preferred embodiment of the present invention, the fields in the load status information message include the current accessible number of the user and the total access capacity, and the method further includes an operation process of the access node actively quitting, specifically including the following steps:
(4-1) the edge node judges whether the current NFV architecture resource is tense, if yes, the step (4-2) is continued, otherwise, the step (1-1) is continued;
(4-2) the access node actively unicast transmits a load state information message;
(4-3) the edge node receives the load state information message sent by the access node, judges whether the values of the field of the current accessible number of the user and the field of the total access capacity are both 0, and deletes the access node in the load management table if the values are both 0; if not, continue with step (1-3).
Wherein, the above-mentioned judging whether the current NFV architecture resource is in shortage is:
judging whether the value of the user current accessible digital field of the received load state information message of the access node is lower than a preset critical value of the system, if so, returning a judgment result of the resource shortage of the NFV architecture; otherwise, returning the judgment result of the resource tension of the NFV architecture.
The load status information message includes the following fields: type, Length, VNF ID, current accessible number of users, total access capacity, memory usage rate and CPU usage rate.
In actual use, as shown in fig. 1, a remote user module is connected to the NFV architecture through an infrastructure. The edge node is mainly responsible for processing load sharing scheduling, and the access node is mainly responsible for processing user access. When all VNF nodes in the system are created, virtual communication network cards are created and added into an internal two-layer OVS network, and two-layer intercommunication is achieved. They can realize the receiving and sending of protocol message through the virtual interface.
1. Load index definition
The load states of all access nodes mainly include: the current accessible number of users, the total access capacity of the users, the memory utilization rate and the CPU utilization rate. Each index has a priority, the load index of each access node can be finally calculated, and the priority can be configured through the MANO module. The load index of the access node calculated according to the load state information of the access node is calculated by the following formula:
wherein, the access node multiplies the three values by 100 and then sends the three values to the edge node. The higher the metric value, the greater the load on the access node. Finally, the edge node can select a specific access node to access a new user by means of this indicator.
Because the protocol messages are all transmitted and received in the internal two-layer OVS network, the messages can be defined simply according to TLV format, and comprise the following fields: type, Length, VNF ID, current accessible number of users, total access capacity, memory usage rate and CPU usage rate. The size of the current accessible number field of the user is 4B, the size of the total access capacity field is 4B, the size of the memory utilization rate field is 1B, the size of the CPU utilization rate field is 1B, the size of the T field is 1B, the size of the L field is 1B, and the size of the VNF ID field is 4B.
2. Node address
After each edge node is added into the OVS network, the IP address information of the edge node needs to be periodically multicast, so that the access node can conveniently acquire the IP address of the edge node and can send the information for calculating the load index by using unicast. Each access node needs to record the IP addresses of all edge nodes in the current network.
The message includes the following fields: t, L and VNF ID, IP address. The size of the T field is 1B, the size of the L field is 1B, the size of the VNF ID field is 4B, and the size of the IP address field is 4B.
3. Load sharing method
When the MANO module creates a VNF node, a unique identification ID (VNF ID) is distributed to the node, an IP address of the same network segment is distributed to the virtual network card, and the IP address is added into an internal two-layer OVS network.
All nodes default roles are access nodes. However, if an edge node needs to be selected for remote users, the MANO module may set the configuration when deploying the NFV system, for example: VNF1 and VNF 2.
And the edge node newly joining the network starts a first period timer and sets a first period, the default is 10 seconds, and the address information message is multicast and sent.
The edge node receives the message and directly discards the message. The access node receives the message and updates the address information to the edge node address management list.
The access node starts a first aging timer and sets a second period, the second period is 3 times of the first period timer of the edge node, the default is 30 seconds, and if the IP address message of the edge node is not received again within the overtime of the first aging timer, the address is aged and deleted from the address management list.
The access node starts a second period timer and sets a third period, the default is 10 seconds, the multicast message is sent to the whole two-layer network, and the edge node is informed of the node load state.
The access node receives the message and directly discards the message. The edge node receives the message, searches a node load management table according to the VNF ID, and if the node load management table is not found, creates a new table entry; and if the table entry is found, updating the table entry.
And when the edge node receives the load state message of the access node, a load management table is established. Manage a table with VNF ID as key and manage a table again according to node load metrics:
if the current accessible number of the user is 0, the node is directly placed at the tail part of the list to indicate that the node has no access capability.
If the current accessible number of the user is greater than 0, calculating the load index of the node, and sequencing by using the index: the load index is lower than that of the list item, and the list item is preferentially used. If the indexes are the same, the current accessible number is large, and the access is preferentially used.
And the edge node starts a second aging timer and sets a fourth period, wherein the fourth period is 3 times of the interval of the second timer of the access node and is defaulted to be 30 seconds, and if the load management table of the access node is not updated within the overtime of the aging timer, the access node is aged and deleted. The node may have been down or overloaded, and the edge node cannot select this access node.
When the edge node receives the on-line request of the remote user module, the edge node selects the node with the lowest load index from the load management table of the access node, and then forwards the on-line request of the user to the access node for on-line processing of the user. Meanwhile, a corresponding relation table of user MAC-access node VNF ID is required to be recorded, so that subsequent protocol messages of the user can be conveniently forwarded to the access node. Before the access node advertises a new load index again, the edge node needs to calculate the latest load index according to the new online number of the user and the load index information advertised before, so as to ensure that the load index does not exceed the access node with suboptimal load index.
When the edge node receives the load index message of the access node again, the corresponding data can be updated again and the data can be reordered.
4. Dynamic expansion and contraction capacity
When the edge node finds that the load indexes of all the access nodes reach a certain high-order threshold value, the edge node informs a MANO module to newly establish a VNF node to enlarge the user access capacity.
When the edge node finds that the load indexes of all the access nodes are lower than a certain low threshold, the access node with the least number of current users can be inserted into a VNF list to be recovered, namely, the new user is not preferentially selected because of low load indexes when the new user is online, and after all users on the access node are offline, the MANO module is informed to recover the node.
5. Priority scheduling for access users
When resources of the whole NFV architecture system are in shortage, the edge node can schedule the online according to the priority of the user after receiving the online request of the user.
High priority users: and according to the previous strategy, timely responding to the user on-line event and distributing the available access nodes for the user.
Low priority users: the online time can be properly prolonged.
Looking at the current NFV system load index threshold: if the access node is lower than the high threshold, immediately allocating an access node for the user to be on-line; if the value is still higher than the high threshold, recording the online request time point t1 of the user, and discarding the online request message.
When the user requests online next time t2, continuously checking the current NFV system load index threshold: if the access node is lower than the high threshold, immediately allocating an access node for the user to be on-line; if the time length of the user trying to go on line is still higher than the high threshold, the time length of the user trying to go on line reaches the specified time length, and as a preferred embodiment of the invention, t2-t1> configurable waiting time length, the user is not allocated to access node going on line.
6. Access node active exit
In addition to this, the designated access node VNF may be manually deleted when the entire NFV architecture is resource-tight. When the VNF of the access node senses the event, it actively sends a load state information packet to the edge node: the current accessible number is 0, and the total access capacity is 0. Then, after the edge node receives the message, it can determine that the access node actively exits the load sharing scheduling, and thus removes it from the load management table. And if a new user is on line subsequently, the node cannot be selected again. If the message of the access node cannot be received by the edge node due to the abnormal conditions of the internal network or the node itself, and the like, the edge node finally ages and deletes the node through the aging timer.
The system and the method for realizing the network service access dynamic load sharing function under the NFV architecture can avoid realizing the load sharing of user access under the NFV architecture through a special load balancer and can realize the dynamic expansion and contraction of VNF resources at the same time. In the technical scheme, the VNF nodes directly pass through an internal two-layer network, the access nodes unicast the load states, and the edge nodes collect the load states of all VNFs, so that the networking requirements are simplified, the convergence speed is improved, the condition that the user access load in the NFV framework is unbalanced is solved, the false death phenomenon of VNF containers is avoided, and the VNFs in the whole system continuously run in a relatively balanced and idle load state.
In this specification, the invention has been described with reference to specific embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.
Claims (20)
1. A system for realizing a network service access dynamic load sharing function under an NFV architecture is characterized in that the system comprises a plurality of access nodes and a plurality of edge nodes, wherein the access nodes and the edge nodes are VNF nodes, and each VNF node is connected with an internal network of the system, and the system comprises:
the edge node is connected with an external remote user module and periodically multicasts the IP address of the edge node to the access node, calculates the load index of the access node according to the received load state of the access node and selects a proper access node for the external remote user module to access;
the access node is connected with each edge node and is used for periodically sending the real-time load state of the access node to the edge node in a unicast mode through a protocol message and recording the IP address of the edge node in the current network;
the system for realizing the network service access dynamic load sharing function under the NFV architecture carries out the following processing procedures:
(1-1) the edge node periodically multicast and sends own IP address information message to each VNF node;
(1-2) the access node receiving the IP address information message periodically unicast and sends a load state information message of the access node to the corresponding VNF node;
(1-3) the edge node which receives the load state information message dynamically calculates the load index of the access node according to the load state information of the access node;
(1-4) the edge node performs access node allocation operation for the external remote user module with online demand according to the load index information, and returns to the step (1-3);
the step (1-1) specifically comprises the following steps;
(1-1.1) the edge node starts a first period timer and sets a first period, and periodically multicast and sends an IP address information message to each VNF node;
(1-1.2) the VNF node receives the IP address information message and judges whether the VNF node is an access node, if so, the step (1-1.3) is continued, and if not, the message is directly discarded;
(1-1.3) the access node updates the received IP address information to a local edge node IP address management table, starts a first aging timer and sets a second period;
(1-1.4) the access node judges whether the IP address message of the edge node is received again in the second period, if so, the step (1-2) is continued, otherwise, the IP address of the edge node is deleted in a local IP address management table of the edge node.
2. The system for implementing network traffic access dynamic load sharing under NFV architecture as claimed in claim 1, wherein said system further comprises a MANO module, said MANO module is used for creating and recovering said VNF node module.
3. The system according to claim 1, wherein the system internal network is a two-layer OVS network.
4. A method for realizing network service access dynamic load sharing control under NFV architecture is disclosed, the method is based on a system for realizing network service access dynamic load sharing function under NFV architecture, the system includes a plurality of access nodes and a plurality of edge nodes, both the access nodes and the edge nodes are VNF nodes, each VNF node is connected with the internal network of the system, wherein:
the edge node is connected with an external remote user module and periodically multicasts the IP address of the edge node to the access node, calculates the load index of the access node according to the received load state of the access node and selects a proper access node for the external remote user module to access;
the access node is connected with each edge node and is used for periodically sending the real-time load state of the access node to the edge node in a unicast mode through a protocol message and recording the IP address of the edge node in the current network;
the method is characterized by comprising the following steps:
(1-1) the edge node periodically multicast and sends own IP address information message to each VNF node;
(1-2) the access node receiving the IP address information message periodically unicast and sends a load state information message of the access node to the corresponding VNF node;
(1-3) the edge node which receives the load state information message dynamically calculates the load index of the access node according to the load state information of the access node;
(1-4) the edge node performs access node allocation operation for the external remote user module with online demand according to the load index information, and returns to the step (1-3);
the step (1-1) specifically comprises the following steps;
(1-1.1) the edge node starts a first period timer and sets a first period, and periodically multicast and sends an IP address information message to each VNF node;
(1-1.2) the VNF node receives the IP address information message and judges whether the VNF node is an access node, if so, the step (1-1.3) is continued, and if not, the message is directly discarded;
(1-1.3) the access node updates the received IP address information to a local edge node IP address management table, starts a first aging timer and sets a second period;
(1-1.4) the access node judges whether the IP address message of the edge node is received again in the second period, if so, the step (1-2) is continued, otherwise, the IP address of the edge node is deleted in a local IP address management table of the edge node.
5. The method of claim 4, wherein the first period is 10 seconds.
6. The method of claim 4, wherein the second period is 3 times the length of the first period.
7. The method according to claim 4, wherein the step (1-2) specifically includes the following steps:
(1-2.1) the access node starts a second period timer and sets a third period, and periodically unicast and sends a load state information message of the access node to each VNF node;
(1-2.2) the VNF node receives the load state information message and judges whether the VNF node is an edge node, if so, the step (1-2.3) is continued, and if not, the message is directly discarded;
(1-2.3) the edge node searches whether the access node exists in a load management table according to the VNF ID, and if so, the access node in the load management table is updated; otherwise, establishing a new load management table access node, starting a second aging timer and setting a fourth period;
(1-2.4) the edge node judges whether the load management table is updated again in the fourth period, if yes, the step (1-3) is continued, otherwise, the access node is deleted in the load management table, and then the step (1-1) is continued.
8. The method according to claim 7, wherein the third period is 10 seconds.
9. The method of claim 7, wherein the fourth period is 3 times as long as the third period.
10. The method according to claim 4, wherein the load status information packet includes a field of a current number of users that can be accessed, and the step (1-3) specifically includes the following steps:
(1-3.1) judging whether the value of the field of the current accessible number of the user is 0, if so, arranging the access node at the tail of the load management table, and then continuing the step (1-1); otherwise, continuing the step (1-3.2);
(1-3.2) the edge node calculates the load index of the access node according to the load state information of the access node, and sorts the load management table according to the load index.
11. The method according to claim 4, wherein the load index of the access node calculated according to the load status information of the access node is calculated by the following formula:
12. the method according to claim 10, wherein the edge node ranks the load management table according to a priority load index of the access node with a low priority, and if the load indexes are the same, ranks the access nodes with a large number of users with a high priority.
13. The method according to claim 4, wherein the step (1-4) specifically includes the following steps:
(1-4.1) the edge node judging whether the external remote user module with the online requirement is a new online user module;
(1-4.2) if yes, the edge node distributes the access node with the lowest load index for the user module, creates a corresponding table of the user module and the corresponding access node, and then continues to the step (1-3);
(1-4.3) if not, the edge node allocates corresponding access nodes for the user modules according to the existing corresponding table of the user modules and the corresponding access nodes, and then the step (1-3) is continued.
14. The method according to claim 4, wherein the system further includes a MANO module, and the method further includes an operation process of dynamic scaling, and specifically includes the following steps:
(2-1) the edge node judges whether the load indexes of all the access nodes are not less than a preset high-level early warning threshold value of the system, if so, the MANO module newly establishes a VNF node, and then the step (1-1) is continued; otherwise, continuing the step (2-2);
(2-2) the edge node judges whether the load indexes of all the access nodes are not more than a low early warning threshold value preset by the system, if so, the MANO module recovers the access nodes with the user accessible number of 0, and then the step (1-1) is continued; otherwise, continuing (1-3.2).
15. The method according to claim 4, wherein the method further includes an operation process of user priority scheduling, and the method includes the following steps:
(3-1) the edge node judges whether the current NFV architecture resource is tense, if yes, the step (3-2) is continued, otherwise, the step (1-1) is continued;
(3-2) the edge node judges whether the user module which is newly on line belongs to a high-priority user;
(3-3) if yes, the edge node directly allocates available access nodes for the user module, and then the step (1-1) is continued;
(3-4) if not, delaying the online time for the user module by the edge node, and then continuing the step (1-1).
16. The method according to claim 15, wherein the step (3-4) further includes the following steps:
(3-4.1) the edge node judges whether the load index of at least 1 access node is lower than a preset high early warning threshold value of the system, if so, the edge node allocates the access node for the user module, and then the step (1-1) is continued; otherwise, recording the first online request time point t1 of the user module, discarding the online request message, and continuing the step (1-1);
(3-4.2) if the edge node receives the on-line request message of the user module again, recording a time point t2 of the on-line request of the user module again;
(3-4.3) the edge node judges whether the load index of at least 1 access node is lower than a preset high early warning threshold value of the system, if so, the edge node allocates the access node for the user module, and then the step (1-1) is continued; otherwise, when the time length of the user module trying to get on line reaches the designated time length, distributing the access node for the user module, and then continuing the step (1-1).
17. The method as claimed in claim 16, wherein the time duration for the user module to attempt to go online in step (3-4.3) is t2-t 1.
18. The method according to claim 4, wherein the fields in the load status information message include a current accessible number of users and a total access capacity, and the method further includes an operation process of an access node actively exiting, specifically including the following steps:
(4-1) the edge node judges whether the current NFV architecture resource is tense, if yes, the step (4-2) is continued, otherwise, the step (1-1) is continued;
(4-2) the access node actively unicast transmits a load state information message;
(4-3) the edge node receives the load state information message sent by the access node, judges whether the values of the field of the current accessible number of the user and the field of the total access capacity are both 0, and deletes the access node in the load management table if the values are both 0; if not, continue with step (1-3).
19. The method according to any one of claims 15 to 18, wherein the determining whether NFV architecture resources are in shortage includes the following steps:
(i) judging whether the value of the field of the current accessible number of the user in the received load state information message of the access node is lower than a preset critical value of the system;
(ii) if yes, returning a judgment result of the resource shortage of the NFV architecture; otherwise, returning the judgment result of the resource tension of the NFV architecture.
20. The method according to any of claims 4 to 18, wherein the load status information packet includes the following fields:
type, Length, VNF ID, current accessible number of users, total access capacity, memory usage rate and CPU usage rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811079677.3A CN109714223B (en) | 2018-09-17 | 2018-09-17 | System and method for realizing network service access dynamic load sharing function under NFV architecture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811079677.3A CN109714223B (en) | 2018-09-17 | 2018-09-17 | System and method for realizing network service access dynamic load sharing function under NFV architecture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109714223A CN109714223A (en) | 2019-05-03 |
| CN109714223B true CN109714223B (en) | 2022-03-22 |
Family
ID=66253863
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811079677.3A Active CN109714223B (en) | 2018-09-17 | 2018-09-17 | System and method for realizing network service access dynamic load sharing function under NFV architecture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109714223B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111629046B (en) * | 2020-05-22 | 2022-04-08 | 中国联合网络通信集团有限公司 | Edge computing cooperation method, edge computing equipment and terminal |
| CN111770302A (en) * | 2020-07-16 | 2020-10-13 | 安徽华速达电子科技有限公司 | Video plug flow load balancing method and system based on optical network unit equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105635199A (en) * | 2014-10-28 | 2016-06-01 | 广州睿哲网络科技有限公司 | Method and device for implementation of self-organization cluster server supporting load balancing |
| CN106412999A (en) * | 2016-11-14 | 2017-02-15 | 上海斐讯数据通信技术有限公司 | Load balancing method and load balancing system |
| CN106797405A (en) * | 2016-12-14 | 2017-05-31 | 华为技术有限公司 | Distributed load equalizing system, health examination method and service node |
| CN107666419A (en) * | 2016-07-28 | 2018-02-06 | 中兴通讯股份有限公司 | A kind of virtual broadband cut-in method, controller and system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10140146B2 (en) * | 2015-10-22 | 2018-11-27 | Genband Us Llc | Network management for elastic virtual network functions |
| US11159609B2 (en) * | 2020-03-27 | 2021-10-26 | Intel Corporation | Method, system and product to implement deterministic on-boarding and scheduling of virtualized workloads for edge computing |
-
2018
- 2018-09-17 CN CN201811079677.3A patent/CN109714223B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105635199A (en) * | 2014-10-28 | 2016-06-01 | 广州睿哲网络科技有限公司 | Method and device for implementation of self-organization cluster server supporting load balancing |
| CN107666419A (en) * | 2016-07-28 | 2018-02-06 | 中兴通讯股份有限公司 | A kind of virtual broadband cut-in method, controller and system |
| CN106412999A (en) * | 2016-11-14 | 2017-02-15 | 上海斐讯数据通信技术有限公司 | Load balancing method and load balancing system |
| CN106797405A (en) * | 2016-12-14 | 2017-05-31 | 华为技术有限公司 | Distributed load equalizing system, health examination method and service node |
Non-Patent Citations (1)
| Title |
|---|
| 《智能管道体系下新一代内容分发网络_CDN3.0_技术概述》;海锦霞等;《通信热点》;20140131;14-18 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109714223A (en) | 2019-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9591641B2 (en) | Method and device for scheduling slot resources | |
| KR101887581B1 (en) | Flow-based packet transport device and packet management method thereof | |
| KR100615351B1 (en) | Server, mobile communication system, positional information managing method, radio base station, mobile station, method for calling in mobile communication system, and mobile communication method | |
| CN103053146B (en) | Data migration method and device | |
| JPH09121217A (en) | Method for burst transfer | |
| CN110784779B (en) | Data acquisition method of electricity consumption information acquisition system | |
| CN102264116A (en) | Node network access method based on distributed time division multiple access wireless ad hoc network | |
| CN109714223B (en) | System and method for realizing network service access dynamic load sharing function under NFV architecture | |
| CN109962760A (en) | A kind of business scheduling method suitable for wireless TDMA ad hoc network | |
| CN102223306A (en) | Method for transmitting massages and device | |
| CN104469855A (en) | Method and device for selecting time slot resources | |
| CN113923125A (en) | Tolerance analysis method and device for multi-service flow converged communication in industrial heterogeneous network | |
| CN104967571A (en) | Bandwidth adjustment method and device | |
| CN100466593C (en) | Method of implementing integrated queue scheduling for supporting multi service | |
| CN108134986B (en) | Message transmission method and device | |
| CN101341692B (en) | Admission control using backup link based on access network in Ethernet | |
| CN103763206A (en) | Network scheduling method and gateway | |
| CN109120549B (en) | Priority-driven switching optimization method under wireless SDN | |
| CN101754389A (en) | Resource control method, device and system | |
| CN112838989A (en) | Data stream management method, network equipment and storage medium | |
| US20220279036A1 (en) | Method, apparatus and system for processing an access request in a content delivery system | |
| CN106851685B (en) | Method and system for controlling bandwidth of mobile terminal | |
| CN101695049A (en) | Method and device for processing businesses in monitoring system | |
| CN103546872A (en) | Paging message sending method in trunking communication system | |
| CN113992609B (en) | Method and system for processing multilink service data disorder |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |