CN114257594A

CN114257594A - Method for distributing network resources to user network side in distributed system

Info

Publication number: CN114257594A
Application number: CN202111573186.6A
Authority: CN
Inventors: 肖霞; 梁晖; 张阳; 曾培; 李灵
Original assignee: Sichuan Netop Telecom Co ltd
Current assignee: Sichuan Netop Telecom Co ltd
Priority date: 2021-12-21
Filing date: 2021-12-21
Publication date: 2022-03-29
Anticipated expiration: 2041-12-21
Also published as: CN114257594B

Abstract

The invention discloses a method for distributing network resources to a user network side in a distributed system, which comprises the following steps: step one, in a plurality of unit rings at a user network side, a node of each unit ring obtains network topology information matched with the node through a topology obtaining mode; step two, in any unit ring, each node elects and determines a main node, a standby node and a common node of each unit ring in a priority mode based on whether the node can reach the central network; and step three, each unit ring reports the network topology information matched with the unit ring to the central network through the main node, and allocates the network resources of the standby node and the common node based on the network resources received from the central network. The invention provides a method for allocating network resources to a user network side in a distributed system, which realizes the automatic configuration of user network IP services.

Description

Method for distributing network resources to user network side in distributed system

Technical Field

The invention relates to the field of network communication. More particularly, the present invention relates to a method for allocating network resources to a user network side in a distributed system.

Background

The distributed system network (as shown in fig. 4) is composed of a central network side and a user network side connected with the central network side, the user network side is composed of a plurality of sub-user networks, each sub-user network (also called a unit ring) comprises a plurality of nodes (each node is a switch), at most 2 nodes in one unit ring are connected with a resource planning system of the central network, and 2 nodes are mutually backups. And a unit ring is formed inside each user network, and each user network meets the communication requirements of the main center and the branch centers. As shown in fig. 1, networking of the entire system is complex, there are many nodes in the network, and resources (such as IP addresses, etc.) need to be allocated to each node, but the network topology acquisition mode and the resource allocation mode of the existing unit ring are not high in automation degree, and there is no way to achieve plug and play, which is not beneficial to automatically managing network resource information, and specifically, the following disadvantages are included:

1. in the process of allocating IP addresses, confusion is easy to generate, and the situation that two or more nodes set the same IP can be caused;

2. it is difficult to troubleshoot when IP conflicts occur. After the manual IP address allocation is adopted, for the IP conflict generated in the network, the conflict source can not be determined, so that the conflict is inconvenient to be checked;

3. when a new node accesses a network, the next IP address to be allocated and gateway information need to be known at the same time, and the set IP does not conflict with other nodes;

4. the existing IP automatic allocation method cannot sense the access and the leaving of the nodes, so that the network topology is automatically calculated.

Disclosure of Invention

An object of the present invention is to solve at least the above problems and/or disadvantages and to provide at least the advantages described hereinafter.

To achieve these objects and other advantages in accordance with the present invention, there is provided a method for network resource allocation to a user network side in a distributed system, comprising:

step one, in a plurality of unit rings at a user network side, a node of each unit ring obtains network topology information matched with the node through a topology obtaining mode;

step two, in any unit ring, each node elects and determines a main node, a standby node and a common node of each unit ring in a priority mode based on whether the node can reach the central network;

and step three, each unit ring reports the network topology information matched with the unit ring to the central network through the main node, and allocates the network resources of the standby node and the common node based on the network resources received from the central network.

Preferably, the method further comprises the following steps:

step four, after finding the main node is offline, the nodes of each unit ring immediately initiate re-election of the main node, and report the change condition of the network topology information to the central network through a new main node after election;

the central network decides whether to send new unit ring network resources to the new host node according to whether the current resource conflict situation exists.

Preferably, in the first step, the topology obtaining means is configured to include:

s10, each node in the unit ring periodically sends monitoring frames to other nodes, or actively sends monitoring frames for 1 time when the physical connection state of the left side and the right side of the node changes;

s11, each node is provided with a corresponding MAC address list which is established based on the source MAC address in the received monitoring frame;

s12, each node pair updates the MAC address list through a judgment algorithm of overtime aging;

and S13, each node describes the position of the node in the cell ring by describing the hop count of other nodes from the counterclockwise direction and the clockwise direction to the node, thereby obtaining the network topology information of each node in the cell ring.

Preferably, in the second step, the determining manner of whether each node can reach the central network is configured to include:

s20, each node sends corresponding Hello message to the central network;

and S21, each node judges whether the node can reach the central network based on whether the node can receive the feedback message of the central network.

Preferably, in step two, the priority election means is configured to include:

s22, in the cell ring, if only one node can reach the central network, directly electing the node as the main node, otherwise, entering S23;

s23, in the unit ring, all nodes which can reach the central network are elected as candidate nodes, and other nodes which cannot reach are elected as common nodes;

and S24, judging among all the candidate nodes based on the priorities of the nodes, selecting the node with the higher priority as the master node, and comparing MAC addresses of the candidate nodes if the priorities of the candidate nodes are the same, and selecting the node with the larger MAC address as the master node.

Preferably, in the step two, under the condition that a new node is added after the master node is elected, whether the back-end network of the master node is still reachable is judged, if so, the master node is not changed, otherwise, the master node electing operation is carried out again.

Preferably, in the third step, the master node periodically sends a Hello packet to the central network, and if the central network does not receive the Hello packet sent by the master node in a period, or the master node does not receive a response from the central network in a period, the master node is considered to be out of contact with the central network.

Preferably, in step three, after receiving the resource allocation request of each child node, the master node performs resource allocation to the standby node and the normal node based on a resource management algorithm running on the master node.

Preferably, the resource management algorithm includes:

s30, respectively calculating the distance from the master node to the backup node from the left side and the right side by taking the master node as the center, recording the hop count if the distance is up, and setting the hop count to 0 if the distance is not up;

s31, calculating right and left demarcation points, if the hop count is 0, determining that the direction is not reachable, if the hop count is 1, determining that the neighbor is a standby node, otherwise, setting the hop count to-1/2;

s32, the main node checks and allocates resources in a way of right side first and left side second, in the detection process, if the hop count from the main node to the node is 0, the direction check and allocation is stopped, if the hop count from the main node to the node is 1, whether the node allocates the spare node resource is further judged, if not, the spare node resource is allocated; if the hop number to the node is more than 1, the load in the direction is determined to be shared to the main node, the rest nodes are the nodes with the load shared to the standby node, and meanwhile, the network IP resources are distributed.

Preferably, in S32, the master node establishes a record linked list for network resource allocation, where the record linked list uses the MAC address of each node as a key value, divides the total address received by the central network into a plurality of sub-address pools corresponding to the nodes, and sets each sub-address pool to be in an unallocated state;

when the main node distributes network IP resources to other nodes, the main node searches and judges whether each node has been distributed by the resources through key values in the distribution history record, if so, the main node directly uses the configuration of the node, sets the corresponding sub-address pool as distributed and processes the next node.

The invention at least comprises the following beneficial effects: in view of the problems and automatic configuration defects of manual configuration of the network IP address of the existing distributed system, the invention calculates the topology of the user network through a custom algorithm, namely a topology tree taking a main node as a center, can sense the access and the departure of a new node, reports a central network request resource based on the topology, and then distributes the resource to other nodes by the main node, thereby realizing the automatic configuration of the IP service, further ensuring that each node in the user network does not need to be manually configured in the application process, realizing the automatic configuration of the service in the system and supporting the plug and play of equipment.

Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention.

Drawings

Fig. 1 is a schematic diagram of transmitting a monitoring frame through a node 2 by taking a node 1 as an example in a unit ring according to the present invention;

FIG. 2 is a schematic diagram of the workflow of the topology acquisition state machine of the present invention;

FIG. 3 is a schematic diagram of topology information of one of the nodes in the cell ring;

fig. 4 is a system composition block diagram of a distributed system network.

Detailed Description

The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.

It will be understood that terms such as "having," "including," and "comprising," as used herein, do not preclude the presence or addition of one or more other elements or groups thereof.

In view of the problems of manual configuration of the network IP address of the distributed system and the defects of automatic configuration, the invention realizes automatic configuration of the IP service of the user network, realizes automatic configuration of the service in the system without manual configuration of each node in the user network, supports plug and play of equipment, and is mainly realized by automatically calculating network topology, sensing node access and departure and resource management:

1. topology computation

Sensing the node resource connection condition to calculate the network topology, electing a master/standby node according to the network state, reporting the network topology and the resources by the master node and a resource planning system, receiving the network resources distributed by the resource planning system, and distributing the distributed network resources to the standby node and the common node by the master node.

2. Node access and departure

Each node finds that the master node is offline and immediately initiates master node re-election, after the master node is elected, a new master node reports topology change, and the resource planning system determines whether to re-issue the network resources of the cell ring according to whether the resource conflict condition exists or not.

3. Resource management algorithm

The resource management algorithm is run by the main node, and is distributed according to the algorithm after receiving the requests of each sub-node.

Specifically, the present invention mainly comprises:

1. topology acquisition

The nodes in the unit ring periodically (2S) send monitoring frames to other nodes, or actively send monitoring frames for 1 time when the physical connection state of the left side and the right side of the node changes, so that the following information can be acquired from the monitoring frames:

1) a node MAC address;

2) arriving over several hops from the left or right side of the node;

3) counting the error and correct frames received by the node from the left side and the right side;

4) physical connection states of the left side and the right side of the node.

The node maintains a MAC address table that is established based on the source MAC address in the received supervision frame. As shown in the above figure, all nodes in the intranet are connected into a ring, and as shown in the node 1 in fig. 1 as an example, the node 1 may receive 2 packets of monitoring frames sent by the node 2 (no deduplication operation is performed on the type of packet through 5 hops from the left side and 1 hop from the right side). Node 1 parses the source MAC address (i.e., node B MAC address) from the supervision frame and adds it as a new entry in the local MAC address list. The invention designs a topology acquisition state machine as shown in fig. 2, and for 2 received monitoring frames, overtime aging is needed, the aging time is 4S, and aging is started after the monitoring frames are received, namely, the monitoring frames are not acquired again in 4S, and the address is deleted from an MAC address list. If the monitoring frame sent by the same node with smaller local distance is received before the aging time, the monitoring frame (sent from the other side) needs to be replaced and the aging is restarted, and if the monitoring frame is longer than the existing distance, the monitoring frame does not need to be replaced. And after the aging time is up, replacing the node with the distance value if the distance value in the other direction is not aged, and deleting the node if the distance value in the other direction does not exist, which indicates that the node is lost.

The node 1 describes the position of the node in the ring network by describing the hop counts of other nodes from the counterclockwise direction and the clockwise direction to the node 1, so that the topological connection relationship of the node is known. The node information in the lower graph describes the position information of all nodes within the unit ring with respect to node 1. Finally, all nodes can calculate a piece of topology information with the nodes as the center, taking the node 1 as an example, and the network topology information is shown in fig. 3.

2. Master/backup node election

A plurality of networks exist in a user network, each network needs to select a master/standby node, and the resources of the standby node and the common node are distributed by a master node. The master/standby node election algorithm is based on neighbor nodes obtained by topology discovery, and the master node election algorithm is as follows:

after the node is started, the node interacts with a resource planning system to acquire network accessibility, selects the node as a common node when the node is inaccessible, and finishes election;

otherwise, the node reads the topology information, if the node is empty, the node is selected as the main node, and the election is finished;

otherwise, judging that the priority of each other node is higher than that of the other node, selecting the other node as a standby node, and finishing election;

otherwise, selecting the self as the main node, and finishing election;

if the remaining nodes have no nodes and the back ends of the nodes can be reached, no standby nodes exist; if only one back end of the remaining nodes can be reached, the remaining nodes are used as standby nodes; if the remaining nodes have two or more back ends reachable, the standby nodes are selected out through the priority.

Priority comparison is used in the election algorithm, which is as follows:

judging the accessibility of a resource planning system, wherein the priority is high if the network is accessible;

then judging the priority, wherein the priority with a larger value is higher;

if the priority is the same, the node IDs are used for comparison (the node IDs take the MAC addresses of the unit ring nodes), and the node ID with the larger node ID has higher priority.

Therefore, if the master node is selected, after the node is newly added, if the back-end network of the original master node is still reachable, the master node is not changed. Otherwise, the master node election is performed again.

3. Network resource acquisition

After the master node election is completed, the master node reports network topology information (specific information is shown in table 1) taking the master node as a center to the central network, the central network obtains network resource allocation through comprehensive calculation, and the network resources are allocated to the master node through the resource planning system.

TABLE 1

And the master node connected with the central network actively sends a Hello message to the central network every 1s, and the central network responds after receiving the Hello message. If the central network does not receive the Hello message for 3s continuously, the central network considers that the Hello message is disconnected with the central network; if the host node does not receive the central network response for 3s continuously, the host node is considered to be out of contact with the central network (the Hello message in the communication process is shown in table 2). After the neighbor relation between the central network and the host node is established, the query is actively initiated to query the currently used resources of the intranet. And the master node receives the query and then responds to the current topology and the node resources. And the central network issues the resources distributed by the resource planning system to the user network.

TABLE 2

4. Network resource allocation

And after the main node acquires the network resources from the central network, the main node allocates IP address pool resources to each node according to the node MAC in the network resources. Each newly accessed node needs to be allocated some network resources, such as IP addresses, etc. The network resources are maintained and distributed by the main node, the main node maintains a network resource pool, when a new access node needs to be available, the resources are distributed from the network resource pool and transmitted to the new access node, and if the network resource pool can not distribute the resources, the network resources are distributed according to the newly acquired network resources; and when a node leaves from the unit ring domain, the resource is recycled to the network resource pool.

The master node divides a total address pool (when configuration is received) received from the central network into a plurality of sub-address pools according to basic units (16 are temporarily set), each sub-address pool corresponds to one node, and the sub-address pools of all the nodes are not allocated during initialization. Traversing each node according to the distance between the node and the main node and the standby node:

1) and setting the resources of the node according to the saved resource configuration.

2) The distances to the standby nodes are calculated from the left side and the right side respectively by taking the main node as the center. If the number of the hop is up, the number of the hop is recorded, and if the number of the hop is not up, the number of the hop is 0.

3) Right and left demarcation points are calculated. The left/right demarcation points are: the hop count is 0, which indicates that the direction is not reachable; if the hop number is 1, the neighbor is the standby node; hop count >1 hop count-1/2.

4) The resources are checked and allocated starting from the right side and starting from the left side.

And acquiring the network resources allocated to the nodes, and comparing the network resources allocated to the nodes according to the node IDs. And comparing with the ring network resource, and if no conflict exists, saving the ring network resource into the node resource configuration. Otherwise, the unallocated allocated nodes are reselected.

If the hop count is 0, stopping the direction checking and distribution;

if the hop number is 1, if the standby node resource is not allocated, the standby node resource is allocated;

if the hop count is greater than 1, the hop count is-1/2, and the quotient + the remainder is the node for which the load is shared in the direction (routing strategy); the rest is the node for load sharing to standby. And allocates network IP resources.

And establishing a network resource distribution record linked list, and taking the node MAC as a key value. And establishing a sub-address pool linked list and setting the sub-address pool linked list to be unallocated. And using the node MAC as a key value, searching whether the node is allocated once or not in the allocation history record through the key value, if so, using the node configuration, setting the corresponding sub-address pool as allocated, and processing the next node.

The above scheme is merely illustrative of a preferred example, and is not limiting. When the invention is implemented, appropriate replacement and/or modification can be carried out according to the requirements of users.

The number of apparatuses and the scale of the process described herein are intended to simplify the description of the present invention. Applications, modifications and variations of the present invention will be apparent to those skilled in the art.

While embodiments of the invention have been disclosed above, it is not intended to be limited to the uses set forth in the specification and examples. It can be applied to all kinds of fields suitable for the present invention. Additional modifications will readily occur to those skilled in the art. It is therefore intended that the invention not be limited to the exact details and illustrations described and illustrated herein, but fall within the scope of the appended claims and equivalents thereof.

Claims

1. A method for allocating network resources to a user network side in a distributed system, comprising:

2. The method for network resource allocation on the user network side in a distributed system as claimed in claim 1, further comprising:

3. The method according to claim 1, wherein in step one, the topology obtaining means is configured to include:

4. The method according to claim 1, wherein in the step two, the determining means for determining whether each node can reach the central network is configured to include:

s20, each node sends corresponding Hello message to the central network;

5. The method for allocating network resources to the user network side in the distributed system as claimed in claim 1, wherein in step two, the priority election means is configured to include:

6. The method according to claim 1, wherein in step two, when the master node is elected and a new node is added, it is determined whether the master node back-end network is still reachable, if so, the master node is not changed, otherwise, the master node election operation is performed again.

7. The method as claimed in claim 5, wherein in step three, the master node periodically sends Hello message to the central network, and if the central network does not receive Hello message sent by the master node in a period, or the master node does not receive response from the central network in a period, the master node is considered to be out of connection with the central network.

8. The method according to claim 1, wherein in step three, the master node performs resource allocation to the standby node and the normal node based on a resource management algorithm running on the master node after receiving the resource allocation requests of each child node.

9. The method for network resource allocation on the user network side in a distributed system according to claim 8, wherein said resource management algorithm comprises:

10. The method for allocating network resources to the user network side in a distributed system as claimed in claim 8, wherein in S32, the master node establishes a record linked list for network resource allocation, the record linked list uses the MAC address of each node as a key value, divides the total address received by the central network into a plurality of sub address pools corresponding to the nodes, and sets each sub address pool to be in an unallocated state;