WO2012092765A1

WO2012092765A1 - Domain name system and method thereof for providing load balancing

Info

Publication number: WO2012092765A1
Application number: PCT/CN2011/077923
Authority: WO
Inventors: 王亮
Original assignee: 中兴通讯股份有限公司
Priority date: 2011-01-04
Filing date: 2011-08-02
Publication date: 2012-07-12
Also published as: CN102572001A

Abstract

The present invention provides a domain name system (DNS) and a method thereof for providing load balancing. The method comprises: after receiving a notification message sent by an IP address server corresponding to a domain name, a DNS querying a load value of the IP address server, and updating a load value of an IP address corresponding to the IP address server; and after receiving a domain name query message sent by a DNS client, querying an IP address corresponding to the domain name, and returning the IP address with the smallest load value to the DNS client. In the present invention, the DNS dynamically queries the load value of the IP address server and dynamically updates data such as the load value of the IP address. When a DNS client sends a domain name query message, a DNS server queries an IP address corresponding to the domain name and returns the IP address with the smallest load value to the DNS client. In this way, the problem of dynamically acquiring a load state of a server corresponding to an IP address is solved, and dynamic load balancing of a network is implemented.

Description

Domain name system and method for providing load balancing thereof

The invention relates to a load balancing technology in a communication network, in particular to a domain name system

(Domain Name System, DNS) and its method of providing load balancing. Background technique

In a network that currently uses DNS, if DNS is required to provide load balancing, DNS can only provide a static-based load balancing mechanism. For example, if the DNS in the Ethernet is required to provide the load balancing function for the Baidu domain name, the load of the IE client accessing the Baidu website needs to be evenly distributed to the IP address corresponding to the Baidu domain name. The specific processing procedure is as follows: DNS enables the load balancing mechanism to query the IP address of the corresponding domain name www, baidu, com in the local cache or local database, and finds three IP addresses: 220.181.37.56, 220.181.37.57, 220.181.37.58, DNS After confirming that all three IP addresses are in a connectable state, the IP address is returned to the DNS client by polling. The DNS client obtains one of the IP addresses and uses the obtained IP address to perform the next service. On the surface, DNS divides the services of these DNS clients into IP addresses. In fact, this static load balancing mechanism cannot sense how the current load of the server corresponding to the three IP addresses is. According to the actual load status of the server, the IP address of the best state is provided. Even if the DNS provides the available IP address by polling, it is not necessarily the IP address of the server with the lightest load.

In the face of the current increasingly large communication network, as the traffic volume of the communication network continues to grow, it can be known that due to the continuous increase of the traffic volume, the specific load status of the server corresponding to the IP address can better realize the DNS network. Load balancing. However, how to implement dynamic load balancing in DNS, no one has proposed relevant technical implementation solutions. Summary of the invention

In view of this, the main object of the present invention is to provide a DNS and a method for providing load balancing, which can implement dynamic load balancing of a network.

In order to achieve the above object, the technical solution of the present invention is achieved as follows:

A domain name system DNS provides a method for load balancing, and the method includes:

After receiving the notification message sent by the IP address server corresponding to the domain name, the DNS queries the load value of the IP address server, and updates the load value of the IP address corresponding to the IP address server;

After receiving the domain name query message sent by the DNS client, the IP address corresponding to the domain name is queried, and the IP address with the lowest load value is returned to the DNS client.

Further, the method further includes: the DNS stores an IP address corresponding to the domain name and a load value of the IP address.

Further, the method further includes: the IP address server corresponding to the domain name calculates its own load value by using a preset load algorithm, and sends a notification message to the DNS when it detects that the self load value changes.

When the IP address server corresponding to the domain name detects that the load value changes, the notification message is sent to the DNS:

The IP address server corresponding to the same domain name calculates its current load value according to a preset period and the same load algorithm, and compares the calculated load value with the load value of its previous cycle, when the comparison result exceeds the preset valve. When the value is sent, a notification message is sent to the DNS.

Wherein, when the comparison result exceeds the preset threshold value: the calculated load value fluctuates by more than ± 5% with respect to the load value of the previous cycle.

The DNS querying the load value of the IP address server is:

The dynamic load query module of the DNS sends a DNS dynamic load query DNS-DLQ message to the IP address server; The IP address server returns a DNS-DLQ response containing the load value of the IP address server to the DNS.

A DNS, the system includes: a dynamic load query module, a DNS server; wherein, the dynamic load query module is configured to: after receiving the notification message sent by the IP address server corresponding to the domain name, query the load value of the IP address server And send an update message;

a DNS server, after receiving the update message, updating a load value of the IP address corresponding to the IP address server; after receiving the domain name query message sent by the DNS client, querying the IP address corresponding to the domain name, and The IP address with the lowest load value is returned to the DNS client.

Further, the DNS server is specifically configured to store an IP address corresponding to the domain name and a load value of the IP address.

Further, the dynamic load query module is specifically configured to send a DNS-DLQ message to the IP address server, and receive a DNS-DLQ response that is returned by the IP address server and includes a load value of the IP address server.

The DNS provided by the present invention and the method for providing the load balancing method can query the load value of the IP address server through the DNS, and dynamically update the data such as the load value of the IP address; when the DNS client sends the domain name query message, the DNS queries the domain name corresponding to the domain name. The IP address returns the IP address with the most load value to the DNS client. Thus, the DNS client obtains the currently available and lightly loaded IP address from the DNS, thereby solving the problem of dynamically obtaining the IP address. The problem of IP address server load status achieves dynamic load balancing of the DNS network. DRAWINGS

1 is a schematic flowchart of a method for providing a load balancing method for a DNS according to an embodiment of the present invention; FIG. 2 is a schematic flowchart of a specific embodiment of a method for providing a load balancing method for a DNS according to an embodiment of the present invention;

FIG. 3 is a schematic structural diagram of a DNS providing load balancing system according to an embodiment of the present invention. detailed description

The basic idea of the present invention is: after receiving the notification message sent by the IP address server corresponding to the domain name, the DNS queries the load value of the IP address server, and updates the load value of the IP address corresponding to the IP address server; After the domain name query message sent by the DNS client, the IP address corresponding to the domain name is queried, and the IP address with the smallest load value is returned to the DNS client.

The present invention will be further described in detail below with reference to the accompanying drawings.

FIG. 1 shows a flow of a DNS providing load balancing method according to the present invention. As shown in FIG. 1, the method includes the following steps:

Step 101: After receiving the notification message sent by the IP address server corresponding to the domain name, the DNS queries the load value of the IP address server, and updates the load value of the IP address corresponding to the IP address server.

Generally, a domain name is configured with at least one IP address, and each IP address corresponds to an IP address server. When an IP address server in the IP address server group corresponding to the domain name detects that its own load value changes, Sending a notification message to the Dynamic Load Query Module (DLQM) of the DNS; wherein the notification message can use a standard DNS-notify message (refer to RFC1996);

Further, the step further includes: the IP address server corresponding to the same domain name calculates its current load value according to a preset period and a load algorithm, and sends a notification message to the dynamic load of the DNS when detecting that the load value thereof changes. Query module

The change of the load value of the self may be: the IP address server compares the calculated load value with the load value of the previous period, and the calculated load value fluctuates with respect to the load value of the previous period. When the preset threshold is set, the load value of the load is changed. Here, the preset threshold is used to reduce the IP address server and the dynamic load query module when the IP address server load value does not change much. Data interaction, the The threshold can be set to ± 5%, 10%, etc. according to actual needs.

When receiving the notification message sent by any IP address server in the IP address server group, the dynamic load query module knows that the load value of the IP address server changes, and sends a query message to the IP address server to query the changed load value. After receiving the query message, the IP address server returns a response containing the load value of the IP address server to the dynamic load query module; after receiving the response message, the dynamic load query module sends a message to the DNS DNS server. To update the load value of the IP address corresponding to the IP address server stored in the DNS server; specifically, the dynamic load query module uses an internal message to update the load value of the IP address and the IP address stored by the DNS server, and the internal message mainly includes the message ID and the IP address. And three parameters of the load value, wherein the IP address is the primary key;

The query message and response may be based on the DNS query type defined in RFC1035, defining Dynamic Load Query (DLQ) messages and responses based on DNS queries, recorded as DNS-DLQ messages and DNS-DLQ responses;

Specifically, the infrastructure of the DNS-DLQ message and the DNS-DLQ response are consistent with the definition of RFC1035, including: Header, Question, Answer, Authority, and Additional. The specific definition of each part can be queried in RFC1035, and will not be described again.

It should be noted that, in the embodiment of the present invention, based on the definition of RFC1035, the DNS-DLQ message adds QTYPE, QCLASS, and QNAME to the Question structure, where QTYPE is DLQ, QCLASS is IN, and QNAME is the IP to be queried. Address; DNS-DLQ response Based on the definition of RFC1035, NAME, TYPE, RDATA are added to the Answer structure, where NAME is the IP address, TYPE is DLQ, and RDATA is the load value; the definition of other parts meets the definition of RFC1035.

Step 102: After receiving the domain name query message sent by the DNS client, the DNS queries the IP address corresponding to the domain name, and returns the IP address with the smallest load value to the DNS client.

Specifically, in this step, the DNS server of the DNS receives the domain sent by the DNS client. Name query message, find the IP address corresponding to the domain name, and read the load value of the IP address. When the IP address corresponding to the domain name is multiple, compare the load value of each IP address, and set the current load value to the most d, the IP address. The address is returned to the DNS client for the next step of the business.

In the embodiment of the present invention, the physical implementation of the DNS server and the dynamic load query module may be set on the same server or on multiple servers; the load value of the IP address and IP address stored by the DNS server may be saved in itself. Real-time updates in the cache area to save the overhead of the physical table, which has the structure of Table 1:

Table 1

The IP Address is a 32-bit Internet address, and the Load Value is a load value. It is a numeric parameter. In the specific implementation, when the DNS initializes this part of the data, the load value corresponding to each IP address in the cache can be set to be the same. value.

FIG. 2 is a flowchart of a specific embodiment of a method for providing a load balancing method according to the present invention. As shown in FIG. 2, the embodiment is described based on a DNS, a DNS client, and an IP address server group corresponding to a domain name, where the DNS includes a DNS. The server and the dynamic load query module, the IP address server group corresponding to the domain name includes three IP address servers, namely: an IP1 server with an IP address of IP1, an IP2 server with an IP address of IP2, and an IP3 server with an IP address of IP3; The embodiment includes the following steps:

Step 1~2, after the IP1 server calculates a change in its own load value through a preset load algorithm, the notification message is sent to the dynamic load query module.

Specifically, in this step, the IP1 server periodically calculates its current load value according to a preset load algorithm, and when it detects that its own load value changes, sends a notification message to the dynamic load query module; Set the calculation period of the IP address server according to the actual needs and the network environment; the load algorithm can be selected according to the needs of the user, for example For example, the CPU usage, memory usage, bandwidth and other parameters of the server can be weighted to obtain the load value of the server, etc.; the notification message can use a standard DNS-notify message (refer to RFC1996);

Wherein, the change of the load value detected by itself may be: The IP1 server compares the calculated load value with the load value of the previous period, and the fluctuation of the calculated load value with respect to the load value of the previous period exceeds the advance When the threshold is set, the load value of the load changes. The preset threshold can be set to ± 5%, ± 10%, etc. according to the actual needs, in order to change the load value of the server at the IP address. In this case, the process of data interaction between the IP address server and the dynamic load query module is reduced.

Step 3~5, after receiving the notification message sent by the IP1 server, the dynamic load query module sends an inquiry message to the IP1 server; and receives a response message returned by the IP1 server, and notifies the DNS server to update the load value of the IP address as IP1;

Specifically, after receiving the notification message, the dynamic load query module learns that the load value of the IP1 server changes, and needs to re-query the load value of the IP1 server, so the query message is sent to the IP1 server; after receiving the query message, the IP1 server receives the query message. Returning a response containing the load value of the IP1 server to the dynamic load query module; after receiving the response message, the dynamic load query module sends a message to the DNS server to update the stored IP address to the load value of IP1; here, the dynamic load The query module uses the internal message to update the IP address stored by the DNS server to the load value of the IP1. The internal message mainly includes three parameters: a message ID, an IP address, and a load value, wherein the IP address is the primary key;

Specifically, in this step, based on the definition of RFC1035, the DNS-DLQ message adds QTYPE, QCLASS, and QNAME to the Question structure, where QTYPE is DLQ, QCLASS is IN, QNAME is IPl, and DNS-DLQ response is in RFC1035. On the basis of the definition, NAME, TYPE, RDATA are added to the Answer structure, where NAME is IPl, TYPE is DLQ, RDATA is the current load value of IPl; other parts are defined to satisfy RFC1035 definition.

Step 6~7, after the IP2 server calculates a change in its own load value through a preset load algorithm, the notification message is sent to the dynamic load query module.

Specifically, in this step, the IP2 server periodically calculates its current load value according to a preset load algorithm, and when it detects that its own load value changes, sends a notification message to the dynamic load query module; The calculation period of the IP address server is set according to the actual needs and the network environment; the load algorithm can be selected according to the needs of the user. For example, the CPU utilization, memory usage, bandwidth and other parameters of the server can be weighted to obtain the load of the server. Values, etc.; notification messages can use standard DNS-notify messages (refer to RFC1996);

Wherein, the change of the load value detected by itself may be: The IP2 server compares the calculated load value with the load value of the previous period, and the fluctuation of the calculated load value with respect to the load value of the previous period exceeds the advance When the threshold is set, the load value of the load changes. The preset threshold can be set to ± 5%, ± 10%, etc. according to the actual needs, in order to change the load value of the server at the IP address. In this case, the process of data interaction between the IP address server and the dynamic load query module is reduced.

Step 8~10, after receiving the notification message sent by the IP2 server, the dynamic load query module sends an inquiry message to the IP2 server; and receives a response message returned by the IP2 server, and notifies the DNS server to update the load value of the IP address as IP2;

Specifically, after receiving the notification message, the dynamic load query module learns that the load value of the IP2 server changes, and needs to re-query the load value of the IP2 server, so the query message is sent to the IP2 server; after receiving the query message, the IP2 server receives the query message. Returning a response containing the load value of the IP2 server to the dynamic load query module; after receiving the response message, the dynamic load query module sends a message to the DNS server to update the stored IP address to the IP2 load value; here, the dynamic load The query module uses internal messages to update the IP address stored by the DNS server to the load of IP2. The internal message mainly includes three parameters: a message ID, an IP address, and a load value, where the IP address is a primary key;

Specifically, in this step, based on the definition of RFC1035, the DNS-DLQ message adds QTYPE, QCLASS, and QNAME to the Question structure, where QTYPE is DLQ, QCLASS is IN, QNAME is IP2, and DNS-DLQ response is in RFC1035. On the basis of the definition, NAME, TYPE, and RDATA are added to the Answer structure, where NAME is IP2, TYPE is DLQ, and RDATA is the current load value of IP2; the definitions of other parts meet the definition of RFC1035.

Step 11~12, after the IP3 server calculates the load value of the load through the preset load algorithm, the notification message is sent to the dynamic load query module.

Specifically, in this step, the IP3 server periodically calculates its current load value according to a preset load algorithm, and sends a notification message to the dynamic load query module when detecting that its own load value changes; The calculation period of the IP address server is set according to the actual needs and the network environment; the load algorithm can be selected according to the needs of the user. For example, the CPU utilization, memory usage, bandwidth and other parameters of the server can be weighted to obtain the load of the server. Values, etc.; notification messages can use standard DNS-notify messages (refer to RFC1996);

Wherein, the change of the load value detected by itself may be: The IP3 server compares the calculated load value with the load value of the previous period, and the fluctuation of the calculated load value with respect to the load value of the previous period exceeds the advance When the threshold is set, the load value of the load changes. The preset threshold can be set to ± 5%, ± 10%, etc. according to the actual needs, in order to change the load value of the server at the IP address. In this case, the process of data interaction between the IP address server and the dynamic load query module is reduced.

Step 13~15, after receiving the notification message sent by the IP3 server, the dynamic load query module sends an inquiry message to the IP3 server; and receives a response message returned by the IP3 server to notify the DNS. The server performs an update of the load value whose IP address is IP3;

Specifically, after receiving the notification message, the dynamic load query module learns that the load value of the IP3 server changes, and needs to re-query the load value of the IP3 server, so the query message is sent to the IP3 server; after receiving the query message, the IP3 server receives the query message. Returning a response containing the load value of the IP3 server to the dynamic load query module; after receiving the response message, the dynamic load query module sends a message to the DNS server to update the stored IP address to the IP3 load value; here, the dynamic load The query module uses the internal message to update the IP address stored by the DNS server to the load value of the IP3. The internal message mainly includes three parameters: a message ID, an IP address, and a load value, wherein the IP address is the primary key;

Specifically, in this step, based on the definition of RFC1035, the DNS-DLQ message adds QTYPE, QCLASS, and QNAME to the Question structure, where QTYPE is DLQ, QCLASS is IN, QNAME is IP3, and DNS-DLQ response is in RFC1035. On the basis of the definition, NAME, TYPE, and RDATA are added to the Answer structure, where NAME is IP3, TYPE is DLQ, and RDATA is the current load value of IP3; the definitions of other parts meet the definition of RFC1035.

It should be understood that, in the actual application, the process of data interaction between each IP address server and the dynamic load query module is not limited in sequence; that is, the calculation and update of the load value of the IP1 server, the IP2 server, and the IP3 server are There is no order limit.

Step 16, the DNS client sends a domain name query message to the DNS server according to the input of the user;

Step 17, after receiving the domain name query message, the DNS server queries the IP address configured in the local cache or the corresponding domain name in the database, and compares the load value of the IP address;

Specifically, in this step, the DNS server first queries the IP address corresponding to the domain name according to the domain name; then queries the load value of each IP address, and compares it to obtain the IP address with the lowest current load value. Step 18: Return the IP address with the lowest load value to the DNS client.

It should be understood that the update of the IP address load value by steps 16-18 and the DNS server is not limited by the above order.

FIG. 3 is a schematic diagram showing the structure of a DNS providing load balancing system according to the present invention. As shown in FIG. 3, the DNS includes: a dynamic load query module 31 and a DNS server 32;

The dynamic load query module 31 is configured to: after receiving the notification message sent by the IP address server in the IP address server group corresponding to the domain name, query the load value of the IP address server, and send an update message; the DNS server 32, After receiving the update message, updating the load value of the IP address corresponding to the IP address server; after receiving the domain name query message sent by the DNS client, querying the IP address corresponding to the domain name, and minimizing the load value IP address is returned to the DNS client

Further, the DNS server 32 is specifically configured to store an IP address corresponding to the domain name and a load value of the IP address.

Further, the dynamic load query module 31 is specifically configured to send a DNS-DLQ message to the IP address server, and receive a DNS-DLQ response returned by the IP address server that includes the load value of the IP address server.

The above is only the preferred embodiment of the present invention and is not intended to limit the scope of the present invention.

Claims

Claim

A method for providing load balancing by a domain name system DNS, wherein the method includes: after receiving a notification message sent by an IP address server corresponding to a domain name, the DNS queries a load value of the IP address server, and queries the The IP address server updates the load value of the IP address;

2. The method according to claim 1, wherein the method further comprises: the DNS storage domain name corresponding to the IP address and the load value of the IP address.

3. The method according to claim 1, wherein the method further comprises: the IP address server corresponding to the domain name calculates its own load value through a preset load algorithm, and sends a notification when it detects that the self load value changes. Message to DNS.

The method according to claim 3, wherein when the IP address server corresponding to the domain name detects that the load value changes, the notification message is sent to the DNS:

5. The method according to claim 4, wherein the comparison result exceeds a preset threshold value: the calculated load value fluctuates by more than ± 5% with respect to the load value of its previous cycle.

6. The method according to claim 1, wherein the DNS queries the load value of the IP address server as:

The DNS dynamic load query module sends a DNS dynamic load query DNS-DLQ message to the IP address server;

The IP address server returns a DNS-DLQ response containing the load value of the IP address server to the DNS.

7. A DNS, wherein the system comprises: a dynamic load query module, a DNS server; wherein

The dynamic load query module is configured to: after receiving the notification message sent by the IP address server corresponding to the domain name, query the load value of the IP address server, and send an update message;

The DNS server is configured to: after receiving the update message, update a load value of the IP address corresponding to the IP address server; after receiving the domain name query message sent by the DNS client, query the IP address corresponding to the domain name, and The IP address with the lowest load value is returned to the DNS client.

The DNS according to claim 7, wherein the DNS server is configured to store an IP address corresponding to the domain name and a load value of the IP address.

The DNS according to claim 7, wherein the dynamic load query module is configured to send a DNS-DLQ message to the IP address server, and receive the server that includes the IP address returned by the IP address server. The load value of the DNS-DLQ response.