WO2018099384A1 - Packet load balancing method, apparatus and system - Google Patents

Abstract

Provided are a packet load balancing method, apparatus and system. The method comprises: a load balancing server receiving a user request packet copied and introduced into a local operator network; and the load balancing server balancing loads of the user request packet to a corresponding redirection server. By means of the present invention, by using a load balancing server to balance loads of a user request packet, copied and introduced into a local operator network, to a redirection server, a user redirection service function is realized, thereby solving the problem in the prior art that dedicated hardware devices are required to realize load balancing and cache control, and saving, for an operator, on the expense for buying load balancing hardware devices.

Description

Message load balancing method, device and system Technical field

The present invention relates to the field of load balancing, and in particular, to a load balancing method, apparatus, and system for Internet replication messages.

Background technique

With the rapid development of the broadband interconnect cache service, the network construction of telecom operators has changed from the traditional basic network construction to the direction of user demand. For various web applications in the Internet business, video downloading and end-to-end access consume a lot of bandwidth. End users occupy most of the resources and need to find a solution to the network bottleneck.

In order to obtain a large amount of value-added revenue from Internet application services, operators generally use the Internet Cache (Cache) strategy to deploy a certain number of cached WebCache devices and peer-to-peer network cache (P2P cache) devices on the original network node. At present, in China Telecom and China Mobile, it is also required to deploy Deep Packet Inspection (DPI) optical splitting devices at the core layer or the aggregation layer to import HTTP and P2P content hotspots into the local carrier network to achieve local cache proximity. Services to improve network bandwidth and data utilization.

However, in order to process the DPI split or port mirrored packets, and respond to the user service request, a large number of content distribution network cache (CDN cache) system devices need to be deployed to handle the packet load balancing. These CDN cache system devices are an urgent problem to be solved.

The current popular load balancing strategy is to use hardware load balancing devices (such as F5, RadWare hardware vendors), as shown in Figure 1, but the disadvantage of this solution is that the hardware load balancing equipment is expensive, and the versatility and scalability are not good.

Summary of the invention

The embodiments of the present invention provide a method, a device, and a system for load balancing a packet, so as to at least solve the problem that a dedicated hardware device is required in the related art to implement load balancing cache control.

According to an embodiment of the present invention, a load balancing method for a message is provided, including:

The load balancing server receives the user request message that is imported into the local carrier network, and the load balancing server loads the user request message to the corresponding redirect server.

In one embodiment, the user request message is copied and introduced at least by one of the following methods: deep packet detection DPI optical access, local domain name server Local DNS access, policy routing access, and port mirroring access.

In an embodiment, the type of the user request message is at least one of the following: a hypertext transfer protocol HTTP, a peer-to-peer network P2P, a local DNS.

In an embodiment, before the load balancing server load balances the user request packet to the corresponding redirect server, the method further includes:

Determining the user request message, and determining a user request message that needs to be redirected according to the destination IP address or MAC address of the user request message and the flag set in the user request message. .

In an embodiment, the user request message that needs to be redirected is determined according to the destination IP address or the MAC address of the user request message and the flag set in the user request message, including:

When the destination IP address or the MAC address of the user request packet is a local address, the user request packet is forwarded according to the protocol stack.

When the destination IP address or the MAC address in the user request packet is not the local address, the flag set in the user request packet is compared with a preset value, and if they match, the The user request packet is a user request packet that needs to be redirected, and is stored in the redirect buffer queue. If not, the user request packet is discarded.

In one embodiment, the load balancing server loads and balances the user request message. The corresponding redirecting server includes: the load balancing server redirects the user request packet that needs to be redirected to the corresponding redirecting server according to the predetermined load balancing policy.

In an embodiment, the load balancing server is multiple, and each load balancing server corresponds to a different type of user request message.

In one embodiment, the redirect server is multiple, and the load balancing server selects a corresponding redirect server based on at least one of the following: a quality of service of the redirect server, a routing weight, a balancing policy, and a hash of the target address. .

In an embodiment, after the load balancing server load balances the user request packet to the corresponding redirect server, the method further includes: the redirecting server determining, according to the user request packet, a corresponding resource cache server And sending a redirect message to the user; the user initiates a user service request to the corresponding resource cache server according to the redirect message.

According to another embodiment of the present invention, a load balancing server is further provided, including:

a packet capture pre-processing module, configured to receive a user request message that is copied and imported to the local carrier network;

The request load balancing module is configured to load balance the user request packet to a corresponding redirect server.

In an embodiment, the user request message is at least introduced by using one of the following methods: DPI optical access, Local DNS access, policy routing access, and port mirroring access.

In an embodiment, the type of the user request message is at least one of the following: HTTP, P2P, local DNS.

In an embodiment, the packet capture pre-processing module includes: a routing policy unit, configured to determine, according to the destination IP address/MAC address of the user request packet, a user request packet that needs to be redirected; The parsing unit is configured to parse the user request packet that needs to be redirected, and further determines the user request packet that needs to be load balanced according to the preset filtering rule; and the request redirection unit is configured to satisfy the redirection and responsible for balancing. The user request message is sent to the request load balancing module.

In one embodiment, the request load balancing module includes: a load balancing decision unit, configured to redirect a user request message that needs to be redirected to a corresponding redirect server according to the configured load balancing policy; The equalization distribution unit is set to distribute the user request message.

In one embodiment, the request load balancing module includes: a heartbeat keep-alive unit, configured to send and receive a TCP_CHECK message between the load balancing server and the redirect server, and the redirect server is serviceable Stateful detection, heartbeat keepalive parameter configuration, redirect server address maintenance, and display of serviceable statistics; balance policy management unit, configured to configure and deliver the load balancing policy, and start and stop the heartbeat keep-alive service stop.

In one embodiment, the load balancing decision unit is further configured to obtain a load balancing policy configuration and a redirect server address list, and obtain a list of serviceable redirect server addresses from the heartbeat keep-alive unit, and count the current redirect server. The available load to increase or decrease the amount of requests from the current redirect server.

According to still another embodiment of the present invention, there is also provided a load balancing system, comprising the load balancing server described in the foregoing, and a redirect server and a resource cache server, wherein

The redirecting server is configured to determine a corresponding resource cache server according to the user request packet from the load balancing server, and send a redirect message to the user;

The resource cache server is configured to receive a service request from a user and provide a corresponding request resource to the user.

In an embodiment, the load balancing server is multiple, and each load balancing server corresponds to a different type of user request message.

In one embodiment, the redirect server is multiple, and the load balancing server selects a corresponding redirect server based on at least one of the following: a quality of service of the redirect server, a routing weight, a balancing policy, and a hash of the target address. .

In one embodiment, the load balancing server is in the same subnet as the redirect server.

In one embodiment, the redirect server and the resource cache server are separate Deploy or set up.

In the embodiment of the present invention, the load balancing server is used to load the user request packet of the local carrier network to the redirect server, and the user redirection service function is implemented. The cost of the equipment.

DRAWINGS

The drawings described herein are intended to provide a further understanding of the invention, and are intended to be a part of the invention. In the drawing:

1 is a schematic diagram of a hardware load balancing framework according to the related art;

2 is a schematic diagram of a load balancing server processing a copy message and a load balancing framework according to an embodiment of the present invention;

3 is a schematic diagram of a load balancing server load balancing device access mode according to an embodiment of the present invention;

4 is a schematic diagram showing the composition of a load balancing server load balancing device according to an embodiment of the present invention;

FIG. 5 is a flowchart showing the operation of a load balancing server load balancing device according to an embodiment of the present invention; FIG.

6 is a flowchart of preprocessing of packet capture by using a server according to an embodiment of the present invention;

FIG. 7 is a flow chart showing the operation of implementing a request load balancing function in an embodiment of the present invention.

detailed description

The invention will be described in detail below with reference to the drawings in conjunction with the embodiments. It should be noted that the embodiments in the present application and the features in the embodiments may be combined with each other without conflict.

2 is a schematic diagram of a load balancing server processing a copy message and a load balancing framework according to an embodiment of the present invention. In this embodiment, the packet capture pre-processing and load balancing are implemented by a load balancing server, which may be a general-purpose server, for example, may be any conventional PC general-purpose server device, and data that is usually deployed in the operator's computer room. In the network, configure 10Gigabit The fiber-optic network card device can have a message input and output function.

In this embodiment, there is no special requirement for the configuration of the load balancing server device. For a node with a relatively large service bandwidth, a network card device with good performance can be selected. For a node with a small amount of traffic, a network card device with moderate performance can be selected. .

According to the actual situation of the current network, the load balancing server receives the packet through the network port of the input stream. The network card does not need to be configured with an IP address. The network interface of the load balancing server that requests the load balancing output stream can configure the local LAN segment IP address. , but requires that the cache server be configured in the same subnet segment. At the same time, a set of load balancing IP address whitelists is configured on the load balancing server for load balancing packet output. Correspondingly, the IP address of the same network segment is configured on the cache server for interface load balancing packets.

In one embodiment, as shown in FIG. 2, the load balancing server receives the code stream that is split by the DPI, and filters and analyzes the code stream by processing the routing policy module and the message parsing module in the replication message mechanism.

Since there are different load balancing devices depending on the bandwidth of the incoming code stream in some usage scenarios, correspondingly, each load balancing device is provided with a different load balancing distribution unit.

At the same time, in the actual application, the redirect server has different roles and roles. The type of packets that need to be captured by each load balancing server may be different. For example, some are responsible for HTTP packet processing and load balancing, and some are responsible for P2P reporting. Context capture and load balancing. Therefore, different packet filtering rules and load balancing policies can be configured for different load balancing requirements. Correspondingly, the packet filtering rules and load balancing policies configured by each load balancing server may also be different.

The load balancing device determines, after receiving the packet that meets the requirement, and forwards the packet to the redirecting server according to the preset policy.

Similarly, due to the limitation of the bandwidth capability of each redirecting server, the system also needs to deploy multiple redirecting servers to balance the packets received by the load balancing server and further parse the packets.

In the embodiment of the present invention, as shown in FIG. 3, the load balancing server receives the DPI optical splitting code stream sent by the network or the domain name request message introduced by the LocalDNS. Among them, the newspaper The text type is generally based on TCP or UDP format, such as HTTP protocol packets, P2P packets, and DNS request packets.

The embodiment of the invention further provides a load balancing server for processing a copy message and a load balancing device, as shown in FIG. 4 . The apparatus includes a message capture pre-processing module 41 and a request load balancing module 42.

The packet capture preprocessing module includes: a routing policy unit 411, a message parsing unit 412, and a request redirection unit 413. The message capture pre-processing module implements control of the incoming code stream by executing configuration commands and setting parameters such as message filtering mode, and the capture instruction includes start and end, and the like. The packet filtering mode includes: packet type, packet eigenvalue, payload of the payload, and configuration of the packet buffer queue.

The routing policy unit 411 is configured to process the replication message module to perform routing judgment on the packet on the operating system kernel protocol stack netfilter module. The kernel module works at the chain link of the kernel protocol stack PRE_ROUTING. Then, after the route is judged, the message enters the kernel protocol stack for processing.

The message parsing unit 412, that is, the HTTP/P2P GET kernel module, is configured to determine and analyze the link layer packet type and the service packet type, and the packet conforming to the load balancing offload condition (NF_ACCEPT) enters the kernel protocol stack. The load balancing distribution process is processed, and if it does not meet the conditions, it is discarded by the message filtering unit (NF_DROP).

The request redirection unit 413 is configured to direct the request of the cache resource to the cache server, parse the valid data packet, and construct a response (eg, http/302) redirect message to the user, thereby implementing the user request. Redirect boot.

In an embodiment, the message capture pre-processing module 41 is responsible for receiving the code stream, and the routing policy unit 411 determines the destination IP address or the destination MAC address of the captured message to determine whether it is the local IP address, and if it is the local machine, the protocol is taken. The stack routing and forwarding process, if not the local device, the message parsing unit 412 sets the packet to be parsed, and offsets the payload of the packet by a preset offset, and a preset value. Compare, if the judgment condition is met Upon entering the buffer queue, the message proceeds to the request redirection unit 413 for processing. Implement load balancing control and offload processing for incoming packets. If the judgment condition is not met, the packet is discarded. The request redirection unit constructs a valid data packet, and selects an available cache server according to the available cache server list and corresponding resource information, and redirects the user request to the cache server. It should be noted that the request redirection unit only modifies the HTTP packet. The related resource content in the load is incorporated into the load balancing cache queue. The message is still output to the cache server by the load balancing module.

The request load balancing module 42 includes a load balancing decision unit 421, a load balancing distribution unit 422, an equalization policy management unit 423, and a heartbeat keep-alive unit 424.

The load balancing decision unit 421 is configured to obtain a load balancing policy configuration and a redirect server address list, and obtain a serviceable redirect server address list from the heartbeat keep-alive unit. The load balancing decision unit also loads the available load of the current redirect server, and is set to increase or decrease the request amount of the current redirect server device.

The load balancing distribution unit 422 is configured to distribute the load balancing message.

The equalization policy management unit 423 is configured to start and stop the load balancing policy configuration delivery and the heartbeat keep-alive service.

The heartbeat keep-alive unit 424 is configured to send and receive TCP_CHECK messages between the load balancing server and the redirect server, to detect service state of the redirect server, configure heartbeat keep-alive parameters, redirect server address maintenance, and Display of service statistics.

The embodiment of the present invention further provides a workflow for processing a copy message and a software load balancing device. As shown in FIG. 5, the following processes are included:

First, the user triggers the request to the load balancing server packet capture preprocessing module through the operation of the front end;

Secondly, the request message processing module enters the process of processing the copy message in the load balancing server, and the operating system kernel route pre-processing (PRE_ROUTING) triggers the message parsing module to view the link layer message type of the code stream in the buffer area, and the judgment chain The layer-layer packet type is sent to the local packet (PACKET_HOST). If the packet capture condition is met, the request redirection module is directly sent. Continue processing. If it is another host message (PACKET_OTHERHOST), set the local link layer message type (PACKET_HOST), and send the message to the upper application to continue processing.

Further, the load balancing server packet parsing unit determines whether the application layer service packet type is an HTTP/P2P packet. If the condition is met, the process proceeds to the next step.

Further, the load balancing server requests the redirecting unit to obtain a list of currently available redirect server addresses;

Further, the load balancing server requests the redirecting unit to initiate a request redirection call to the redirect server side according to the obtained service address and the service URL;

Further, the load balancing server redirects the server's QoS, routing weights, balancing policies, and target address hashes to perform load balancing policy selection, such as whether the redirect server is currently available, whether it times out, etc., and calls load balancing offloading. The interface sends the packet to multiple redirect servers and simultaneously calls the load balancing statistics interface to write to the statistics.

Further, returning the load balancing service request processing result;

In this implementation manner, the main purpose of the load balancing server to update the statistical service quality is to automatically calculate the available load of the current service, that is, increase or decrease the request amount of the current redirect service device, and timely list from the current redirect server when the unavailable is found. Excluded.

In an actual application, the implementation process of the embodiment of the present invention may include processing a copy packet process and a load balancing workflow, which are separately described below.

FIG. 6 is a flowchart of preprocessing of packet capture by using a server according to an embodiment of the present invention. As shown in FIG. 6, the method includes the following steps:

In step S601, the load balancing server load balancing device is connected to the network, where the inbound network port is connected to the DPI code stream or the DNS packet.

In step S602, the DPI optical splitting device is deployed in the network or the load balancing code stream is introduced through the DNS.

Step S603, the load balancing server packet capture pre-processing module determines that the load is average The balance server sends the packet whose destination MAC address or destination IP address is not the local packet to the packet parsing unit for processing.

Step S604: The message parsing unit in the message pre-preprocessing module parses the message, and after determining the predetermined value setting after a predetermined number of bytes of offset, the packet discarding process does not satisfy the preset condition. If the protocol type satisfies TCP or UDP and the payload of the message is HTTP, P2P, and LocalDNS, the request is redirected to the unit for processing.

Step S605, the request redirection unit in the message capture pre-processing module constructs a valid data packet, selects an available redirect server according to the available redirect server IP address list and corresponding resource information, and redirects the user request to load balancing. On the cache server.

In this embodiment, in order to process the copy message function, functions of starting and uninstalling the HTTP/P2P kernel module, and function commands for starting and stopping the message fetching are also provided. This can be done with a shell script provided by the Linux system.

In this embodiment, the redirection server is deployed separately or in combination with the cache server. Depending on the packet type and function role, a redirection server with different functions can be configured.

FIG. 7 is a working flowchart of implementing a request load balancing function according to an embodiment of the present invention. As shown in FIG. 7, the method includes the following steps:

In step S711, the redirection server is connected to the network, and the device and the load balancing device are in the same subnet to implement network interworking.

Step S712: The user sends an HTTP/P2P request message, and the load balancing server receives the DPI split/DNS request message.

Step S713: The redirecting server accesses the network port to receive the output packet of the load balancing server, and the redirecting server receives the user service response packet from the network port.

Step S714, configuring the load balancing policy, and redirecting the server address list and the heartbeat keep-alive parameter.

Step S715, the load balancing server filters the application layer packet, for the packet class. The type is HTTP/P2P, which is sent to the load balancing module, and vice versa.

Step S716: The load balancing server forwards the packet to the load balancing module according to the load balancing condition.

Step S717: The request load balancing module obtains a load balancing policy, redirects the server address list, the weight, and the redirect server service state to perform load balancing output on the code stream.

In this embodiment, in order to manage the request load balancing function, a function of configuring a load balancing policy and a function of starting and stopping the heartbeat keep-alive service are also provided.

Embodiments of the present invention also provide a storage medium. The storage medium may be arranged to store program code for performing the flow steps in the above embodiments:

Optionally, in this embodiment, the foregoing storage medium may include, but not limited to, a USB flash drive, a Read-Only Memory (ROM), a Random Access Memory (RAM), a mobile hard disk, and a magnetic memory. A variety of media that can store program code, such as a disc or a disc. For specific examples in this embodiment, reference may be made to the examples described in the foregoing embodiments and the optional embodiments, and details are not described herein again.

It will be apparent to those skilled in the art that the various modules or steps of the present invention described above can be implemented by a general-purpose computing device that can be centralized on a single computing device or distributed across a network of multiple computing devices. Alternatively, they may be implemented by program code executable by the computing device such that they may be stored in the storage device by the computing device and, in some cases, may be different from the order herein. The steps shown or described are performed, or they are separately fabricated into individual integrated circuit modules, or a plurality of modules or steps thereof are fabricated as a single integrated circuit module. Thus, the invention is not limited to any specific combination of hardware and software.

The above description is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Any modifications, equivalent substitutions, improvements, etc., are intended to be included within the scope of the present invention.

Claims (23)

A load balancing method for a message, including: The load balancing server receives the user request message that is copied and imported into the local carrier network. The load balancing server load balances the user request packet to a corresponding redirect server. The method of claim 1, wherein the user request message is introduced at least by one of the following ways: Deep packet inspection, DPI optical access, local DNS server local DNS access, policy routing access, and port mirroring access. The method according to claim 1, wherein the type of the user request message is at least one of the following: a hypertext transfer protocol HTTP, a peer-to-peer network P2P, a local DNS. The method of claim 1, wherein the load balancing server before the load balancing the user request message to the corresponding redirecting server further comprises: Determining the user request message, and determining a user request message that needs to be redirected according to the destination IP address or MAC address of the user request message and the flag set in the user request message. . The method according to claim 4, wherein the user request message of the redirect service is determined according to the destination IP address or MAC address of the user request message and the flag set in the user request message. ,include: When the destination IP address or the MAC address of the user request packet is a local address, the user request packet is forwarded according to the protocol stack. The method of claim 5 further comprising: When the destination IP address or the MAC address in the user request packet is not the local address, the flag set in the user request packet is compared with a preset value, if If the matching is performed, the user request packet is a user request packet that needs to be redirected, and is stored in the redirect buffer queue. If not, the user request packet is discarded. The method of claim 1, wherein the load balancing server load balances the user request message to a corresponding redirect server, including: The load balancing server redirects the user request packet that needs to be redirected to the corresponding redirect server according to the predetermined load balancing policy. The method of claim 1, wherein the load balancing server is a plurality, and each load balancing server corresponds to a different type of user request message. The method according to claim 1, wherein the number of redirect servers is multiple, and the load balancing server selects a corresponding redirect server based on at least one of: a quality of service of the redirect server, a routing weight, an equalization policy, The target address is hashed. The method of claim 1, wherein the load balancing server after load balancing the user request message to the corresponding redirect server further comprises: Determining, by the redirecting server, the corresponding resource cache server according to the user request packet, and sending a redirect message to the user; The user initiates a user service request to the corresponding resource cache server according to the redirect message. A load balancing server that includes: a packet capture pre-processing module, configured to receive a user request message that is copied and imported to the local carrier network; The request load balancing module is configured to load balance the user request packet to a corresponding redirect server. The load balancing server according to claim 11, wherein the user request message is introduced at least by one of the following ways: Deep packet inspection, DPI optical access, local DNS server local DNS access, policy routing access, and port mirroring access. The load balancing server according to claim 11, wherein the type of the user request message is at least one of the following: a hypertext transfer protocol HTTP, a peer-to-peer network P2P, and a local DNS. The load balancing server according to claim 11, wherein the message capture preprocessing module comprises: a routing policy unit, configured to determine, according to the destination IP address or the MAC address of the user request packet, a user request packet that needs to be redirected; The packet parsing unit is configured to parse the user request packet that needs to be redirected, and further determine the user request packet that needs to be load balanced according to the preset filtering rule. The request redirection unit is configured to send the user request message that satisfies the redirection and responsible for the equalization to the request load balancing module. The load balancing server according to claim 11, wherein the request load balancing module comprises: The load balancing decision unit is configured to redirect the user request packet that needs to be redirected to the corresponding redirect server according to the configured load balancing policy; The load balancing distribution unit is set to distribute the user request message. The load balancing server of claim 15, wherein the request load balancing module comprises: a heartbeat keep-alive unit, configured to send and receive a TCP_CHECK message between the load balancing server and the redirect server, and detect the serviceable state of the redirect server, configure a heartbeat keep-alive parameter, and redirect the server address Maintenance, and display of serviceable statistics; The balance policy management unit is configured to configure and deliver the load balancing policy, and start and stop the heartbeat keep-alive service. The load balancing server according to claim 16, wherein The load balancing decision unit is further configured to obtain a load balancing policy configuration and a redirect server address list, and obtain a list of serviceable redirect server addresses from the heartbeat keep-alive unit, and count the available load of the current redirect server to improve Or reduce the amount of requests from the current redirect server. A load balancing system, comprising the load balancing server according to any one of claims 11-17, and a redirect server and a resource cache server, wherein The redirecting server is configured to determine a corresponding resource cache server according to the user request packet from the load balancing server, and send a redirect message to the user; The resource cache server is configured to receive a service request from a user and provide a corresponding request resource to the user. The load balancing system of claim 18, wherein the load balancing server is a plurality, and each load balancing server corresponds to a different type of user request message. The load balancing system according to claim 18, wherein the number of redirection servers is plural, and the load balancing server selects a corresponding redirection server based on at least one of: redirection server quality of service, routing weight, and equalization Policy, target address hashing. The load balancing system of claim 18, wherein the load balancing server is in the same subnet as the redirecting server. The load balancing system according to claim 18, wherein the redirect server and the resource cache server are separately deployed or combined. A computer readable storage medium storing a computer program that, when executed, performs the method of any one of claims 1 to 10.

Images