WO2021253889A1

WO2021253889A1 - Load balancing method and apparatus, proxy device, cache device and serving node

Info

Publication number: WO2021253889A1
Application number: PCT/CN2021/080746
Authority: WO
Inventors: 王永强; 年彦东
Original assignee: 北京金迅瑞博网络技术有限公司; 北京金山云网络技术有限公司
Priority date: 2020-06-17
Filing date: 2021-03-15
Publication date: 2021-12-23
Also published as: CN111464661A; CN111464661B

Abstract

The present disclosure relates to a load balancing method and apparatus, a proxy device, a cache device and a serving node. The method comprises: storing a file in blocks, and specifically scattering the file content to different cache parties relatively evenly in a block form; and when a request party requests the file, mapping a content request of the file into at least one block request; and obtaining, on the basis of the at least one block request, from at least one cache party, at least one block content available for constituting a complete target file, so as to realize a response to the request party. As the size of the block content corresponding to each block request is relatively uniform, and accordingly, the traffic of the requested file is evenly distributed to different cache parties, the internal network traffic of different cache parties is relatively balanced, which causes no bottleneck to the use of the bandwidth of an external network, improving the external network service capability of a network such as a CDN.

Description

Load balancing method, device, proxy equipment, cache equipment and service node

This disclosure requires the priority of a Chinese patent application filed with the Chinese Patent Office, the application number is 202010555608.6, and the invention title is "load balancing method, device, proxy device, cache device and service node" on June 17, 2020, and its entire content Incorporated in this disclosure by reference.

Technical field

The present disclosure belongs to the field of computer technology, and in particular relates to a load balancing method, device, proxy device, cache device, and service node.

Background technique

CDN (Content Delivery Network, Content Delivery Network) provides a multi-mirror cache of files running on the public network. The CDN network is composed of multiple CDN nodes, and a CDN node is a cluster composed of multiple physical devices (servers). The CDN network uses functions such as load balancing, content distribution, and scheduling to enable users to obtain required content from nearby servers in the network to reduce network congestion and improve user access response speed and hit rate.

Among them, each physical device in the CDN node contains a reverse proxy nginx and a cache storage. The reverse proxy nginx is used to provide load balancing and request forwarding functions, so as to request file content such as url (uniform resource locator, unified resource Locator) The request is forwarded to the cache of the physical device that caches the requested content. For the same URL request, no matter which device is processed by nginx, it will be forwarded to the cache storage of the same device through a consistent hashing algorithm. This ensures that the content of a URL is stored in the cache of one device. Reduce duplicate storage. That is, in the current technology, the reverse proxy nginx can only distribute the content of different URLs (that is, the content of different files) to different caches and store the same URL (file) content in the same cache.

The traffic between nginx and cache is carried out through the intranet, but the content length corresponding to different URLs is usually different, and the access frequency of different URLs is also different. The corresponding URLs with large content length and high access frequency are hashed to the cache, Will consume a higher intranet bandwidth, on the contrary, only need to consume a lower intranet bandwidth. At present, in a CDN network, the internal network bandwidth consumed by a physical device is the same as the external network bandwidth. The uneven distribution of the internal network bandwidth of CND nodes will limit the utilization of external network bandwidth and bring about the use of external network bandwidth. The bottleneck is caused, the maximum bandwidth consumed by the external network cannot reach the maximum saturation (different physical devices are configured with the same maximum available external network bandwidth), and the external network service capabilities of networks such as CDN are correspondingly reduced.

Summary of the invention

In view of this, the embodiments of the present disclosure provide a load balancing method, device, proxy device, cache device, and service node. The purpose is to achieve intranet bandwidth (consumption) between different cache parties (different caches) in service nodes such as CDN. Bandwidth) balance, break the bottleneck of external network bandwidth usage, and improve the external network service capabilities of CDN and other networks.

The specific technical solutions are as follows:

A load balancing method applied to an agent, the method comprising: obtaining a content request carrying a data resource address of a target file; splitting the content request into at least one block request including corresponding block location information; wherein The block location information is determined by the agent according to the data resource address; the at least one block request is distributed to at least one cache respectively corresponding to the block location information in the at least one block request Obtain at least one block content fed back by at least one cache party that has received the corresponding block request in the at least one block request; the absolute difference of the data length of the different block content is less than a predetermined threshold, and the division of the same file When there are multiple pieces of content, the first number of cache parties storing multiple pieces of content is multiple; and the at least one piece of content is fed back to the requester.

A load balancing method applied to a caching party, the method comprising: obtaining a block request including target block location information sent by an agent, where the target block location information is at least one determined by the agent according to a data resource address One of the block location information; the data resource address is carried in the content request set as the request target file received by the agent, and the block request is set to request that the target file corresponds to the target block The block content of the location information; obtain the block content according to the target block location information; send the block content to the agent.

A load balancing device, which is applied to an agent, includes: a first obtaining unit, configured to obtain a content request carrying a data resource address of a target file; and a splitting unit, configured to split the content request into At least one segmentation request corresponding to the segmentation location information; wherein the segmentation location information is determined by the agent according to the data resource address; the distribution unit is configured to distribute the at least one segmentation request to the At least one cache party corresponding to the block location information in the at least one block request; the second acquiring unit is configured to obtain feedback from at least one cache party that has respectively received the corresponding block request in the at least one block request At least one piece of content; the absolute difference between the data lengths of different pieces of content is less than a predetermined threshold, and when there are multiple pieces of content of the same file, the first number of caches storing the multiple pieces of content is more A feedback unit, configured to feed back the at least one piece of content to the requester.

A load balancing device applied to a caching party, the device comprising: a third obtaining unit, configured to obtain a block request including target block location information sent by the agent, and the target block location information is based on the agent's One of at least one piece location information determined by the data resource address; the data resource address is carried in the content request set as the request target file received by the agent, and the piece request is set as the request for the target file The block content corresponding to the target block location information; the fourth obtaining unit is configured to obtain block content according to the target block location information; the sending unit is configured to send the block content to the agent .

An agent device includes: a first memory configured to store a first computer instruction set; a first processor configured to execute the instruction set stored in the first memory to implement the application as described in any one of the preceding items The agent's load balancing method.

A cache device includes: a magnetic disk, configured to store the data content of a file; a second memory, configured to store a second computer instruction set; a second processor, configured to execute the instruction set stored on the second memory, Implement the load balancing method applied to the cache side as described in any of the above.

A computer-readable storage medium in which a first set of computer instructions is stored, and when the first set of computer instructions is executed by a processor, the load applied to an agent as described in any one of the above is realized Balanced approach.

A computer-readable storage medium having a second set of computer instructions stored in the computer-readable storage medium, and when the second set of computer instructions is executed by a processor, the load applied to the cache side as described in any one of the preceding items is realized Balanced approach.

A service node includes a plurality of physical devices, the physical devices include a proxy and a cache; the proxy in the service node executes the load balancing method applied to the proxy as described in any one of the above and the At least one cache party in the service node interacts; the cache party in the service node interacts with the agent in the service node by executing the load balancing method applied to the cache party as described in any one of the above items.

The load balancing method, device, proxy device, cache device, and service node provided by the embodiments of the present disclosure store files in blocks, and specifically disperse the content of the files to different cache parties relatively evenly in blocks. When the requesting party When requesting a file, the content request of the file is mapped to at least one block request, and at least one block content that can be used to form a complete target file is obtained from at least one cache party based on the at least one block request, so as to achieve a response to the requester. Since the size of the block content corresponding to each block request is relatively uniform, the traffic of the requested file is allocated to different caching parties accordingly, so the intranet traffic of different caching parties will be relatively balanced, and will not be external. The use of network bandwidth has brought bottlenecks to improve the external network service capabilities of CDN and other networks.

Description of the drawings

FIG. 1 is a schematic flowchart of a load balancing method applied to a proxy provided by an embodiment of the present disclosure;

2 is a schematic diagram of another flow chart of a load balancing method applied to an agent provided by an embodiment of the present disclosure;

FIG. 3 is a schematic flowchart of a load balancing method applied to a caching party according to an embodiment of the present disclosure;

FIG. 4 is a flowchart of an implementation process of obtaining block content according to target block location information provided by an embodiment of the present disclosure;

FIG. 5 is a processing logic diagram of an application example of obtaining a target file from a CDN node provided by an embodiment of the present disclosure;

Fig. 6 is a schematic diagram of another process of a load balancing method applied to a caching party provided by an embodiment of the present disclosure;

FIG. 7 is a schematic structural diagram of a load balancing device applied to an agent provided by an embodiment of the present disclosure;

FIG. 8 is a schematic structural diagram of a load balancing device applied to a caching party according to an embodiment of the present disclosure;

FIG. 9 is a schematic diagram of another structure of a load balancing device applied to a caching party according to an embodiment of the present disclosure;

FIG. 10 is a schematic diagram of a structure of an agent device provided by an embodiment of the present disclosure;

FIG. 11 is a schematic diagram of a structure of a cache device provided by an embodiment of the present disclosure.

detailed description

The technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present disclosure.

The embodiments of the present disclosure disclose a load balancing method, device, proxy device, cache device, storage medium, and service node. By storing files in blocks, the content of the files is relatively evenly distributed to different caches in the form of blocks. , To achieve a balanced allocation of the traffic consumed by file content requests to different cache parties. In this way, for intranet application scenarios, such as the interaction scenario between the reverse proxy nginx and cache storage in the CDN node, the internal consumption of different cache parties The network bandwidth distribution will be relatively balanced, and it will not cause the uneven distribution of the internal network bandwidth of different cache parties due to the excessive/too small data volume of different file contents or the different access frequency as in related technologies, so that the internal network bandwidth usage distribution Will not bring bottlenecks to the use of external network bandwidth.

The following first describes the load balancing method provided by the embodiments of the present disclosure. Referring to FIG. 1, a schematic flow chart of the load balancing method when applied to a proxy is shown, which may include the following steps:

Step 101: Obtain a content request carrying the data resource address of the target file.

A CDN node is a cluster composed of multiple physical devices (servers). Each physical device of a CDN node includes a reverse proxy nginx and a cache storage, as the main body of execution of the method shown in Figure 1, and the proxy can be But it is not limited to the reverse proxy nginx on any physical device (such as a server) in the CDN node, which is responsible for the load balancing and request forwarding of the cache storage within the node, such as the CDN node.

Clients running on user electronic devices such as mobile phones and tablet computers, when they need to get the content of the target file from the network, if they need to get the page content of a webpage, they will send the distributed CDN node (the CDN network uses load balancing, Content distribution, scheduling and other functions enable users to obtain the required content from the nearest server in the network.) Send a content request for the target file. The content request at least carries the data resource address of the target file. In the public network environment, the content carried The data resource address is usually the uniform resource locator url of the file. For the reverse proxy nginx of the physical device in the CDN node, it will receive a content request carrying the data resource address of the target file, such as url.

Step 102: Split the content request into at least one segment request including corresponding segment location information; wherein the segment location information is determined by the agent according to the data resource address.

In the embodiments of the present disclosure, the files are stored in different caches in blocks (for the specific case where the file size does not exceed the set data length, the distributed storage is not performed, and the files are directly stored in one cache). When there are multiple pieces of content in the same file, the first number of cache parties storing the multiple pieces of content is multiple, and the absolute difference of the data length of different pieces of content is less than the set threshold to ensure different The distribution balance of bandwidth usage of the cache.

In one embodiment, a fixed data length can be set (of course, the length can also be changed according to the strategy), the data length is the smallest unit for realizing file block and distributed storage, and the file content is dispersed in the unit of the data length Storage, the stored content of each block does not exceed the set data length.

For the obtained content request for the complete target file, the agent splits it into at least one block request to achieve different block content acquisition of the target file.

The function of the block location information carried in the block request is to facilitate the cache storage of a physical device in the corresponding caching party, such as a CDN node, to index and read the block corresponding to the block request from its content collection content.

Step 103: Distribute the at least one block request to at least one cache party respectively corresponding to the block location information in the at least one block request.

Step 104: Obtain at least one piece of content fed back by at least one cache party that has respectively received a corresponding piece of the at least one piece request.

After receiving the block request, the cache party indexes and reads the corresponding block content from the stored content set according to the block location information carried in the block request, and feeds back the read block content to the sender The proxy for this block request, such as the reverse proxy nginx.

Among them, the absolute difference of the data length of the different block content is less than a predetermined threshold, so as to balance the load of each cache party as much as possible. In a case where there are multiple block contents of the same file, the first number of cache parties storing the plurality of block contents is multiple.

Step 105: Feed back the at least one piece of content to the requester.

After obtaining at least one piece of content corresponding to the at least one piece request fed back by at least one caching party, the agent may assemble the at least one piece of content to obtain the target file required by the requesting party, in an embodiment, And feedback the target file to the requesting party.

In this implementation, the agent can specifically assemble each block content in order to obtain the complete target file requested by the requesting party, and then feedback the complete target file obtained by the sequential assembly to the requesting party, for example, the CDN node receives it The reverse proxy nginx of the content request of the target file will feed back the complete target file obtained by sequential assembly to the requesting client.

Alternatively, in an embodiment, the agent may also directly send the at least one piece of content to the requester, so that the requester obtains the requested target file by assembling the at least one piece of content. When there are multiple pieces of at least one piece of content, in order to facilitate the requester to assemble multiple pieces of content in sequence, when sending multiple pieces of content to the requester, the agent can also carry each piece of content at the same time The corresponding block number.

In this embodiment, the file is stored in blocks and the content of the file is relatively evenly distributed to different caching parties in blocks. When the requesting party requests the file, the file content request is mapped to at least one block request based on At least one segment request obtains at least one segment content that can be used to form a complete target file from at least one cache party, so as to implement a response to the requester. Since the size of the block content corresponding to each block request is relatively uniform, the traffic of the requested file is allocated to different caching parties accordingly, so the intranet traffic of different caching parties will be relatively balanced, and will not be external. The use of network bandwidth has brought bottlenecks to improve the external network service capabilities of CDN and other networks.

Referring to FIG. 2, in one embodiment, the load balancing method applied to the agent of the present disclosure can be implemented through the following processing procedures:

Step 201: Obtain a content request carrying the data resource address of the target file.

In this embodiment, the data resource address of the target file includes the url of the target file. In addition, step 201 is the same as step 101 in the foregoing embodiment. For details, please refer to the related description or description of step 101 in the foregoing embodiment. In order to avoid repetition, a detailed description of step 201 is not provided here.

Step 202: Determine the file body data length of the target file.

In the back-end storage, a file includes two parts: the file header and the file body. The file header is used to record the relevant description information of the file, such as the data length of the data content in the file body; the file body is Include the actual data content of the file.

After the agent receives the content request carrying the data resource address of the target file, such as url, in order to facilitate the splitting of the content request into several block requests, the file header of the target file needs to be read from the cache to facilitate the extraction of the target from it. The actual data length of the data content in the file body of the file is convenient for subsequent processing. In order to simplify the explanation and description, the actual data length of the data content in the file body of the target file is simply recorded as "file body data length".

In this step, the process of determining the file body data length of the target file may include:

1) Determine the file header intranet address and file header location information of the target file according to the data resource address;

In one embodiment, after the file header location information is determined, a file header request including the file header location information may be generated, that is, the file header location information is carried in the file header request.

For obtaining the file header of the target file, if the proxy obtains the reverse proxy nginx of the content request, on the one hand, it generates the intranet address of the file header according to the url carried in the content request for locating, routing and storing the file header. The caching side of the file header of the target file, on the other hand, generates file header location information according to the url, which is used as an index to implement indexing and hitting of the file header of the target file on the located caching side.

As an optional implementation manner, the process of generating the internal network address of the file header of the target file includes: splicing the url and the number of the file header carried in the content request of the target file to obtain splicing information; and performing a consistent hash operation on the splicing information , Get the file header intranet address of the target file.

For a file, the number of the file header is set to 0 in this embodiment, and the number of each block in the file body increases sequentially from 1 until the number of the nth block of the file body is n, where n It is a natural number greater than 1. It is easy to understand that this numbering rule is not unique, and there can be many other numbering rules to achieve similar purposes. For example, the number of the file header is set to 1, and the number of each block in the file body increases sequentially from 2. ; Or, the number of the file header is set to a, and the number of each block in the file body increases sequentially from b, etc. Of course, the file header/block number can also be a character composed of multiple numbers and/or letters , No longer show them one by one.

Since the url and the number of the file header/block are spliced together, the difference in the splicing information corresponding to the file header and different blocks is small, and it is easy to map the file header and different blocks to a smaller number by performing a consistent hash operation on it. The content of the same file can’t be evenly distributed to more cache parties, such as evenly distributed to more cache storage in the CDN node. After the inventor’s research and verification, the url and number (file The splicing information of the header/block number) is directly hashed consistently, and each hash result is normally distributed, resulting in a large number of blocks being scattered to a few cache parties.

In order to improve this situation, in this embodiment, after the url and the file header/block number are spliced, it is proposed to perform information digest calculation on the splicing result, and perform consistent hashing on the information digest calculated from the information digest to This can achieve the purpose of evenly distributing the file header and different blocks of the same file to more cache parties.

Therefore, as another optional implementation manner, the process of generating the internal network address of the file header of the target file may also be: splicing the url and the number of the file header carried in the content request of the target file to obtain the splicing information; Perform digest calculation on the spliced information; perform consistent hashing on the calculated information digest to obtain the file header intranet address of the target file;

The digest algorithm used can be, but is not limited to, md5 (message digest 5 algorithm), md4 (message digest 4 algorithm), etc.

In addition, in one embodiment, the process of generating the file header location information of the target file may include: combining the url of the target file with at least one of the number of the file header and the corresponding position of the file header in the target file. Assemble to obtain the file header position information of the file header.

That is, the file header location information of the target file must include at least the url of the target file, and the number of the file header (such as "0") and the corresponding position of the file header in the target file (such as the file header in the file Corresponding range: at least one of 0k～128k).

2). Send the file header location information to the cache party indicated by the file header intranet address.

In the implementation, specifically, the file header request carrying the file header location information can be sent to the cache party indicated by the file header intranet address; in addition, the file header intranet address can also be carried in the file as part of the file header location information. There is no restriction on this in the header request.

3) Obtain the file body data length of the target file determined and fed back by the caching party according to the file header location information.

In an embodiment, the file header determined and fed back by the cache party according to the file header location information may be obtained, and the file header carries the file body data length of the target file.

The cache party indicated by the internal network address of the file header, after receiving the file header request of the target file, indexes the corresponding file header from the data set cached by itself according to the file header location information carried in the request, such as url and file header number. , And read it.

The agent receives the file header of the target file read and fed back by the caching party, and reads the file body data length of the target file from the received file header.

Step 203: Determine a second amount of block content included in the file body of the target file according to the predetermined data length and the file body data length of the target file.

The predetermined data length, for example, can be 128k or other data lengths different from 128k.

Now take 128k as an example. Assuming that the file body data length of the target file is h, the second quantity of the block content included in the file body of the target file can be determined as

Step 204: Determine a second number of block location information according to the data resource address, and generate a second number of block requests each including different block location information.

In the implementation, since the cache party that stores the corresponding block content of the target file needs to be located and routed first, the corresponding block content can be obtained in the located cache party according to the block location information of the block content. Therefore, in addition to determining the number of The block location information of the second quantity may further include the following processing: determining the number of block intranet addresses of the second quantity according to the data resource address.

The function of the divided intranet address of the divided content is to locate and route the cache of the corresponding divided content of the target file.

The function of the block location information of the block content is to be used as an index to implement indexing and hitting of the corresponding block content of the target file on the located cache side. In one embodiment, in this embodiment, the caching party, such as the cache storage on the physical device in the CDN node, uses the key-value method to store the file header and block content of the file. Location information (such as url+number+range corresponding to the number, etc.) is used as the "key" of the data content (that is, the "value" of the file header/block content). Based on the storage method, the location of the file header or block content is obtained After the information, the location information can be used as the "key" to obtain its corresponding "value".

Similar to the generation process of the file header intranet address, as an optional implementation, the process of generating the block intranet address may include: splicing the url of the target file and the block number of the block content (such as 1 or 2 or 3). ...Or n, etc., where n is a natural number greater than 1) to obtain the splicing information; perform a consistent hash operation on the splicing information to obtain the block intranet address of the block content.

In order to evenly distribute the content of different blocks of the same file to more cache parties and achieve a higher degree of load balancing, as an alternative implementation, the block intranet address of the block content can also be generated through the following process: Combine the url of the target file and the block number of the block content to obtain the spliced information; use a predetermined digest algorithm to perform digest calculation on the spliced information; perform a consistent hash operation on the calculated information summary to obtain the block content of the block Intranet address.

The process of generating the block location information of the block content may include: performing at least one of the url of the target file, the block number of the block content, and the content location of the block content in the target file. Assemble, obtain the block location information of the block content.

That is, the block location information of the block content of the target file must at least include the url of the target file, and the block number of the block content and the corresponding content position of the block content in the target file (such as number x Corresponding content in the file body: at least one of 128k～256k).

Step 205: Distribute the at least one block request to the cache party indicated by the block intranet address corresponding to the corresponding block request.

Wherein, after receiving the block request, the cache party obtains and feeds back the block content indicated by the block location information in the block request to the agent. For example, the block location information is used as the "key", and the corresponding "value" is obtained, and feedback is provided to the agent.

Step 206: Obtain at least one piece of content fed back by at least one cache party that has respectively received a corresponding piece of the at least one piece request.

The absolute difference of the data lengths of different block contents is less than the set threshold. In the case where there are multiple block contents of the same file, the first number of cache parties storing the plurality of block contents is multiple.

Step 207: Feed back the at least one piece of content to the requesting party.

In an embodiment, after obtaining at least one piece of content corresponding to the at least one piece request fed back by at least one caching party, the agent may assemble the at least one piece of content to obtain the target file required by the requester, and Feed back the target file to the requesting party.

In this implementation, the agent can specifically assemble the content of each segment based on the number corresponding to the content of each segment, such as 1, 2, 3...n, where n is a natural number greater than 1, and assemble the content of each segment in order to obtain the completeness requested by the requesting party. The target file, and the complete target file assembled in sequence is fed back to the requester, such as the reverse proxy nginx in the CDN node that receives the content request of the target file, and the complete target file assembled in sequence is fed back to the requesting client Wait.

In this embodiment, the file is stored in blocks, the file content is relatively evenly distributed to different cache parties, and the content request of the file is mapped to at least one block request, so that the traffic of the requested file can be distributed to different caches in a balanced manner. The caching side, as far as possible, enables the use of internal network bandwidth to be evenly distributed, which improves the external network service capabilities of CDN and other networks. In addition, this embodiment also performs digest operations on the splicing information of url and file header/block number information, and performs consistent hashing on the obtained information digest, so that the content of different blocks of the same file can be stored in more caches. The more balanced distribution of the internal network can further reduce the limitation of the internal network bandwidth usage and the external network bandwidth utilization rate.

Next, the load balancing method applied to the caching side of the embodiment of the present disclosure will be described. Refer to FIG. 3, which shows a schematic flowchart of the load balancing method when applied to the caching side, including the following steps:

Step 301: Obtain a block request sent by the agent that includes the location information of the target block.

The target segment location information is one of at least one segment location information determined by the agent according to the data resource address; the data resource address is carried in the content request for the target file received by the agent, and The block request is used to request the block content corresponding to the target block location information in the target file.

In this step, specifically, the block request sent by the agent based on the corresponding block intranet address can be obtained.

Wherein, the corresponding block intranet address is one of the second number of block intranet addresses determined by the agent according to the data resource address; the process of determining the second number of block intranet addresses includes: The agent determines the block number of each block content of the second quantity; generates the block intranet address of the block content according to the uniform resource locator and the block number of the block content, and obtains the block number of the second number Block intranet address. For a more specific determination process of the second number of block intranet addresses, please refer to the relevant description above, which will not be described in detail here.

Combining with the previous description, it can be determined that the target block location information in the block request may include the url of the complete target file corresponding to the block content, as well as the number of the block content and the content corresponding to the number in the target file. At least one of location (such as content range).

Step 302: Obtain the block content according to the target block location information.

As shown in Figure 4 (where "Y" in the figure represents "Yes" and "N" represents "No"), the process of obtaining the block content according to the target block location information can be implemented as follows:

Step 401: Determine whether the block content indicated by the target block location information exists locally on the caching party.

Based on the operation and maintenance characteristics of the CDN network, the data content of the file is not cached in the CDN network when the data content of the file is requested for the first time or is cleared based on the clearing strategy, which will cause the file data content to hit the CDN network If it fails, correspondingly, the CDN network needs to pull (ie obtain) the data from the data source site of the file and cache the data (combined with the CDN network protocol and operation and maintenance characteristics, in the embodiment of the present disclosure, storing the data content on the caching side is equivalent to storing the data on the caching side Cache data content).

In view of this feature, in this embodiment, the cache party, such as the cache storage on a physical device in the CDN node, after obtaining the block request, first uses the target block location information carried in the request as an index to determine the cache party’s local Whether there is the block content indicated by the target block location information.

Step 402: In a case where it is determined that it exists, obtain the segmented content from the local.

If the result of the determination indicates that it exists, the caching party can use the target block location information carried in the request as an index, index and read the corresponding block content from the stored content set.

Step 403: In a case where it is determined that it does not exist, perform predetermined back-to-source processing, and obtain the block content corresponding to the block request through the back-to-source processing.

If the result of the determination indicates that it does not exist, the back-to-origin process is executed to obtain the requested segmented content.

The process of back-to-origin processing includes:

1) Determine the back-to-origin caching party corresponding to the URL in the target block location information;

Among them, the cache party that receives the block request and fails the local hit, extracts the url from the block location information of the block request, which is the url of the target file corresponding to the block, and performs consistency on the extracted url. It is hoped to calculate the intranet address of the target file; then the cache party indicated by the intranet address of the target file is used as the back-to-source cache party.

The back-to-origin caching party is responsible for pulling and retrieving the target file from the source site when the target file is not stored in each caching party (for example, the target file is requested for the first time or has been cleared from each caching party based on a clearing strategy). Store the data content of the target file. Subsequently, when the stored target file meets the clearing condition, the back-to-source caching party clears the data of the target file from its own content collection.

2) Obtain the block content corresponding to the block request from the back-to-source caching party, and cache the obtained block content in the corresponding location of the cache party indicated by the target block location information.

In implementation, you can also use the block location information in the block request (such as url + block number + content range corresponding to the number) as the "key" to obtain the block content corresponding to the block request from the back-to-origin cache, and Cache the obtained block content at the corresponding location of the cache party indicated by the block intranet address/block location information, that is, cache the read block content at the corresponding location of the cache party that failed the local hit; then, When the block content of the target file is requested again, since the cache party pointed to by the block intranet address already stores a copy of the block content, a local hit can be successfully performed.

In the same way, the cache party that receives other block requests for the target file will also converge to the back-to-origin cache party due to its own hit failure, read the required block content from it, and store a copy of the block content in itself In this way, two copies of the content of a file are stored in the node. One copy is stored in the back-to-origin caching side used to converge back to the source, and the other is divided into different caches (non-back-to-origin) in the form of blocks. In the caching side), during the coexistence of the two storages, the agent will request the contents of each block of the target file in blocks based on the routing strategy to achieve load balancing, and for the complete data of the file stored by the back-to-source caching party The content will be cleared based on the clearing strategy when the clearing condition is met, and the clearing condition may be, but not limited to, the unrequested duration reaching the set duration, etc.

Step 303: Send the segmented content to the agent.

In one embodiment, after obtaining the divided content, the cache party may send the obtained divided content to the agent that issued the request for dividing.

It should be noted that for the target file that is not stored by the cache party (target file that is not stored in the CDN node), the file header will also fail locally when the file header is requested, and the cache party that receives the file header request and fails the local hit is based on The same principle as above converges to the back-to-source caching side. In this case, the target file does not exist in the back-to-source caching side. Accordingly, the data content of the target file needs to be pulled from the source site, and the back-to-source caching side completes the data pull And after storing the complete data content, the cache party that receives the file header request obtains the required file header from the back-to-source cache party and stores a copy in itself. Later, when it continues to distribute block requests to different cache parties to obtain the target file When the content of the block, each cache party that receives the block request does not store the required block content, and thus converges to the back-to-source cache party, and reads the required block from the complete target file stored by the back-to-source cache party The content is stored separately.

As shown in Figure 5, the processing logic of an application instance that obtains the target file from the CDN node is provided. After obtaining the url request, the reverse proxy nginx performs the following processing to obtain the target content:

Step 1. Send a file header request carrying url and file header range to cache1 indicated by the file header internal network address;

Step 2. After cache1 receives the file header request, if it needs to return to the source, hash the url consistency to cache2 (if it does not need to return to the source, directly store the response nginx based on cache1 itself);

Step 3. Cache2 pulls and caches the content of the file from the origin site, and returns the file header to cache1;

Step 4. Cache1 responds to nginx with the file header to inform nginx of the data length of the file body; at the same time, cache1 caches the file header;

Step 5. In the process of distributing the block request, nginx sends a block request containing the url and block range generated according to the data length of the file body to the cache3 corresponding to the block intranet address;

Step 6. If you need to return to the source, cache3 will hash the URL consistency to cache2 (if you don't need to return to the source, it will directly store the response nginx based on cache3 itself);

Step 7. Cache2 returns the corresponding block content to cache3;

Step 8. Cache3 responds to nginx with the segmented content, and cache3 caches the segmented content;

Steps 9-12 and steps 9-12 are similar to steps 5-8, and steps 9-12 and 5-8 can be executed in parallel, so they will not be repeated;

Step 13. Nginx assembles the block content returned by each cache to obtain the complete target file and feed it back to the client.

In this embodiment, in the process of converging back to the source, the back-to-source cache party pulls the complete data content of the file from the source station, and each cache party that receives the block request reads and stores the required data from the back-to-source cache party. The block content can enable the block request and routing of the target file data content to be realized when the target file is requested again, and the internal network load balance can be realized as quickly as possible. In addition, a back-to-source cache party is used to converge back The source, without expanding the amount of return to the source to a higher degree, will not bring a higher additional burden to the CDN node.

In practical applications, based on the load balancing strategy, a caching party such as the cache storage on a physical device in a CDN node may cache one or more block contents of the same file, or cache block contents and file headers of the same file . In this way, referring to FIG. 6, the load balancing method applied to the cache side may further include the following processing before step 301:

Step 301': If the file header position information of the target file sent by the agent is obtained, the file body data length of the target file is obtained according to the file header position information, and the file body data length is sent to the agent.

The file header location information can be carried in the file header request, and the caching party can specifically obtain the file header request sent by the agent, and extract the file header location information from it.

The length of the file body data can be carried in the file header.

The caching party can specifically use the file header location information (such as url+file header number) as the index (key), obtain the corresponding file header (value) from its stored content collection, and feed it back to the requesting file header The agent is used by the agent to extract the length of the file body data from the obtained file header. For example, the agent determines the second quantity of the block content of the target file according to the file body data length and the set data length, and then the target file The content request of the file is split into a second number of multiple block requests, etc. For details, please refer to the previous description, which will not be repeated here.

Corresponding to the above-mentioned load balancing method, an embodiment of the present disclosure also discloses a load balancing device. As shown in FIG. 7, when applied to an agent, the load balancing device includes:

The first obtaining unit 701 is configured to obtain a content request carrying a data resource address of the target file;

The splitting unit 702 is configured to split the content request into at least one segment request including corresponding segment location information; wherein the segment location information is determined by the agent according to the data resource address;

The distributing unit 703 is configured to distribute the at least one block request to at least one cache party respectively corresponding to the block location information in the at least one block request;

The second acquiring unit 704 is configured to acquire at least one piece of content fed back by at least one cache party that has respectively received the corresponding piece of the at least one piece of request; the absolute difference of the data length of the different piece of content is less than the set value Threshold, in the case that there are multiple pieces of content in the same file, the first number of cache parties storing the multiple pieces of content is multiple;

The feedback unit 705 is configured to feed back the at least one piece of content to the requester.

In an optional implementation manner of the embodiment of the present disclosure, the splitting unit 702 is specifically configured as follows:

Determining the file body data length of the target file;

Determine the second quantity of the block content included in the file body of the target file according to the predetermined data length and the data length of the file body;

According to the data resource address, a second number of block intranet addresses and block location information are determined, and a second number of block requests each including different block location information are generated.

In an optional implementation manner of the embodiment of the present disclosure, the splitting unit 702 determining the file body data length of the target file includes:

Determining the file header intranet address and file header location information of the target file according to the data resource address;

Sending the file header location information to the caching party indicated by the file header's intranet address;

Obtain the file body data length of the target file determined and fed back by the caching party according to the file header location information.

In an optional implementation manner of the embodiment of the present disclosure, the splitting unit 702 is further configured to: determine a second number of block intranet addresses according to the data resource address;

The data resource address includes the uniform resource locator of the target file;

The splitting unit 702 determines, according to the data resource address, a second number of block intranet addresses or block location information, including:

Determine the block number of each block content of the second quantity;

The block intranet address or block location information of the block content is generated according to the uniform resource locator and the block number of the block content, and the block intranet address or block location information of the second number is obtained.

In an optional implementation manner of the embodiment of the present disclosure, the splitting unit 702 generates the block intranet address or block location information of the block content according to the uniform resource locator and the block number of the block content ,include:

Splicing the uniform resource locator and the block number of the block content to obtain splicing information;

Calculating an information abstract corresponding to the spliced information by using a predetermined abstract algorithm;

Perform a consistent hash operation on the information digest to obtain the block intranet address of the block content;

Assemble the uniform resource locator with at least one of the segment number of the segmented content and the content location of the segmented content in the target file to obtain the segmented location of the segmented content information.

In an optional implementation manner of the embodiment of the present disclosure, the distribution unit 703 is specifically configured as follows:

Distribute the at least one block request to the cache party indicated by the block intranet address corresponding to the corresponding block request; wherein, after receiving the block request, the cache party obtains block location information in the block request The indicated block content is fed back to the agent.

In an optional implementation manner of the embodiment of the present disclosure, the feedback unit 705 is specifically configured as follows:

Assemble the at least one piece of content to obtain the target file, and send the target file to the requesting party; or,

Send the at least one piece of content to the requester, so that the requester obtains the target file by assembling the at least one piece of content.

For the load balancing device applied to the agent disclosed in the embodiments of the present disclosure, since it corresponds to the load balancing method applied to the agent disclosed in the respective embodiments above, the description is relatively simple, and the related similarities are Please refer to the description of the load balancing method applied to the proxy in the respective embodiments above, and the details will not be described here.

As shown in Figure 8, in the case of being applied to the caching side, the load balancing device includes:

The third obtaining unit 801 is configured to obtain a block request including target block location information sent by the agent, where the target block location information is one of at least one block location information determined by the agent according to the data resource address The data resource address is carried in a content request received by the agent and set to request a target file, and the block request is set to request the block content corresponding to the target block location information in the target file;

The fourth obtaining unit 802 is configured to obtain block content according to the target block location information;

The sending unit 803 is configured to send the divided content to the agent.

In an optional implementation manner of the embodiment of the present disclosure, the fourth acquiring unit 802 is specifically configured as follows:

Determining whether the block content indicated by the target block location information exists locally on the caching party;

In the case where it is determined to exist, obtain the segmented content from the local;

In a case where it is determined that it does not exist, a predetermined back-to-source process is performed, and the block content corresponding to the block request is obtained through the back-to-source process.

In an optional implementation manner of the embodiment of the present disclosure, the fourth acquiring unit 802 performs predetermined back-to-origin processing, including:

Determining the back-to-origin cache party corresponding to the data resource address in the target block location information;

Obtain the block content corresponding to the block request from the back-to-source caching party, and cache the block content in a corresponding location of the cache party indicated by the target block location information.

In an optional implementation manner of the embodiment of the present disclosure, the data resource address includes a uniform resource locator of the target file;

The fourth obtaining unit 802 determining the back-to-source cache party corresponding to the data resource address in the target block location information includes:

Performing a consistent hash operation on the uniform resource locator of the target file to obtain the intranet address of the target file;

Determining the cache party indicated by the intranet address of the target file as the back-to-source cache party;

In an optional implementation manner of the embodiment of the present disclosure, the back-to-source caching party obtains and caches the target file's information from the source site of the target file when the target file is not stored in each caching party. Data content; when the cached target file meets the clearing condition, clear the target file from the back-to-source cache party.

In an optional implementation manner of the embodiment of the present disclosure, referring to FIG. 9, the device may further include:

The fifth obtaining unit 804 is configured to obtain the file header request including the file header position information of the target file sent by the agent before obtaining the file header request including the file header position information of the target file sent by the agent, according to the file header The location information acquires the file body data length of the target file, and sends the file body data length to the agent.

In an optional implementation manner of the embodiment of the present disclosure, the third acquiring unit 801 is specifically configured as follows:

Obtain the block request sent by the agent based on the corresponding block intranet address;

Wherein, the corresponding block intranet address is one of the second number of block intranet addresses determined by the agent according to the data resource address; the process of determining the second number of block intranet addresses includes : The agent determines the block number of each block content of the second quantity; generates the block intranet address of the block content according to the uniform resource locator and the block number of the block content, and the obtained number is the second The number of block intranet addresses.

For the load balancing device applied to the caching side disclosed in the embodiments of the present disclosure, since it corresponds to the load balancing method applied to the caching side disclosed in the respective embodiments above, the description is relatively simple, and the related similarities are Please refer to the description of the load balancing method applied to the caching party in the respective embodiments above, and the details will not be described here.

The embodiment of the present disclosure also discloses a proxy device, which can be, but is not limited to, a reverse proxy nginx on any physical device (such as a server) in a CDN network node, as shown in FIG. 10, a schematic structural diagram of the proxy device , The agent equipment at least includes:

The first memory 1001 is configured to store the first computer instruction set;

The first set of computer instructions can be implemented in the form of a computer program.

The first memory 1001 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

The first processor 1002 is configured to implement any load balancing method applied to the agent as described above by executing the instruction set stored in the first memory.

The first processor 1002 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), and a field programmable Gate array (FPGA) or other programmable logic devices, etc.

In addition, the proxy device can also include components such as a communication interface and a communication bus. The first memory, the first processor and the communication interface communicate with each other through the communication bus.

The communication interface is set to communicate between the proxy device and other devices (such as other physical devices in the CDN node). The communication bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, and a control bus.

In this embodiment, the first processor in the proxy device executes the first computer instruction set stored in the first memory to implement block storage of the file and specifically distribute the file content to different cache parties relatively evenly. When the requester requests a file, the content request of the file is mapped to at least one block request, and at least one block content that can be used to form a complete target file is obtained from at least one cache party based on the at least one block request, so as to achieve the request to the requester. the response to. Since the size of the block content corresponding to each block request is relatively uniform, the traffic of the requested file is allocated to different caching parties accordingly, so the intranet traffic of different caching parties will be relatively balanced, and will not be external. The use of network bandwidth has brought bottlenecks to improve the external network service capabilities of CDN and other networks.

The embodiment of the present disclosure also discloses a cache device. The cache device can be, but is not limited to, cache storage on any physical device (such as a server) in the CDN network node. The cache device includes at least:

Disk 1101, set to store the data content of the file;

The second memory 1102 is configured to store a second computer instruction set;

The second set of computer instructions can be implemented in the form of a computer program.

The second processor 1103 is configured to implement any load balancing method applied to the cache side as described above by executing the instruction set stored in the second memory.

In the cache device of this embodiment, the second processor 1103 executes the second set of computer instructions in the second memory 1102 to implement block storage of files and specifically disperses the content of the files to different cache parties relatively evenly. At the same time, it realizes the block request and routing of files, which can make the use of internal network bandwidth evenly distributed, will not bring bottleneck to the use of external network bandwidth, and improve the service capability of the external network.

The embodiment of the present disclosure also discloses a computer-readable storage medium, the computer-readable storage medium stores a first set of computer instructions, and when the first set of computer instructions is executed by a processor, any one of the above is applied to the agent. Load balancing method.

Correspondingly, the embodiment of the present disclosure also discloses another computer-readable storage medium, and the second computer instruction set is stored in the second computer instruction set when the second computer instruction set is executed by the processor. Realize any of the above load balancing methods applied to the caching side.

The instructions stored in the above-mentioned two computer-readable storage media can realize the block storage of the file and the specific content of the file is relatively evenly distributed to different cache parties when running, and can realize the block division of the file at the same time. Requests and routing can make the use of internal network bandwidth evenly distributed, will not bring bottleneck to the use of external network bandwidth, and improve the service capability of the external network.

The embodiment of the present disclosure also discloses a service node, which includes a plurality of physical devices, and the physical devices include a proxy party and a cache party;

The agent in the service node interacts with at least one cache party in the service node by executing any load balancing method applied to the agent as described above;

The cache party in the service node interacts with the agent in the service node by executing any load balancing method applied to the cache party as described above.

The service node can be a CDN node. Correspondingly, the aforementioned agent can be a reverse proxy nginx on a physical device (such as a server) in the CDN node, and the cache party can be a cache on a physical device (such as a server) in the CDN node. storage.

Regarding the load balancing-based communication interaction process between the proxy party and the cache party, please refer to the introduction of the previous related embodiments for details, which will not be described in detail here.

It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. For the same and similar parts between the various embodiments, refer to each other. Can.

For the convenience of description, when describing the above system or device, the functions are divided into various modules or units to be described separately. Of course, when implementing the present disclosure, the functions of each unit may be implemented in the same one or more software and/or hardware.

It can be seen from the description of the above embodiments that those skilled in the art can clearly understand that the present disclosure can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution of the present disclosure can be embodied in the form of a software product in essence or a part that contributes to the related technology. The computer software product can be stored in a storage medium, such as ROM/RAM, magnetic disk, An optical disc, etc., includes a number of instructions to enable a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments of the present disclosure.

Finally, it should be noted that in this article, relational terms such as first, second, third, and fourth are only used to distinguish one entity or operation from another entity or operation. Any such actual relationship or sequence between these entities or operations must be required or implied. Moreover, the terms "including", "including" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device that includes a series of elements includes not only those elements, but also those that are not explicitly listed Other elements of, or also include elements inherent to this process, method, article or equipment. If there are no more restrictions, the element defined by the sentence "including a..." does not exclude the existence of other identical elements in the process, method, article, or equipment that includes the element.

The above are only the preferred embodiments of the present disclosure. It should be pointed out that for those of ordinary skill in the art, without departing from the principles of the present disclosure, several improvements and modifications can be made, and these improvements and modifications are also It should be regarded as the protection scope of this disclosure.

Industrial applicability

In related technologies, the internal network bandwidth consumed by a physical device of the CDN network is the same as the external network bandwidth. The unbalanced distribution of the internal network bandwidth of the CND node will limit the utilization of the external network bandwidth and bring about the use of external network bandwidth. The bottleneck has caused the maximum bandwidth consumed by the external network to be unable to reach the maximum saturation (different physical devices are configured with the same maximum available external network bandwidth), which correspondingly reduces the external network service capabilities of networks such as CDN.

In view of the problems in the related technology, the embodiments of the present disclosure store files in blocks, and specifically disperse the content of the files to different cache parties relatively evenly in blocks. When the requesting party requests the file, the content of the file is requested The mapping is at least one block request, and at least one block content that can be used to form a complete target file is obtained from at least one cache party based on the at least one block request, so as to implement a response to the requester. Since the size of the block content corresponding to each block request is relatively uniform, the traffic of the requested file is allocated to different caching parties accordingly, so the intranet traffic of different caching parties will be relatively balanced, and will not be external. The use of network bandwidth has brought bottlenecks to improve the external network service capabilities of CDN and other networks.

Claims

A load balancing method applied to an agent, and the method includes:

Get the content request carrying the data resource address of the target file;

Splitting the content request into at least one segment request including corresponding segment location information; wherein the segment location information is determined by the agent according to the data resource address;

Distributing the at least one block request to at least one cache party respectively corresponding to the block location information in the at least one block request;

Obtain at least one piece of content fed back by at least one cache party that has respectively received the corresponding piece request in the at least one piece of request; the absolute difference of the data length of different piece of content is less than a predetermined threshold, and the piece of content of the same file In the case of multiple pieces, the first number of cache parties storing multiple pieces of content is multiple;

The at least one piece of content is fed back to the requesting party.
The method according to claim 1, wherein the splitting the content request into at least one segment request including corresponding segment location information comprises:

Determining the file body data length of the target file;

Determine the second quantity of the block content included in the file body of the target file according to the predetermined data length and the data length of the file body;

According to the data resource address, a second number of block location information is determined, and a second number of block requests each including different block location information are generated.
The method according to claim 2, wherein said determining the file body data length of the target file comprises:

Determining the file header intranet address and file header location information of the target file according to the data resource address;

Sending the file header location information to the caching party indicated by the file header's intranet address;

Obtain the file body data length of the target file determined and fed back by the caching party according to the file header location information.
The method according to claim 2 or 3, further comprising: determining a second number of block intranet addresses according to the data resource address;

The data resource address includes the uniform resource locator of the target file;

Determining the second number of block intranet addresses or block location information according to the data resource address includes:

Determine the block number of each block content of the second quantity;

The block intranet address or block location information of the block content is generated according to the uniform resource locator and the block number of the block content, and the block intranet address or block location information of the second number is obtained.
The method according to claim 4, wherein the generating the block intranet address or block location information of the block content according to the uniform resource locator and the block number of the block content comprises:

Splicing the uniform resource locator and the block number of the block content to obtain splicing information;

Calculating an information abstract corresponding to the spliced information by using a predetermined abstract algorithm;

Perform a consistent hash operation on the information digest to obtain the block intranet address of the block content;

Assemble the uniform resource locator with at least one of the segment number of the segmented content and the content location of the segmented content in the target file to obtain the segmented location of the segmented content information.
The method according to claim 4 or 5, wherein the distributing the at least one block request to at least one cache party respectively corresponding to the block location information in the at least one block request comprises:

Distributing the at least one block request to the cache party indicated by the block intranet address corresponding to the corresponding block request;

Wherein, after receiving the block request, the cache party obtains the block content indicated by the block location information in the block request and feeds it back to the agent.
The method according to any one of claims 1 to 6, wherein the feeding back the at least one piece of content to the requesting party comprises:

Assemble the at least one piece of content to obtain the target file, and send the target file to the requesting party; or,

Send the at least one piece of content to the requester, so that the requester obtains the target file by assembling the at least one piece of content.
A load balancing method applied to a caching party, the method comprising:

Obtain a segmentation request including target segmentation location information sent by the agent, where the target segmentation location information is one segmentation location information of at least one segmentation location information determined by the agent according to the data resource address; the data resource The address is carried in a content request set to request a target file received by the agent, and the block request is set to request the block content corresponding to the target block location information in the target file;

Acquiring block content according to the target block location information;

Sending the segmented content to the agent.
The method according to claim 8, wherein said obtaining the block content according to the location information of the target block comprises:

Determining whether the block content indicated by the target block location information exists locally on the caching party;

In the case where it is determined to exist, obtain the segmented content from the local;

In a case where it is determined that it does not exist, a predetermined back-to-source process is executed, and the block content corresponding to the block request is obtained through the back-to-source process.
The method according to claim 9, wherein said executing predetermined back-to-origin processing comprises:

Determining the back-to-origin cache party corresponding to the data resource address in the target block location information;

Obtain the block content corresponding to the block request from the back-to-source caching party, and cache the block content in a corresponding location of the cache party indicated by the target block location information.
The method according to claim 10, wherein the data resource address includes a uniform resource locator of the target file;

The determining the back-to-source cache party corresponding to the data resource address in the target block location information includes:

Performing a consistent hash operation on the uniform resource locator of the target file to obtain the intranet address of the target file;

Determine the cache party indicated by the intranet address of the target file as the back-to-source cache party.
The method according to claim 10 or 11, wherein the back-to-origin caching party obtains and caches the target file's data from the source site of the target file when the target file is not stored in each caching party. Data content; when the cached target file meets the clearing condition, clear the target file from the back-to-source cache party.
The method according to any one of claims 8-12, wherein, before acquiring the block request including the location information of the target block sent by the agent, the method further comprises:

If the file header position information of the target file sent by the agent is obtained, the file body data length of the target file is obtained according to the file header position information, and the file body data length is sent to the agent.
The method according to any one of claims 8-13, wherein the block request including the location information of the target block sent by the obtaining agent comprises:

Obtain the block request sent by the agent based on the corresponding block intranet address;

Wherein, the corresponding block intranet address is one of the second number of block intranet addresses determined by the agent according to the data resource address; the data resource address includes the address of the target file Uniform resource locator; the process of determining the intranet address of the second number of blocks includes: the agent determines the block number of each block content of the second number; according to the uniform resource locator and the block content The block number generates the block intranet address of the block content, and the block intranet address with the second number is obtained.
A load balancing device applied to an agent, the device comprising:

The first obtaining unit is configured to obtain a content request carrying a data resource address of the target file;

A splitting unit, configured to split the content request into at least one segment request including corresponding segment location information; wherein the segment location information is determined by the agent according to the data resource address;

A distributing unit, configured to distribute the at least one block request to at least one cache party corresponding to the block location information in the at least one block request;

The second obtaining unit is configured to obtain at least one piece of content fed back by at least one cache party that has respectively received the corresponding piece request in the at least one piece of piece request; the absolute difference of the data length of the different piece of content is less than a predetermined threshold, In the case where there are multiple pieces of content in the same file, the first number of cache parties storing multiple pieces of content is multiple;

The feedback unit is configured to feed back the at least one piece of content to the requester.
A load balancing device applied to a caching party, the device comprising:

The third obtaining unit is configured to obtain a block request including target block location information sent by the agent, where the target block location information is one block of at least one block location information determined by the agent according to the data resource address Location information; the data resource address is carried in the content request received by the agent and set to request the target file, and the block request is set to request the block content corresponding to the target block location information in the target file ；

The fourth obtaining unit is configured to obtain the block content according to the target block location information;

The sending unit is configured to send the segmented content to the agent.
An agent device, including:

The first memory is configured to store the first computer instruction set;

The first processor is configured to implement the load balancing method according to any one of claims 1-7 by executing the instruction set stored in the first memory.
A cache device, including:

Disk, set to store the data content of the file;

The second memory is set to store the second computer instruction set;

The second processor is configured to implement the load balancing method according to any one of claims 8-14 by executing the instruction set stored in the second memory.
A computer-readable storage medium, the computer-readable storage medium stores a first set of computer instructions, and when the first set of computer instructions is executed by a processor, the load according to any one of claims 1-7 is realized Balanced approach.
A computer-readable storage medium having a second set of computer instructions stored in the computer-readable storage medium, and when the second set of computer instructions is executed by a processor, the load according to any one of claims 8-14 is realized Balanced approach.
A service node includes multiple physical devices, and the physical devices include a proxy party and a cache party;

The agent in the service node interacts with at least one caching party in the service node by executing the method according to any one of claims 1-7;

The cache party in the service node interacts with the agent party in the service node by executing the method according to any one of claims 8-14.