RU2722464C1 - Method of expanding a cdn network using a peer-to-peer network - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: invention is intended for combining CDN and peer-to-peer networks. In the method, the peer-to-peer network control subsystem generates a list of nodes of a peer-to-peer network of a given size, addresses of the CDN nodes and a list of nodes of the peer-to-peer network are sent to the client. Then the client sends to the nodes from the list a request to load the next segment of viewed Live content, to which they respond when there is a segment in their cache and free resources. After the response, resources are reserved, then the client sends a request for loading data to the random node from among the responders and increases the error counter for non-responding nodes; if necessary, data is re-requested at the CDN node, when the error counter reaches the threshold, the node is stopped to be used, and if the number of nodes in the list is less than a given threshold, the subsystem for controlling the peer-to-peer network requests the missing number of nodes, if the number of nodes in the list is greater than a given threshold, then the required number of nodes is selected randomly, without considering nodes whose error counter exceeds the predetermined threshold and nodes from the list transmitted by the client, if the node has insufficient resources for processing requests, priority is given to clients, which more often load data, from the overflowed cache removing the oldest by time addition of segments.

EFFECT: technical result consists in improvement of efficiency of processing data when combining said networks by reducing load on network CDN.

6 cl, 1 dwg

Description

Technical field

The claimed technical solution relates to the field of digital information transmission and is intended for combining a CDN network and peer-to-peer networks.

State of the art

A conventional method for delivering content in CDNs is known from the prior art, which consists in selecting for each user the CDN server that is optimal from the point of view of a particular metric and sending the requested content from the selected server to the user. This method works well under relatively uniform load conditions; for example, when a large number of users request different content at different times. However, Live content is characterized by a large number of requests for the same content at certain points in time (for example, during the broadcast of a major sporting event) and a low number of requests at other points in time. This leads to the fact that part of the time the CDN server will be idle, and part of the time they will be overloaded, which will lead to a decrease in the quality of user service.

One way to improve the quality of user service that does not require the purchase, installation and maintenance of a large number of additional servers is to use peer-to-peer networks as an additional way to deliver content to end users. This method is effective for delivering Live content, since many users simultaneously request the same content, and therefore, peer-to-peer network participants do not need to use large storage for caching content, which allows effectively connecting almost any end-user device to the peer-to-peer network, which , in turn, allows you to deliver Live content to a large number of users without creating additional load on the CDN network.

An analogue of the claimed invention is a method for combining a peer-to-peer network and a CDN network (US patent US20080155061A1 for an invention, IPC G06F15 / 16; G06F15 / 173, publ. 06/26/2008, [1]), in which the end devices of users are connected to a CDN network and can be Designated by the CDN network load balancer as a content source for other users.

A known method of unified content delivery and content planning based on CDN and P2P (Chinese patent CN101267379 for invention, IPC H04L12 / 16, H04L29 / 08, publ. September 17, 2008, [2]). As in the claimed technical solution, the method [2] is carried out using a unified management subsystem and user client. The video streaming management server performs the corresponding processing and scheduling of resources, determines the CDN streaming media endpoint that provides the service to the client.

In analogue [2], the control subsystem informs the P2P subsystem that the live broadcast source is a specific node in the CDN subsystem, and is located in the P2P subnode. The management subsystem sets the path for the distribution of the live stream in the CDN subsystem. The user client requests live content after receiving the publishing point, and the request is imported into the content planning subsystem in a unified content transfer system.

The specified analogue [2] is the set of essential features the closest analogue of the same purpose to the claimed technical solution. Therefore, it is adopted as a prototype.

A technical problem, the solution of which is provided when implementing or using a technical solution, is the need to improve the quality of service.

The disadvantage of the method [1] is the high load on the load balancing system of the CDN network associated with the need for continuous monitoring of the status of all nodes of the peer-to-peer network to ensure correct balancing. The methods of patents [1] and [2] do not contain mechanisms to prevent overload of peer-to-peer network nodes with a large number of requests from other peer-to-peer network participants. In addition, these methods [1] and [2] can lead to a deterioration in the quality of service if the data transfer rate between nodes of the peer-to-peer network is insufficient to transmit the next segment of the streaming broadcast before the playback of the current segment is completed.

Disclosure of the claimed technical solution.

The technical result provided by the claimed technical solution is to reduce the load on the CDN network.

Another technical result is the prevention of overload of peer-to-peer network nodes while increasing the data transfer rate between two peer-to-peer network nodes.

The essence of the claimed technical solution lies in the fact that the method of expanding the CDN network is carried out using a peer-to-peer network, consisting of a peer-to-peer network management subsystem that runs on one or more servers, and a peer-to-peer network client subsystem that runs on client devices. When a new client connects and upon receipt of a data update request, the ad-hoc control subsystem generates a list of ad-hoc network nodes viewing the broadcast with the same identifier, the connection parameters of which correspond to the new client’s connection parameters by the criteria that are selected so as to form a list of a given size. Moreover, the set of criteria depends on the Live content being viewed. Next, the client receives the addresses of the CDN nodes providing access to the requested content, as well as the generated list of peer-to-peer network nodes. Then the client sends a request to all nodes from the list about the possibility of downloading the next segment of the Live content being viewed, to which they respond if the requested segment is in their local cache, as well as if there are free computing and network resources. It differs in that:

- after the response, the resources are reserved for a specified period of time, then the client sends a request to download data to a random node from the number of responding, and also increases the error counter for unresponsive nodes, and in the case of no answers, transmission errors, or low download speed, the data is re-requested from one from CDN nodes;

- when the error counter reaches a certain threshold, the corresponding peer-to-peer network node is marked as “bad” by this client and ceases to be used, and information about it is transmitted to the peer-to-peer network management subsystem to exclude this node from future lists, moreover, if the number of nodes is in the client’s list it turns out to be less than a specified threshold, the missing peer network is requested from the peer-to-peer network management subsystem;

- if the number of nodes in the list is greater than the specified threshold, then the required number of nodes is randomly selected from them, while the nodes whose error counter exceeds the specified threshold, as well as the nodes present in the list sent by the client, if any, are not considered;

- if the node does not have enough resources to process all incoming requests, priority is given to customers who often download data;

- in case of a cache overflow, the oldest segments are added in time of adding.

In special cases, it is permissible to carry out the technical solution as follows.

When reserving resources on a peer-to-peer network node, it is preferable to take into account the ratio of the number of data download requests to the number of requests about the possibility of downloading data for a given period of time.

The user can rewind the video to view the fragment of the broadcast. If after this the user returns to the live broadcast, then the segments corresponding to the viewed fragment of the broadcast record are not deleted from the cache for a specified period of time.

To a randomly selected part of peer-to-peer network nodes, the peer-to-peer network control subsystem sends configuration parameters, some of which are randomly changed within specified limits. At the same time, the management subsystem monitors the operation of these nodes, and if these configuration parameters provide the best broadcast quality, or reduce the load on the client device, then they begin to be used for all new nodes of the peer-to-peer network.

Servers providing the operation of the peer-to-peer network management subsystem are the CDN cache for all types of Live content viewed by clients. In this case, peer-to-peer network nodes, in case of impossibility to download data from other peer-to-peer network nodes, first download data from the CDN cache of the management subsystem, and if it is not possible to load data from the CDN cache, they are re-requested from one of the CDN nodes. At the same time, server resources providing the operation of the management subsystem are priority allocated for the operation of the management subsystem.

On servers providing the operation of the peer-to-peer network management subsystem, one or more subsystems of the client of the peer-to-peer network are additionally executed, moreover, server resources are prioritized for the operation of the peer-to-peer network management subsystem. In this case, in the event of a lack of resources, the execution of peer-to-peer client subsystem instances is stopped, which view the broadcast that has the largest number of viewers among peer-to-peer network participants.

The above essence is a combination of essential features of the claimed technical solution, ensuring the achievement of all the claimed technical results.

Brief Description of the Drawings

The figure shows a diagram of a method for expanding a CDN network using a peer-to-peer network.

Implementation of a technical solution.

The expansion of the CDN network is carried out using the following subsystems:

- peer-to-peer client subsystems (1);

- peer-to-peer network management subsystems (2);

- subsystems (networks) of CDN (3);

- peer-to-peer network configuration subsystems (4).

A peer-to-peer network management subsystem (2) provides a software module that runs on one or more dedicated servers, or CDN network servers (3), or a combination thereof. This module stores the necessary objects for the subsystem to work in the server’s RAM, on a disk drive, or in another data storage, including a list of peer-to-peer network nodes, their error counters, the identifier of the Live content being viewed for each node, and connection parameters for each node. Connection parameters of a peer-to-peer network node may contain values of one or more metrics, such as the autonomous system number, geo-coordinates, device type, maximum cache size, type and speed of Internet connection. The identifier of the viewed Live content uniquely identifies the Live broadcast, as well as all its parameters, including broadcast quality (resolution, frame rate, bit rate), the selected audio track, the selected camera angle and others.

When a new client connects to the peer-to-peer network subsystem (1), the peer-to-peer network management subsystem (2) receives information from the peer-to-peer network subsystem (1) about the Live content being viewed and the connection parameters. After that, the peer-to-peer network management subsystem (2) generates a list of peer-to-peer network nodes viewing Live content with the same identifier, the connection parameters of which correspond to the parameters for connecting a new client to the peer-to-peer network client subsystem (1). One or more compliance criteria, as well as the maximum and minimum size of the generated list of nodes corresponding to the viewed Live content, are requested from the peer-to-peer network configuration subsystem (4).

If the configuration subsystem of the peer-to-peer network (4) does not have a configuration corresponding to the requested Live content, the default configuration is used. Moreover, if an insufficient number of nodes appears in the generated list, the peer-to-peer network management subsystem (2) can either use the less stringent compliance criterion or interrupt the formation of the list of nodes. If there are too many nodes in the generated list, the peer-to-peer network management subsystem (2) can either use the more stringent matching criterion or select the required number of nodes from the generated list randomly. After that, the new client of the peer-to-peer network subsystem (1) is sent the generated list of peer-to-peer network nodes (if one has been generated), configuration parameters received from the peer-to-peer network configuration subsystem (4), and one or more CDN nodes providing access to the requested Live client content. A list of CDN nodes is requested from the CDN subsystem (3).

Upon receipt of a request from the client subsystem of the peer-to-peer network (1) for the formation of a list of nodes of a given size, a list of nodes is formed containing no more than the requested number of elements. Moreover, if the client’s request contains a list of nodes, then nodes from this list are not considered as candidates when creating a new list.

Upon receipt of information from the client subsystem of the peer-to-peer network (1) about changing connection parameters or changing the identifier of the viewed Live content, the new data is saved and the error counter is reset to the initial value that is requested from the peer-to-peer network configuration subsystem (4).

Upon receipt from the client subsystem of the peer-to-peer network (1) a list of “bad” peers of the peer-to-peer network, their error counter increases by the amount requested from the peer-to-peer network configuration subsystem (4). If the error counter reaches the threshold value (requested from the peer-to-peer network configuration subsystem), the corresponding client ceases to be considered as a candidate when generating new lists of peer-to-peer network nodes.

Upon receipt of a request to update the configuration from the client subsystem of the peer-to-peer network (1), configuration parameters are requested from the peer-to-peer network configuration subsystem (4) and the result is sent to the client.

In the event that the peer-to-peer network management subsystem (2), or part of it is performed on dedicated servers, one or more peer-to-peer network client subsystem (1) instance can also be executed on these servers. At the same time, server resources are primarily allocated for the operation of the peer-to-peer network management subsystem (2), and when the number of free resources decreases, the number of executable instances of the peer-to-peer network subsystem is reduced (1), and those instances of the client subsystem that view the broadcast first stop running having the largest number of viewers among peer-to-peer network participants.

Examples of specific performance.

Example 1. Among all peer-to-peer network nodes, the peer-to-peer network control subsystem (2) randomly selects a fraction of nodes specified in the configuration, and passes configuration parameters to each selected node, some of which are randomly changed within specified limits. In this case, the control subsystem (2) monitors the operation of these nodes, and if these configuration parameters provide better broadcast quality, or reduce the load on the client device, then the control subsystem (2) saves this configuration in the configuration subsystem of the peer-to-peer network (4) and starts using it for all new peer-to-peer nodes.

The peer-to-peer network client subsystem (1) provides a software module running on the device of a client participating in the peer-to-peer network, which requests the ability to download the next portion of Live-content data (the next video segment) from all nodes of the peer-to-peer network from its list of nodes (if any ) Further, the peer-to-peer network client subsystem (1) expects to receive responses within the time interval specified in the configuration, after which it requests the next video segment from a random node from the number of respondents, and also increases the error counter by the value specified in the configuration for each non-responding node. If no response was received from any node in a specified period of time, or if there is no list of peer-to-peer network nodes, the next video segment is requested from the CDN network (3). If a video segment was requested from a peer-to-peer network node, but its download is too slow to ensure continuous playback of Live content, or if the download was interrupted, or if a message was received that it was not possible to transmit the requested data, this segment is re-requested from the CDN network (3), and for this peer-to-peer network node, the error counter increases by the value specified in the configuration. The resulting segment is stored in a local cache located in RAM, on a disk drive, or in another data storage. At the same time, if there is not enough free space in the local cache to save the received segment, the oldest segments in time of receiving segments are sequentially deleted until enough space is freed up in the cache to save the new segment.

Example 2. If a Live broadcast supports rewinding, and the user rewound the video to view some fragment of the broadcast, and then went back to live, then the segments corresponding to the viewed fragment of the broadcast record will not be deleted from the cache during the interval specified in the configuration time.

When the error counter reaches the threshold specified in the configuration, the corresponding node is marked as “bad” and excluded from the list of peer-to-peer network nodes. If after this there are fewer nodes in the list of peer-to-peer network nodes than the threshold value specified in the configuration, a request is made from the peer-to-peer network control subsystem (2) for an additional list of size equal to the difference between the value specified in the configuration and the current size of the list. At the same time, the current list of nodes and the list of "bad" nodes can also be transmitted in the request. The nodes from the list received from the peer-to-peer network management subsystem (2) are added to the current list of nodes. If the list is not received from the peer-to-peer network management subsystem (2), or if, after adding the received nodes, the size of the list is still less than the threshold value specified in the configuration, a second request is made after the time interval specified in the configuration.

Upon receipt of a request about the possibility of downloading certain data from another peer-to-peer network node, the peer-to-peer client subsystem (1) checks for the requested data in the local cache. If there is data, the availability of free computing and network resources is checked to ensure the transfer of requested data without affecting the quality of playback of content on this device, as well as without affecting the quality of data transfer to other participants of the peer-to-peer network. If there are not enough resources to service all incoming requests, priority is given to those peer-to-peer network nodes that have a higher proportion of the number of data download requests to the number of requests about the possibility of downloading data for a specified period of time in the configuration. If there is data in the cache and the availability of resources, a response is sent to the incoming request, confirming the ability to download data, and backups of computing and network resources for a specified period of time in the configuration. In all other cases, the response is not sent and the reservation of resources is not made.

Example 3. When reserving resources, the proportion of the number of data loading requests to the number of requests for the possibility of downloading data for a specified period of time is taken into account.

Upon receipt of a request to download data from another peer-to-peer network node, a repeated check is made of the availability of data in the cache and the availability of free computing and network resources. If the check is passed, the requested data is sent; otherwise, a message is sent indicating the impossibility to transmit the requested data. In case of transmission errors, retransmission is not performed.

When changing the connection parameters, as well as when changing the identifier of the viewed Live content, a corresponding request is sent to the peer-to-peer network management subsystem (2). In addition, after a period of time specified in the configuration, a request to update the configuration is sent to the peer-to-peer network management subsystem (2).

Example 4. Dedicated servers that provide the management subsystem (2) also play the role of a CDN cache for all types of Live content viewed by clients. In this embodiment, peer-to-peer network nodes, in case of impossibility to download data from other peer-to-peer network nodes, first try to load data from the CDN cache of the control subsystem (2). If the data was not received within the specified period of time in the configuration, or if the CDN cache reported that it was not possible to provide the requested data, the data is re-requested from one of the CDN nodes. At the same time, the resources of the server providing the operation of the management subsystem (2) are allocated priority for the operation of the management subsystem (2). Data is loaded into the CDN cache from the CDN (3).

The CDN subsystem (3) is a particular CDN network that can provide users with the requested Live content.

A peer-to-peer network configuration subsystem (4) is a set of configuration profiles stored in a configuration file, database, or other information storage.

From the foregoing, it will be clear to a specialist that the claimed invention reduces the load on the CDN network (3), does not require constant monitoring of the status of all peer-to-peer network nodes, does not allow overloading of peer-to-peer network nodes, and also does not allow deterioration in the quality of service due to insufficient data transfer speed between two peer-to-peer nodes.

Although preferred embodiments of the present invention have been described above, one skilled in the art will appreciate that the present invention is not limited to these examples. Moreover, various changes and modifications can be made within the essence and scope of the present invention.

Industrial applicability

The inventive method of reducing the load on the CDN network is implemented using industrially produced devices and materials, and will find wide application in the field of digital content distribution.

Claims (8)

1. A method of expanding a CDN network using a peer-to-peer network, consisting of a peer-to-peer network management subsystem running on one or more servers, and a peer-to-peer network client subsystem running on client devices, which, when connecting to a new client and receiving a request for data updating, the peer-to-peer network management subsystem generates a list of peer-to-peer network nodes viewing the broadcast with the same identifier, the connection parameters of which correspond to the new client’s connection parameters according to the criteria that are selected so as to form a list of a given size, and the set of criteria depends on the Live content being viewed, Next, the client receives the addresses of the CDN nodes providing access to the requested content, as well as the generated list of peer-to-peer network nodes, then the client sends a request to all nodes from the list about the possibility of downloading the next segment of the viewed Live content, to which they respond with the presence of the requested segment in their local cache, as well as the availability of free computing and network resources, characterized in that after the answer the resources are reserved for a specified period of time, then the client sends a request to load data to a random node from the number of respondents, and also increases the error counter for Unanswered nodes, moreover, in the absence of answers, transmission errors, or low download speed, data is re-requested from one of the CDN nodes; when the error counter reaches a certain threshold, the corresponding peer-to-peer network node is marked by this client as "bad" and is no longer used, and information about it is transmitted to the peer-to-peer network management subsystem to exclude this node from future lists, moreover, if the number of nodes in the client’s list is less given threshold, the peer network management subsystem is requested the missing number of nodes; if the number of nodes in the list is greater than the specified threshold, then the required number of nodes is randomly selected from them, while the nodes whose error counter exceeds the specified threshold, as well as nodes present in the list sent by the client, if any, are not considered; if the node does not have enough resources to process all incoming requests, priority is given to clients who download data more often; in case of a cache overflow, the oldest added segments are deleted. 2. The method according to claim 1, characterized in that when reserving resources on a peer-to-peer network node, the ratio of the number of requests for downloading data to the number of requests for the possibility of downloading data for a given period of time is taken into account. 3. The method according to claim 1, characterized in that when the user rewinds the video back to view the broadcast fragment and then returns to the live broadcast, the segments corresponding to the viewed fragment of the broadcast record are not deleted from the cache for a specified period of time. 4. The method according to claim 1, characterized in that to the randomly selected part of the peer-to-peer network nodes, the peer-to-peer network control subsystem sends configuration parameters, part of the values of which are randomly changed within the specified limits, while the control subsystem monitors the operation of these nodes, and if the data configuration parameters provide better broadcast quality, or reduce the load on the client device, then they begin to be used for all new peer-to-peer network nodes.
5. The method according to claim 1, characterized in that the servers providing the operation of the peer-to-peer network management subsystem are the CDN cache for all types of Live content viewed by clients, and peer-to-peer nodes, in case of impossibility to download data from other peer-to-peer network nodes, first download data from the CDN cache of the management subsystem, and if it is not possible to download data from the CDN cache, they are re-requested from one of the CDN nodes, while the server resources providing the operation of the management subsystem are priority allocated for the operation of the management subsystem. 6. The method according to claim 1, characterized in that the servers providing the operation of the peer-to-peer network management subsystem additionally execute one or more peer-to-peer network client subsystems, moreover, server resources are prioritized for the operation of the peer-to-peer network management subsystem, and in case of insufficient resources execution of peer-to-peer client subsystem instances that view the broadcast with the largest number of viewers among peer-to-peer network participants.

Images