JP2022135642A - Transfer device for content distribution network and program - Google Patents

Abstract

To provide a technique that prevents congestion in pre-acquisition processing.SOLUTION: A transfer device for a content distribution network divides a content into one or more chunks and distributes the chunks, and comprises: determination means that, when receiving a plurality of first request packets for requesting chunks of a request content from a client device, determines whether or not to perform pre-acquisition processing of the request content based on the plurality of first request packets; and transmission means that, when determining to perform the pre-acquisition processing of the request content, generates and transmits a plurality of second request packets for requesting chunks, of the chunks of the request content, which are different from the chunks requested by the plurality of first request packets. The transmission means controls the transmission interval between the plurality of second request packets based on the round-trip delay between the client device and the transmission means.SELECTED DRAWING: Figure 6

Description

The present invention relates to a transfer device and program for a content distribution network that divides and distributes content.

Networks have been proposed that distribute content based on names that indicate the content. An outline of a content centric network (CCN: Content Centric Networking), which is one of such networks, will be described below.

In CCN, a server device that publishes content divides the content into one or more chunks or data portions called objects, and a client device acquires the content in units of chunks. Also, in CCN, a communication device that has transferred a chunk (hereinafter referred to as a transfer device) can hold (cache) the chunk. When this transfer device receives an interest packet (request packet) requesting a chunk cached by the device from the client device, the transfer device does not transfer the interest packet to the server device, but holds the interest packet. A data packet containing the chunk can be sent towards the client device from which the Interest packet was sent.

An example of the operation of the transfer device will be described below. The transfer device manages CS (Contents Store), FIB (Forwarding Information Base), and PIT (PIT: Pending Interest Table). CS is information indicating chunks cached by the device itself. The FIB is information indicating the relationship between an Interest packet and an interface to which the Interest packet should be transferred. The PIT is information indicating the relationship between the chunk requested by the forwarded Interest packet and the interface that received the forwarded Interest packet. The PIT is also information indicating that an Interest packet has been transferred and that a chunk requested by the transferred Interest packet is waiting to be received.

When a forwarding device receives an Interest packet, it searches the CS to determine whether it caches the chunk requested by the Interest packet. If it is cached, it transmits a data packet containing the chunk cached by its own device to the client device that sent the interest packet. On the other hand, if the chunk requested by the received Interest packet is not cached, the forwarding device searches the PIT, forwards an Interest packet requesting the same chunk as the Interest packet, and receives the chunk. Determine whether it is in a waiting state. In the state of waiting for reception, the transfer device does not transfer the received Interest packet, and updates the PIT by associating the receiving interface of the Interest packet with the chunk requested by the Interest packet. On the other hand, if the received interest packet is not waiting for the requested chunk, the transfer device transfers the interest packet from the interface determined based on the FIB and updates the PIT. Further, upon receiving a data packet including a chunk, the transfer device determines the transfer destination interface of the data packet based on the PIT and the chunk included in the data packet, and transmits the data packet from the determined interface. At the same time, the information about the chunk is deleted from the PIT. Also, the transfer device updates the CS when it caches the chunks included in the transferred data packets.

Patent Literature 1 discloses a configuration for shortening the acquisition time of content by a client device. According to Patent Document 1, a boundary forwarding device is provided within a CCN. The edge forwarding device is connected to a link with a large delay (hereinafter referred to as a long distance link). A border forwarding device initiates a "pre-acquisition" of requested content when multiple interest packets requesting the same chunk of content (requested content) and being forwarded to the long-distance link meet a predetermined condition. The "pre-acquisition processing" of the requested content means that, among the chunks of the requested content, a plurality of chunks temporally later than the chunk requested by the received and transferred interest packets are received from the client device. It means the operation of acquiring in advance before receiving a packet.

JP 2020-123776 A

Patent document 1 shortens the content acquisition time by the client device by acquiring in advance a plurality of chunks that are highly likely to be requested by the client device in the future through pre-acquisition processing. However, the generation and transmission of a large number of interest packets by the edge forwarding device in the pre-acquisition process can cause congestion.

The present invention provides techniques for reducing congestion in pre-acquisition processing.

According to one aspect of the present invention, when a transfer device of a content distribution network that distributes content by dividing it into one or more chunks receives a plurality of first request packets requesting chunks of requested content from a client device, determining means for determining whether or not to perform pre-acquisition processing of the requested content based on a plurality of first request packets; a transmission means for generating and transmitting a plurality of second request packets each requesting a chunk different from the chunk requested by the first request packet of the transmission means, wherein the transmission means makes a round with the client device The transmission interval of the plurality of second request packets is controlled based on the trip delay.

According to the present invention, congestion can be suppressed in pre-acquisition processing.

1 is a block diagram of a content distribution network according to one embodiment; FIG. 1 is a configuration diagram of a boundary transfer device according to an embodiment; FIG. 4 is a flowchart of processing in a processing unit according to one embodiment; 4 is a flowchart of queue management processing in the pre-acquisition unit according to one embodiment; 4 is a flowchart of interest packet generation processing in the pre-acquisition unit according to one embodiment. 4 is a sequence diagram showing an outline of pre-acquisition processing according to one embodiment; FIG. 1 is a configuration diagram of a boundary transfer device according to an embodiment; FIG.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. It should be noted that the following embodiments do not limit the invention according to the claims, and not all combinations of features described in the embodiments are essential to the invention. Two or more of the features described in the embodiments may be combined arbitrarily. Also, the same or similar configurations are denoted by the same reference numerals, and redundant explanations are omitted.

FIG. 1 is a configuration diagram of a content distribution network according to this embodiment. In the following, the present embodiment will be described assuming that the content distribution network is CCN. However, the present invention is not limited to the CCN, but divides the content into one or more pieces, distributes the content in units of each data portion obtained by the division, and caches the transferred data portion in the transfer device. It can be applied to any content delivery network capable. Also, in the following description, each data portion obtained by dividing the content is referred to as a "chunk" according to CCN. However, the name of each data part obtained by dividing one content is arbitrary.

As shown in FIG. 1, the content distribution network includes a server device 3 that publishes content, client devices 11 and 12 that acquire content according to CCN, and a border transfer device 2 . The client devices 11 and 12 and the border forwarding device 2 are connected to a local network 4 consisting of the usual forwarding devices of the CCN mentioned above. In FIG. 1, client devices 11 and 12, boundary transfer device 2 and local network 4 are provided in Japan. On the other hand, the server device 3 is provided outside the country, for example, in England. The boundary transfer device 2 and the server device 3 are configured to be able to communicate via a global network composed of normal transfer devices and the boundary transfer device 2 according to this embodiment. The content distribution network also has a server device provided in Japan, but it is omitted because it is not relevant to the description of the present embodiment. In addition to the server device 3, the content distribution network has server devices installed in various countries other than Japan, but they are omitted for the sake of simplification of explanation.

In this embodiment, when a client device such as the client devices 11 and 12 installed in Japan acquires content published by a server device such as the server device 3 installed in a country other than Japan, the client device in Japan The FIB of the forwarding device of the local network 4 is set so that the interest packet transmitted by the device is forwarded to the border forwarding device 2 . In this embodiment, to simplify the explanation, it is assumed that only one boundary transfer device 2 is installed in Japan, but a configuration in which multiple boundary transfer devices 2 are installed in Japan is also possible. can. For example, the boundary transfer device 2 is installed in Tokyo and Osaka, and among the interest packets requesting the content published by the server device 3, the interest packet sent by the client device in eastern Japan is sent by the boundary transfer device installed in Tokyo. 2, and the FIB of the transfer device of the local network 4 is set so that the interest packet sent by the client device in western Japan is transferred to the boundary transfer device 2 installed in Osaka. .

The client devices 11 and 12 manage the window size W and the number U of interest packets for which no data packet is received as a response among the transmitted interest packets. The window size W defines the maximum allowable value of number U. For example, with window size W=4, client devices 11 and 12 can initially send up to four interest packets in succession. Then, after transmitting the four Interest packets, the client devices 11 and 12 transmit new Interest packets until receiving a data packet in response to any of the four Interest packets. Can not do it. Also, upon receiving a data packet as a response to one of the four Interest packets, the client devices 11 and 12 can transmit one new Interest packet.

In addition, the client devices 11 and 12 increase or decrease the window size W according to whether or not they have received data packets within a predetermined period as responses to the transmitted interest packets. For example, let the initial value of the window size W be four. In this case, the client devices 11 and 12 double the window size W to 8 after transmitting four interest packets and receiving data packets for all of them within a predetermined period of time. Since the window size W=8, client devices 11 and 12 are able to send 8 interest packets in succession. After sending eight interest packets, the client devices 11 and 12 double the window size W to 16 when they receive data packets for all of them within a predetermined period of time. Thus, in this embodiment, the client devices 11 and 12 double the window size W when all responses to the transmitted interest packets are received within a predetermined period of time. In other words, the client devices 11 and 12 set the window size W to 4, 8, 16, 32, 64, 128 when the amount of traffic on the content distribution network is small and the responses to the transmitted interest packets are received within a predetermined period. , 256 . On the other hand, the client devices 11 and 12 halve the window size W if they do not receive at least one response to the transmitted interest packet within a predetermined period. By gradually increasing the window size W step by step in this way, it is possible to appropriately control the amount of interest packets to be transmitted according to the traffic conditions of the content distribution network. There are various methods for controlling the window size W other than the method described above. be able to.

For example, assume that the client device 11 acquires certain content from the server device 3 . It is assumed that none of the transfer devices (including the boundary transfer device 2) between the client device 11 and the server device 3 cache the chunk of the content. In other words, it is assumed that all interest packets requesting chunks of the content transmitted by the client device 11 are transferred to the server device 3 and all chunks are delivered from the server device 3 . Also, the physical lines of all links between the client device 11 and the server device 3 are sufficiently large, and the traffic volume of all links between the client device 11 and the server device 3 is sufficiently small. and In this case, the client device 11 can sequentially increase the window size W. However, since the distance between Japan and the United Kingdom is long, the link between the client device 11 and the server device 3 is not congested. The round-trip delay (RTT) is also around 200ms. In other words, the client device 11 can only increase the window size W approximately every 200 ms. Thus, when the RTT is large, it takes a long time to increase the window size W to a size corresponding to the possible throughput with the server device 3, and the time required to acquire the content becomes long. Assuming that the server device 3 is installed in Japan, the RTT is about several ms to several tens of ms, and the client device 11 sets the window size W to correspond to the possible throughput up to the server device 3. It can be increased in size in a short time.

FIG. 2 is a block diagram of the boundary transfer device 2. As shown in FIG. The processing unit 22 has a plurality of interfaces and basically performs the same operation as a normal transfer device. That is, the processing unit 22 manages the FIB, PIT, and CS, and transfers interest packets and data packets. Also, when a data packet is received and transferred, the chunk stored in the data packet is cached according to a predetermined criterion. In addition to the same operation as a normal transfer device, the processing unit 22, when transferring a certain Interest packet, acquires the Interest packet based on the chunk name of the chunk requested by the Interest packet. Determine whether it is a processing target. Note that if the interest packet is to be processed by the pre-acquisition unit 21 , the processing unit 22 transmits the interest packet to the pre-acquisition unit 21 . Note that when transmitting an Interest packet to the pre-acquisition unit 21, the Interest packet is not transmitted only to the pre-acquisition unit 21, but is transferred from an interface determined based on the FIB. The interest packet is transmitted to the pre-acquisition unit 21 .

In this embodiment, the chunk name has a hierarchical structure, and the highest level of the hierarchical structure indicates the country name. For example, the name of the content published by the server device in Japan starts with jp indicating Japan, and the chunk name of number 1 of the content whose name is "A" is, for example, "/jp/.../A/ 1". Similarly, the name of the content published by the server device 3 in England starts with uk indicating the United Kingdom, and the chunk name of number 1 of the content whose name is "A" is, for example, "/uk/.../A /1". Also, content whose size is equal to or less than the maximum allowable size of one chunk, such as an html file, is assumed to have the same content name and chunk name. For example, it is assumed that the content whose name is "/uk/.../AAA.html" published by the server device 3 in England also has a chunk name of "/uk/.../AAA.html". do. Among the interest packets to be transferred, the processing unit 22 requests a chunk whose requested chunk name indicates a foreign country and has a chunk number, that is, one chunk of the content divided into a plurality of chunks. interest packet is determined to be processed by the pre-acquisition unit 21 .

FIG. 3 is a processing flow when the processing unit 22 receives an interest packet. The processing unit 22 waits until an interest packet is received in S10. Upon receiving the interest packet, the processing unit 22 determines in S11 whether or not the interest packet should be forwarded based on the FIB. For example, S11 becomes No when an interest packet requesting a cached chunk or an interest packet requesting a chunk indicated in the PIT is received. If the interest packet should not be forwarded based on the FIB, the processing unit 22 repeats the processing from S10. On the other hand, if the interest packet should be forwarded based on the FIB, the processing unit 22 determines whether or not the interest packet is to be processed by the pre-acquisition unit 21 in S12. Whether or not the chunk is to be processed by the pre-acquisition unit 21 is determined as described above based on the chunk name of the chunk requested by the interest packet. If the interest packet is not to be processed by the pre-acquisition unit 21, the processing unit 22 repeats the processing from S10. On the other hand, if the interest packet is to be processed by the pre-acquisition unit 21, the processing unit 22 transmits the interest packet to the pre-acquisition unit 21 in S13. In addition to the processing shown in FIG. 3, the processing unit 22 performs the same processing as a normal transfer device on interest packets.

FIG. 4 is a flowchart of queue management processing in the pre-acquisition unit 21. As shown in FIG. In S<b>20 , the pre-acquisition unit 21 waits until an interest packet is received from the processing unit 22 . Upon receiving an interest packet from the processing unit 22, the pre-acquisition unit 21 manages in S21 whether or not there is a queue corresponding to the content of the chunk requested by the received interest packet. If there is no queue corresponding to the content of the chunk requested by the Interest packet, the pre-acquisition unit 21 determines in S22 that the number of the chunk requested by the Interest packet (hereinafter referred to as RN) is equal to or less than a threshold. determine whether If RN is greater than the threshold, the pre-acquisition unit 21 repeats the process from S20. On the other hand, if the RN is equal to or less than the threshold, the pre-acquisition unit 21 creates a queue corresponding to the content and stores the RN in association with the queue in S23. Then, in S24, the pre-acquisition unit 21 starts counting the timer corresponding to the created queue.

On the other hand, if there is a queue corresponding to the content of the chunk requested by the interest packet in S21, the pre-acquisition unit 21 determines in S25 that the request chunk number of the received interest packet corresponds to the queue of the content. Determine if it is greater than RN. If the number of the request chunk of the received Interest packet is greater than the RN corresponding to the content queue, the pre-acquisition unit 21 sets the RN corresponding to the content queue to the number of the request chunk of the received Interest packet. update to On the other hand, if the request chunk number of the received interest packet is less than or equal to the RN corresponding to the content queue, the pre-acquisition unit 21 does not change the RN corresponding to the content queue.

FIG. 5 is a flowchart of interest packet transmission processing in the pre-acquisition unit 21 . In S30, the pre-acquisition unit 21 determines whether or not there is a queue whose count value of the timer has reached a predetermined value. If there is no queue whose count value of the timer has reached the predetermined value, the pre-acquisition unit 21 repeats the processing from S30. If there is a queue whose timer count value has reached a predetermined value, the pre-acquisition unit 21 determines in S31 whether the RN value corresponding to the queue is equal to or less than a threshold value. In this embodiment, the threshold used in S31 is the same as the threshold used in S22 of FIG. However, the threshold used in S31 and the threshold used in S22 of FIG. 4 may be different values. If the RN value corresponding to the queue is greater than the threshold, the pre-acquisition unit 21 discards the queue in S33 and repeats the process from S30. On the other hand, if the value of the RN corresponding to the queue is equal to or less than the threshold, the pre-acquisition unit 21 selects a content chunk corresponding to the queue whose number is one greater than the value of the RN corresponding to the queue. An interest packet requesting chunks from the value to a value larger by a predetermined value X is generated and transmitted to the processing unit 22 . After that, in S33, the pre-acquisition unit 21 discards the queue and repeats the process from S30.

It should be noted that the processing unit 22 only performs transfer processing on the interest packets received from the pre-acquisition unit 21, and does not perform the processing described with reference to FIG. Also, the processing unit 22 performs the same transfer processing as a normal transfer device for chunks (data packets) received as a response to the interest packet received from the client device 1 . On the other hand, the processing unit 22 caches the chunk received as a response to the interest packet received from the pre-acquisition unit 21 and does not transfer it to the pre-acquisition unit 21 . If there is no capacity to cache all the chunks received as responses to the interest packets received from the pre-acquisition unit 21, the shorter the time after receiving the chunks, the more preferentially they are cached.

A case where the client device 11 acquires content with a content name of "uk/aaa/videoA.mp4" will be described below. The content is divided into 10000 chunks, and the chunk names are "uk/aaa/videoA.mp4/1" to "uk/aaa/videoA.mp4/10000". It is also assumed that the client device 11 has not acquired any content so far and the window size W is a very small initial value, for example 4. It is also assumed that the threshold held by the pre-acquisition unit 21 is 300 and the predetermined value X is 150. In the initial state, none of the transfer devices existing from the client device 11 to the server device 3 cache any chunk of "uk/aaa/videoA.mp4".

Since the window size W=4, the client device 11 continuously sends four interest packets requesting "uk/aaa/videoA.mp4/1" to "uk/aaa/videoA.mp4/4". Send. Each interest packet is to be transferred by the boundary transfer device 2 and to be processed by the pre-acquisition unit 21 (S11 and S12 in FIG. 3), so it is transmitted to the pre-acquisition unit 21. FIG.

Since the pre-acquisition unit 21 does not manage any queue, when receiving an interest packet requesting "uk/aaa/videoA.mp4/1" which is a chunk with a number smaller than the threshold, in S23 of FIG. A queue is created, RN=1 is held in association with the queue, and the timer corresponding to the queue starts counting in S24 of FIG. Subsequently, when the pre-acquisition unit 21 receives an interest packet requesting "uk/aaa/videoA.mp4/2", in S25 of FIG. Update RN to 2. Similarly, by receiving interest packets requesting "uk/aaa/videoA.mp4/3" and "uk/aaa/videoA.mp4/4", the content "uk/aaa/videoA.mp4" is queued. RN is updated to 4.

When the count value of the timer of the queue corresponding to the content "uk/aaa/videoA.mp4" reaches a predetermined value (S30 in FIG. 5), the pre-acquisition unit 21, in S31, sets the RN corresponding to the queue to the threshold value. It is determined whether the number is less than or equal to 300. In this example, since RN=4 and the threshold value is 300 or less, the pre-acquisition unit 21, in S32, requests chunks with numbers 5 to 154 of the content "uk/aaa/videoA.mp4". A packet is generated and transmitted to the processing unit 22 . That is, the pre-acquisition unit 21 generates an interest packet requesting “uk/aaa/videoA.mp4/154” from “uk/aaa/videoA.mp4/5” and transmits it to the processing unit 22 .

As a result, the edge transfer device 2 has transmitted an interest packet requesting "uk/aaa/videoA.mp4/154" from "uk/aaa/videoA.mp4/1" to the server device 3. . Of these, "uk/aaa/videoA.mp4/1" to "uk/aaa/videoA.mp4/4" were transmitted by the client device 11, and the boundary transfer device 2 Data packets received in response to packets are forwarded normally. On the other hand, since the interest packet requesting "uk/aaa/videoA.mp4/154" from "uk/aaa/videoA.mp4/5" was generated by the pre-acquisition unit 21, the processing unit 22 Cache "uk/aaa/videoA.mp4/4" through "uk/aaa/videoA.mp4/154" received as responses to these interest packets.

When the client device 11 receives "uk/aaa/videoA.mp4/4" from "uk/aaa/videoA.mp4/1", it doubles the window size W to 8, and "uk/aaa/videoA .mp4/5" to "uk/aaa/videoA.mp4/12", but since these are cached in border forwarding device 2, border forwarding device 2 immediately retrieves these chunks. It can be sent to the client device 11 . After that, the client device 11 doubles the window size W to 16 and transmits an interest packet requesting "uk/aaa/videoA.mp4/28" from "uk/aaa/videoA.mp4/13". However, since they are cached in the boundary transfer device 2, the boundary transfer device 2 can send these chunks to the client device 11 immediately.

After that, when the client device 11 transmits an interest packet requesting "uk/aaa/videoA.mp4/155", the boundary transfer device 2 creates a new queue corresponding to the content "uk/aaa/videoA.mp4". created in For example, at this time, if the window size W of the client device 11 is 128, the pre-acquisition unit 21 sets "uk/aaa/ will receive an Interest packet requesting up to videoA.mp4/252". Therefore, in S32 of FIG. 5, the border transfer device 2 generates and transmits an interest packet requesting "uk/aaa/videoA.mp4/253" to "uk/aaa/videoA.mp4/402". Therefore, the boundary transfer device 2 can prefetch and cache "uk/aaa/videoA.mp4/253" to "uk/aaa/videoA.mp4/402", and the client device 11 can cache "uk/aaa /videoA.mp4/253" to "uk/aaa/videoA.mp4/402" can be acquired in a short time.

In this way, when the window size W is smaller than the threshold, the boundary transfer device 2 prefetches and caches the content acquired by the client device 11, thereby quickly increasing the window size W of the client device 11. Therefore, the content download time by the client device 11 can be shortened. After that, when the client device 11 transmits an interest packet requesting "uk/aaa/videoA.mp4/403", S22 in FIG. 4 becomes No and no queue is created.

Next, a case where the client device 11 acquires content with a content name of "uk/aaa/video.html" will be described. Since the size of the content is less than or equal to the maximum amount of data in one chunk, the chunk name is equal to the content name and has no chunk number. In this case, the client device 11 transmits an interest packet requesting "uk/aaa/video.html", but since there is no chunk number, it is not subject to processing by the pre-acquisition unit 21 (S12 in FIG. 3). ). Therefore, the interest packet is not sent to the pre-acquisition unit 21 .

Next, a case where the client device 12 acquires content with a content name of "uk/aaa/videoB.mp4" will be described. The content is divided into 10000 chunks, and the chunk names are "uk/aaa/videoB.mp4/1" to "uk/aaa/videoB.mp4/10000". It is also assumed that the client device 12 has already acquired other content, and the window size W is a value, for example, 1024, corresponding to the throughput of content acquisition up to that point. It is also assumed that the threshold held by the pre-acquisition unit 21 is 300 and the predetermined value X is 150. In the initial state, none of the transfer devices existing from the client device 12 to the server device 3 cache any chunk of "uk/aaa/videoB.mp4".

Since the window size W=1024, the client device 12 continuously transmits 1024 interest packets requesting "uk/aaa/videoB.mp4/1" to "uk/aaa/videoB.mp4/1024". to send. Each interest packet is to be transferred by the boundary transfer device 2 and to be processed by the pre-acquisition unit 12 , so it is transmitted to the pre-acquisition unit 21 . Upon receiving the interest packet requesting "uk/aaa/videoB.mp4/1", the pre-acquisition unit 21 creates a queue in S23 of FIG. 4, records RN=1 in association with the queue, At S24 in FIG. 4, the timer corresponding to the queue starts counting. Subsequently, when the pre-acquisition unit 21 receives an interest packet requesting "uk/aaa/videoB.mp4/2", in S25 of FIG. Update RN to 2. Similarly, by receiving an interest packet requesting "uk/aaa/videoB.mp4/3" to "uk/aaa/videoB.mp4/1024", the content "uk/aaa/videoB.mp4" is queued. The corresponding RN is updated to 1024.

When the count value of the timer corresponding to the queue of the content "uk/aaa/videoB.mp4" reaches a predetermined value (S30 in FIG. 5), the pre-acquisition unit 21 determines in S31 that the RN of the queue is 300, which is the threshold. Determine if: In this example, since RN=1024, which is larger than the threshold value of 300, the pre-acquisition unit 21 discards the queue of the content "uk/aaa/videoB.mp4" in S33. That is, the pre-acquisition unit 21 does not generate and transmit an interest packet requesting chunks of the content "uk/aaa/videoB.mp4". Thus, when the window size W of the client device 12 is sufficiently large (larger than the threshold value), the boundary transfer device 2 does not prefetch chunks because the effect of prefetching by the boundary transfer device 2 is small.

Next, a case where the client device 12 acquires content with a content name of "uk/aaa/videoC.mp4" will be described. The content is divided into 10000 chunks, and the chunk names are "uk/aaa/videoC.mp4/1" to "uk/aaa/videoC.mp4/10000". Also, "uk/aaa/videoC.mp4/1" to "uk/aaa/videoC.mp4/4000" are cached in the transfer device within the local network 4, and the client device 12 accesses "uk/aaa/ It is assumed that videoC.mp4/1" to "uk/aaa/videoC.mp4/4000" have already been obtained from the local network 4. Further, it is assumed that the window size W of the client device 12 is a value, for example, 1024, corresponding to the throughput of content acquisition up to that point. All the transfer devices existing from the client device 12 to the server device 3 do not cache any of "uk/aaa/videoC.mp4/4001" to "uk/aaa/videoC.mp4/10000". and Furthermore, it is assumed that the threshold held by the pre-acquisition unit 21 is 300 and the predetermined value X is 150.

Since the window size W=1024, the client device 12 continuously transmits 1024 interest packets requesting "uk/aaa/videoC.mp4/4001" to "uk/aaa/videoC.mp4/5024". to send. Each interest packet is to be transferred by the boundary transfer device 2 and to be processed by the pre-acquisition unit 21 , so it is transmitted to the pre-acquisition unit 21 . The pre-acquisition unit 21 receives an interest packet requesting "uk/aaa/videoC.mp4/4001", but since the requested chunk number is 1001, which is greater than the threshold of 300, a queue is created. not (S22). Thus, the window size W of a client device requesting chunks larger than the threshold is usually large enough, and thus the effect of prefetching by the boundary forwarding device 2 is small, so that the boundary forwarding device 2 does not prefetch chunks. Not performed.

As described above, the edge forwarding device 2 transmits a large number of Interest packets (X=150 Interest packets in the above example) through the pre-acquisition process. When the boundary transfer device 2 continuously transmits a large number of interest packets and the server device 3 transmits a large number of data packets in response, congestion occurs in the section from the server device 3 to the boundary transfer device 2. can give rise to Therefore, the pre-acquisition unit 21 controls the transmission timing of interest packets transmitted by the pre-acquisition process so as not to cause congestion in the section from the server device 3 to the boundary transfer device 2 . Since the size of an interest packet is very small compared to the size of a data packet, if congestion due to a large number of data packets is prevented from occurring in the section from the server device 3 to the border transfer device 2, Congestion due to a large number of interest packets does not occur in the section from the edge transfer device 2 to the server device 3 .

The control of the transmission timing of the interest packet by the pre-acquisition unit 21 will be described below. FIG. 6 outlines the sequence of pre-acquisition processing. As described above, the maximum number X (150 in the above example) of interest packets to be transmitted in the pre-acquisition process is set in the pre-acquisition unit 21 . Also, the client devices 11 and 12 typically transmit multiple interest packets in succession, ie, in bursts, according to a window size W. FIG. Hereinafter, the burst transmission of a plurality of Interest packets by the client devices 11 and 12 according to the window size W will be referred to as "burst transmission" of Interest packets.

For example, when the client device 11 whose window size W is the initial value of 4 starts acquiring certain content, the client device 11 first bursts an interest packet requesting chunks numbered 1 to 4. . As a result, as described above, the boundary transfer device 2 acquires and caches the chunks of numbers 5 to 154 of the content in the pre-acquisition process.

When the client device 11 receives a data packet as a response to the burst-transmitted interest packet, the client device 11 repeats the burst transmission of the interest packet while increasing the window size W as described above. In the above example, the client device 11 performs burst transmission of interest packets while increasing the window size W to 8, 16, 32, 64, 128, and so on. Note that the client device 11 acquires chunks numbered 5 to 12 by burst transmission of interest packets when the window size W=8. Also, the client device 11 acquires chunks numbered 13 to 28 by burst transmission of interest packets when the window size W=16. Further, the client device 11 acquires chunks numbered 29 to 60 by burst transmission of interest packets when the window size W=32. Furthermore, the client device 11 acquires chunks numbered 61 to 124 by burst transmission of interest packets when the window size W=64. Furthermore, the client device 11 obtains chunks numbered 125 to 252 by burst transmission of interest packets when the window size W=128. Among them, chunks numbered 125 to 154 are acquired by the pre-acquisition process.

In this way, in the above example, the client device 11 performs burst transmission of interest packets five times, thereby acquiring chunks numbered 5 to 154 acquired by the boundary transfer device 2 in the pre-acquisition process.

FIG. 6 is a sequence diagram generalizing the above contents. The round-trip delay between the client device 11 and the border transfer device 2 is t1, and the _round -trip delay between the border transfer device ₂ and the server device 3 is t2. At S1 (time 0), the client device 11 performs burst transmission of interest packets. These interest packets are received by the edge forwarding device 2 at time t ₁ /2 and trigger the edge forwarding device 2 to start the pre-acquisition process. As described with reference to FIGS. 4 and 5, the boundary transfer device 2 actually waits until the count value of the timer reaches a predetermined value from time t ₁ /2 before starting pre-acquisition processing. The period is much smaller than the round-trip delay and can be ignored. The interest packet burst-transmitted by the client device 11 in S1 is transferred from the boundary transfer device 2 to the server device 3 in S2. The server device 3 transmits a plurality of data packets in S3 in response to the burst transmission of the interest packets received in S2. The border transfer device 2 forwards the plurality of data packets to the client device 11 at S4. Client device 11 receives these data packets at time t ₂ +t ₁ .

Thereafter, the client device 11 repeats burst transmission of interest packets while increasing the window size W (S5-S6). Since the boundary transfer device 2 prefetches the chunk by pre-acquisition processing, the client device 11 acquires the chunk cached by the boundary transfer device 2 from S5. Therefore, the client device 11 receives a data packet at time t ₂ +2t ₁ as a response to the burst transmission of the interest packet in S5 (first time). Similarly, the client device 11 receives a data packet at time t ₂ +(m+1) t ₁ as a response to the burst transmission of the Interest packet in S6 (m-th time).

For example, in the pre-acquisition process starting at time t ₁ /2, the edge transfer device 2 acquires X chunks to be sent in response to m burst transmissions (S5-S6) of interest packets by the client device 11. It shall be. It is assumed that the values of X and m (integers) are set in the boundary transfer device 2 in advance. In this case, the boundary transfer device 2 should acquire X chunks by time t ₂ +mt ₁ +t ₁ /2. Here, the round trip delay between the border transfer device ₂ and the server device 3 is t2. Therefore, the edge forwarding device 2 should transmit X interest packets for pre-acquisition processing by time (mt ₁ +t ₁ /2). Since the start time of the pre-acquisition process is t ₁ /2, the border forwarding device 2 transmits interest packets at approximately equal intervals between time t ₁ /2 and time (mt ₁ +t ₁ /2). By doing so, it is possible to suppress the occurrence of congestion due to bursty transmission of data packets.

That is, the edge forwarding device 2 averagely transmits X interest packets during the transmission period obtained by the period (mt ₁ +t ₁ /2−t ₁ /2)=mt ₁ . In other words, the edge forwarding device 2 sends X Interests with a period of mt ₁ /X. It is assumed that the boundary transfer device 2 measures the round-trip delay with the client device in advance.

Note that the number X of chunks acquired in the pre-acquisition process may be controlled based on the transmission speed S (upper limit of throughput) available on the route from the server device 3 to the boundary transfer device 2 . First, assume that the size of the chunks is constant and thus the size of the data packets is constant at D. Also, let X0 be the reference number (initial number) of chunks that the boundary transfer device ₂ acquires in the pre-acquisition process. Assuming that X ₀ interest packets are transmitted by the pre-acquisition process, the total size of the data packets acquired by the boundary transfer device 2 in response is X ₀ ×D. When this data packet is received in the period mt ₁ , on the route from the server device 3 to the border transfer device 2, traffic with a transmission speed of X ₀ ×D/mt ₁ is generated on average by pre-acquisition. The edge forwarding device 2 compares S with X ₀ ×D/mt ₁ , and if S≧X ₀ ×D/mt ₁ , transmits X ₀ interest packets by pre-acquisition processing. On the other hand, if S<X ₀ ×D/mt ₁ , the pre-acquisition process reduces the number of interest packets to be sent. The reduced number X may be one that satisfies S≧X×D/mt ₁ , eg, the maximum value of X that satisfies X≦S×mt ₁ /D.

With the above configuration, it is possible to suppress the occurrence of congestion due to pre-acquisition processing.

In this embodiment, the processing unit 22 determines whether or not the chunk name requested by the interest packet is to be processed by the pre-acquisition unit 21 . More specifically, one of the conditions for processing by the pre-acquisition unit 21 is that the chunk name requested by the interest packet indicates a foreign country. However, it is also possible to make it one of the conditions for processing by the pre-acquisition unit 21 that the name is not a country name but a name that indicates a certain region. Alternatively, one of the conditions for processing by the pre-acquisition unit 21 may be that the transfer destination interface indicated by the FIB is a predetermined interface instead of the chunk name. This predetermined interface is, for example, an interface that connects to a communication link with a transmission delay equal to or greater than a predetermined value.

Also, the processing unit 22 that operates as a transfer device and the pre-acquisition unit 21 that performs pre-acquisition processing can be implemented as separate devices.

FIG. 7 is another configuration diagram of the boundary transfer device according to this embodiment. The transfer unit 23 manages the FIB, PIT, and CS, transfers interest packets and data packets, and manages caches, like a normal transfer device.

The determination unit 24 monitors interest packets (hereinafter referred to as request packets) received from the client device. Then, when receiving a plurality of first request packets requesting the same chunk of requested content from the client device, the determination unit 24 determines whether or not to perform pre-acquisition processing of the requested content based on the plurality of first request packets. .

For example, the determination unit 24 determines that a plurality of first request packets are to be transferred and that the installation location of the server device 3 that publishes the requested content is a predetermined location (predetermined country or region). can be one of the conditions for determining that the pre-acquisition processing of the requested content is to be performed. The installation location of the server device 3 can be determined based on the chunk names requested by the plurality of first request packets.

In addition, the determination unit 24 determines that the communication link to which the plurality of first request packets are transferred is a predetermined communication link among the plurality of communication links connected to the boundary transfer device 2. It can be one of the conditions for determining to perform processing. The predetermined communication link may be, for example, a communication link with a transmission delay equal to or greater than a predetermined value, that is, a communication link equal to or greater than a predetermined length.

Further, the determination unit 24 can set the fact that the number of the plurality of first request packets received in the predetermined period is less than a predetermined value as one of the conditions for determining that the pre-acquisition processing of the requested content is to be performed. The predetermined period corresponds to the time from when the timer is started until the count value of the timer reaches a predetermined value (S30 in FIG. 5). If the number of first request packets received in a predetermined period is less than a predetermined value, it means that the window size of the client device is small. can do.

Further, the determination unit 24 determines that the maximum value of the chunk numbers requested by the plurality of first request packets received in the predetermined period is smaller than the threshold as a condition for determining that the pre-acquisition processing of the requested content is to be performed. can be one.

When the determination unit 24 determines to perform the pre-acquisition processing of the requested content, the transmission processing unit 26 selects a plurality of chunks of the requested content, each requesting a chunk different from the chunk requested by the plurality of first request packets. A second request packet is generated and transmitted via the transfer unit 23 . The number X ₀ of the plurality of second request packets is preset in the transmission processing unit 26 . At this time, the transmission processing unit 26 controls the transmission intervals of the plurality of second request packets based on the _round -trip delay t1 with the client device that has transmitted the plurality of first request packets. For example, the transmission processing unit 26 obtains a transmission period based on the _round -trip delay t1, and periodically transmits a plurality of second request packets over this transmission period. The transmission period can be, for example, a period obtained by multiplying the _round -trip delay t1 by an integer (m times). Note that the value of m is preset in the transmission processing unit 26 .

Also, the transmission processing unit 26 can make the number of the plurality of second request packets variable. In this case, the size D of each data packet and the upper limit value S of throughput are preset in the transmission processing unit 26 . For example, when S<X ₀ ×D/mt ₁ , the transmission processing unit 26 reduces the number of multiple second request packets from X ₀ . The reduced number X may be one that satisfies S≧X×D/mt ₁ , eg, the maximum value of X that satisfies X≦S×mt ₁ /D. With this configuration, it is possible to suppress the occurrence of congestion due to pre-acquisition processing. Note that the boundary transfer device 2 may acquire the upper limit value S of the throughput by measurement.

Note that, regarding the plurality of second request packets, chunks with numbers larger than the maximum number of chunks requested by the plurality of first request packets, that is, chunks later in time than the chunks requested by the plurality of first request packets. shall request chunks. For example, chunks requested by a plurality of second request packets range from a number obtained by adding 1 to the maximum number of chunks requested by a plurality of first request packets to a number obtained by adding the number of the plurality of second request packets. Can be chunked.

Further, the boundary transfer device 2 according to the present invention is a program that, when executed by the one or more processors of a device/computer having one or more processors, causes the device/computer to operate and function as the boundary transfer device 2. It can be realized by These computer programs can be stored in a computer-readable storage medium or distributed via a network.

With the above configuration, congestion can be suppressed in the pre-acquisition process. Therefore, it will be possible to contribute to Goal 9 of the Sustainable Development Goals (SDGs) led by the United Nations, "Build resilient infrastructure, promote sustainable industrialization, and foster innovation."

24: determination unit, 26: transmission processing unit

Claims (13)

A transfer device for a content distribution network that distributes content by dividing it into one or more chunks,
determining means for determining, when receiving a plurality of first request packets requesting chunks of requested content from a client device, whether or not to perform pre-acquisition processing of the requested content based on the plurality of first request packets;
when determining to perform pre-acquisition processing of the requested content, generating a plurality of second request packets each requesting a chunk different from chunks requested by the plurality of first request packets, among chunks of the requested content. a transmission means for transmitting the
with
The transfer device, wherein the transmission means controls transmission intervals of the plurality of second request packets based on a round-trip delay with the client device. 2. The transfer device according to claim 1, wherein said transmission means increases the transmission interval of said plurality of second request packets as the round-trip delay to said client device increases. 3. The transmitting means according to claim 1, wherein said transmission means periodically transmits said plurality of second request packets over a transmission period determined based on a round-trip delay with said client device. transfer device. 4. The transfer device according to claim 3, wherein said transmission period is a period obtained by multiplying said round-trip delay by an integer. 5. The transfer device according to claim 1, wherein the number of said plurality of second request packets is a predetermined value. The transmission means divides the total size of data packets received as a response to the transmission of the plurality of second request packets by the transmission period to be equal to or less than the upper limit value. 5. The transfer device according to claim 3, wherein the number of . 7. The transfer device according to claim 6, wherein said upper limit value is an upper limit value of throughput in a section from a server device that publishes said requested content to said transfer device. The judging means determines in advance that the plurality of first request packets are to be transferred and that the installation location of the server apparatus disclosing the requested content is a predetermined location. 7. The transfer device according to any one of claims 1 to 6, wherein the transfer device is one of the conditions for determining to perform the processing. The determination means determines that a communication link to which the plurality of first request packets are to be transferred is a predetermined communication link among a plurality of communication links connected to the transfer device, and performs the request content pre-acquisition process. 7. The transfer device according to any one of claims 1 to 6, wherein the transfer device is one of the conditions for determining to perform. The determining means is characterized in that one of the conditions for determining that the pre-acquisition process of the requested content is to be performed is that the number of the plurality of first request packets received in a predetermined period is smaller than a predetermined value. 10. The transfer device according to any one of claims 1 to 9. The chunk name of each of the one or more chunks into which the content is divided has a chunk number,
A condition for determining that the pre-acquisition process of the requested content is to be performed when the maximum value of the number of chunks requested by the plurality of first request packets received in the predetermined period is smaller than a threshold value. 11. The transfer device according to claim 10, wherein the transfer device is one of: 12. The transfer device according to claim 11, wherein the plurality of second request packets request chunks having a larger number than the maximum number of chunks requested by the plurality of first request packets. A program that causes a computer to function as the transfer device according to any one of claims 1 to 12.

Images