Network delay guarantee system and method based on global completion time keeping service sequence

Abstract

The invention relates to a network time delay guarantee system and method based on global completion time keeping a service sequence. According to the present invention, a network delay assurance system comprises: a packet processing unit configured to generate new metadata by storing an arrived packet and calculating a completion time of the stored packet; a scheduling unit configured to extract a packet having a minimum completion time by comparing completion times of the stored packets; and a data packet output unit configured to output the scheduled data packets through an output port.

Description

Technical Field

[0001] The present invention relates to a network delay guarantee system and method based on a global completion time that maintains service order (service order-preserving), and more specifically to a system and method for guaranteeing network delay. The system and method are based on global completion time information generated at a head node. Without managing the status information of each flow, the completion time of a data packet at a core node of the network is obtained and a schedule is set accordingly, thereby guaranteeing an upper limit on the network delay. Background Art

[0002] Since the 1990s, a technique has been proposed that treats flows as fluids (packets belonging to the same application, with the same source and destination, are treated as a whole) and, on this basis, provides all flows with exactly the requested amount of service through appropriate scheduling at relay nodes, thereby guaranteeing latency. Latency guarantee is closely related to flow protection. The degree of flow protection is inversely proportional to the upper limit of latency. There are three levels of flow protection techniques.

[0003] The first-level flow protection technology is a technology that limits the interference between flows to the maximum level of the packet size. When this technology is applied, the upper limit of the delay at each node is proportional to the maximum packet length. However, there are problems with the complexity of managing and recording a large amount of state information for each flow, so the first-level flow protection technology is not actually applied.

[0004] The second level flow protection technique does not maintain detailed individual states reflecting flow characteristics, but only records the fixed demand and service history of the flow and prescribes future services accordingly. When this technique is applied, the delay cap at each node is proportional to the sum of the maximum packet sizes of each flow.

[0005] The third level flow protection technology stores flows in queues grouped by priority and simply serves the queues by priority. When this technology is applied, the upper limit of latency is determined by the sum of the maximum burst sizes of each flow. The maximum burst size refers to the total amount of data that each flow is allowed to send from the original source at one time. The maximum burst size usually represents the concept of a large set of packets that can be allowed. Therefore, in some cases, it may be difficult to accept an upper limit of latency that is proportional to the sum of packets. In addition, when there are loops in the network topology of the network to which these technologies are applied, the maximum latency is amplified when passing through the node, and it may be possible that the upper limit of latency cannot be guaranteed, depending on the level of network utilization.

[0006] Therefore, methods of applying the best performing first-level techniques to the Internet have been studied. Generalized processor sharing (GPS) provides a paradigm for fair service of flows as fluids, and packetized GPS (PGPS, or weighted fair queues) that implement this paradigm in a packet environment have played a pioneering role in such packet-based schedulers. These are collectively referred to as fair queue schedulers. Fair queues determine the order in which packets are served in ascending order of completion time obtained by the following formula 1.

[0007] [Formula 1] F(p)=max{F(p-1),V(A(p))}+L(p) / r

[0008] Here, p is the pth packet of the flow, A(p) is the time when packet p arrives at the node, L(p) is the length of p, and r is the service rate assigned to the flow to which p belongs. V(t) is called the virtual time function and can be calculated in a variety of ways, such as the product of the actual time t and the ratio of the sum of r of the flows being served and the link capacity, or the current time. Virtual time prevents unfair situations, that is, when the service speed of the flow is faster than the allocation speed, the flow that starts later has a smaller completion time and is thus provided with better service than the existing flow for a considerable time. F(p) represents the completion time of p calculated fairly, and the nodes receive service in the order of the smallest value. The completion time can be calculated when the data packet arrives at the node, so it can be recorded and used in the node as metadata of the data packet before it is stored in the buffer. In general, a queue is provided for each flow, and the queue is managed in a FIFO or PIFO manner, and the scheduler is implemented in the form of checking the queue header (Head of Queue, HoQ) of each flow and serving the queue with the shortest completion time. Alternatively, all packets can be put into one queue and the packets can be inserted into the middle of the queue based on the completion time value. The completion time is the expected completion time when the packet is fairly served. The key point of formula 1 is that in the worst case, that is, when all flows are activated and the link is fully utilized, the service is provided with an interval of L(p) / r compared to the previous packet belonging to the flow. At the same time, by using a work-keeping scheduler, link resources can be prevented from being idle and wasted.

[0009] In order to calculate Formula 1, the F(p-1) of the flow, that is, the completion time of the previous packet, must be remembered. When a packet is received, it is necessary to find out which flow the packet belongs to and the completion time of the most recent packet of the corresponding flow. The completion time F(p-1) of the most recent packet is a value that represents the so-called "flow state". For core nodes that manage millions of flows, the fact that state information must be remembered and read implies considerable complexity, which has become the main reason why the fair queue scheduler is not actually used on the Internet. It is impossible to manage millions of flow states in real time in core nodes.

[0010] To solve this problem, a method has been proposed in which, without managing the state information of the flow at the core node, the flow state can be obtained at the core node using only the packet state by allowing the necessary information to be written at the head node and the necessary information to be modified.

[0011] However, previous studies have attempted to reflect the time information required for packet service on the completion time at each node, which means that the service order of packets of different sizes in the same flow may be reversed.

[0012] To prevent this, the concept of eligible time for data packet service is introduced to guarantee the service order between data packets. Therefore, the data packets are operated in a non-work-holding mode. This can guarantee the upper limit of the delay, but the disadvantage is that the average delay will increase significantly, so a solution to this problem is needed.

[0013] Meanwhile, a method of recording such information associated with a data packet in a data packet header and modifying the information at each node has not been accepted as an Internet standard due to its complexity. Summary of the invention Technical issues

[0014] The technical purpose to be achieved by the present invention is to provide a system and method for ensuring network delay. The system and method are based on the global completion time information generated at the head node. Without the need to manage the status information of each flow, the completion time of the data packet is obtained from the network core node and a schedule is set accordingly, thereby ensuring the upper limit of the network delay. Technical Solution

[0015] According to an embodiment of the present invention, a network delay guarantee system based on a global completion time maintaining a service order includes: a packet processing unit, which is configured to generate new metadata by storing arriving packets and calculating the completion time of the stored packets; a scheduling unit, which is configured to extract a packet with a minimum completion time by comparing the completion time of the stored packets; and a packet output unit, which is configured to output the scheduled packets through an output port.

[0016] The scheduling unit may extract a data packet with a minimum completion time by storing a plurality of flow sets classified according to a predefined standard in a separate FIFO (First In First Out) queue and by comparing completion times of Head of Queue (HoQ) of the plurality of FIFO queues with each other.

[0017] Here, in the predefined criteria, flows having at least one of the maximum burst size, the maximum data packet size and the average service rate being the same within a preset error range may be classified.

[0018] The scheduling unit may use a priority queue to extract the data packet with the smallest completion time.

[0019] The data packet processing unit may adopt a fair queue method and use the following formula to calculate the completion time of the first node.

[0020] F 0 (p) = max{F 0 (p-1),A 0 (p)}+L(p) / r

[0021] Here, F 0 (p) is the completion time calculated when data packet p enters the first node 0, A 0 (p) represents the actual time or virtual time when the data packet p enters the head node 0, L(p) represents the length of the data packet p, and r represents the service rate allocated to the flow to which the data packet p belongs.

[0022] The packet processing unit can use the following formula to calculate the completion time at the core node h (F h (p)).

[0023] F h (p) = F h-1 (p)+d h (p)

[0024] Here, d h(p) represents a function of a core node h and a data packet and is defined as an increment of a completion time calculated based on the core node h, wherein the core node h and the core node h-1 refer to nodes through which the data packet p passes, and the data packet p passes through the core node h immediately after passing through the core node h-1.

[0025] In addition, h (p) can be calculated based on the value W h-1 (p) and U h-1 (p) is determined by the rules defined between h-1 (p) represents the minimum delay of the data packet p at the core node h-1, and U h-1 (p) represents the maximum delay that the data packet p can experience at the core node h-1.

[0026] In addition, h (p) can be determined according to the following formula.

[0027]

[0028] here, represents the maximum packet length of all flows in the core node h-1, R h-1 represents the link capacity in the core node h-1, L i represents the maximum packet length of flow i, and r i Represents the service rate allocated to the flow i to which the data packet p belongs.

[0029] At the same time, h (p) can be calculated based on the value W h-1 and U h-1 The rules defined between them determine, W h-1 represents the minimum delay of any data packet in the core node h-1, and U h-1 It represents the maximum delay that any data packet can experience in the core node h-1.

[0030] According to another embodiment of the present invention, a network delay guarantee method using a network delay guarantee system includes: a step of generating new metadata by storing arrived data packets and calculating the completion time (FT) of the stored data packets; a step of extracting a data packet with a minimum completion time by comparing the completion times of the stored data packets with each other; and a step of outputting the scheduled data packets through an output port. Beneficial Effects

[0031] In this way, according to the present invention, the completion time obtained from the head node is recorded in the data packet as metadata, and based on this completion time, the completion time can be updated in a simple manner in the downstream node without managing the state information of each flow, so that the scheduling order of the data packet can be determined. Therefore, the upper limit of the network delay can be guaranteed without managing the flow state information.

[0032] Furthermore, according to the present invention, the scheduler has a characteristic of operating by adding only node-specific status information to the completion time at the head node, and therefore, the service order of data packets of all flows passing through the same path can be kept unchanged, and there is an advantage in that the data packet with the minimum completion time can be easily found out simply by comparing the HoQs of a plurality of FIFO queues or by using a priority queue. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a configuration diagram showing a network delay guarantee system according to an embodiment of the present invention.

[0034] Figure 2 is a flow chart illustrating a network delay guarantee method using a network delay guarantee system according to an embodiment of the present invention.

[0035] Figure 3 Reference Figure 2 Figures illustrating example implementations.

[0036] Figure 4 is a diagram showing a network topology for performance comparison.

[0037] Figure 5 is the maximum end-to-end delay of type C flow and d h A three-dimensional (3D) surface plot of the relationship between and utilization.

[0038] <Description of Reference Numerals>

[0039] 100: Delay guarantee system

[0040] 110: Packet processing unit

[0041] 120: Scheduling unit

[0042] 130: Data packet output unit. DETAILED DESCRIPTION

[0043] The embodiments of the present disclosure will be described in detail below in conjunction with the accompanying drawings so that those skilled in the art can easily implement the present disclosure. However, the present disclosure can be implemented in various different forms and is not limited to the embodiments described herein. In addition, in order to clearly illustrate the present disclosure in conjunction with the accompanying drawings, parts that are not related to the description are omitted, and similar parts are given similar reference numerals throughout the specification.

[0044] Throughout the specification, when a part is said to “include” a certain component, it does not mean to exclude other components but may further include other components unless explicitly stated otherwise.

[0045] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement the present disclosure.

[0046] In the following, reference will be made to Figure 1 The delay guarantee system according to the embodiment of the present invention is described in more detail.

[0047] Figure 1 is a configuration diagram showing a network delay guarantee system according to an embodiment of the present invention.

[0048] like Figure 1 As shown, the network delay guarantee system 100 according to the embodiment of the present invention includes a data packet processing unit 110 , a scheduling unit 120 and a data packet output unit 130 .

[0049] First, the packet processing unit 110 stores the arrived packet, calculates the completion time (FT) of the stored packet, and generates new metadata. Then, the packet processing unit 110 stores the generated metadata in the header of the packet.

[0050] The scheduling unit 120 compares the completion times of the stored data packets to extract the data packet with the minimum completion time. The scheduling unit 120 adjusts the output order of the data packets by using the priority information of the data packet traffic and the minimum completion time.

[0051] Here, the scheduling unit 120 may store a plurality of flow sets classified according to a predefined criterion in a separate first-in-first-out (FIFO) queue, and compare the completion time of the headers (head of queue (HoQ)) of the FIFO queues to extract a data packet with a minimum completion time. In this case, the predefined criterion may classify flows that have the same maximum burst size, maximum data packet size, and average service rate within a predefined error range.

[0052] Alternatively, the scheduling unit 120 may use a priority queue to extract the data packet with the smallest completion time.

[0053] Finally, the data packet output unit 130 may output the scheduled data packets through an output port.

[0054] The following will refer to Figure 2 and Figure 3 The delay guarantee method using the delay guarantee system according to the embodiment of the present invention is described in more detail.

[0055] Figure 2 is a flowchart showing a network delay guarantee method using a network delay guarantee system according to an embodiment of the present invention, and Figure 3 Reference Figure 2 Figures illustrating example implementations.

[0056] like Figure 2 and Figure 3 As shown, the data packet processing unit 110 classifies the incoming data packets according to the priorities of the traffic (S210).

[0057] The packet processing unit 110 classifies all traffic into high priority traffic and low priority traffic, and classifies data packets into data packets having high priority traffic and data packets having low priority traffic.

[0058] Then, the packet processing unit 110 acquires the completion time based on the metadata included in the header of the packet (S220).

[0059] The network delay guarantee system 100 according to the embodiment of the present invention uses the completion time information obtained from the head node to obtain the completion time in the downstream node, rather than obtaining a new completion time for each node.

[0060] Therefore, the data packet processing unit 110 obtains the completion time of the current node using the function of the completion time of the previous node and the data packet contained in the metadata.

[0061] To illustrate this point again, the packet processing unit 110 calculates the completion time of the head node by adopting the fair queue method represented by Formula 2.

[0062] [Formula 2] F 0 (p) = max{F 0 (p-1),A 0 (p)}+L(p) / r

[0063] Here, F 0 (p) is the completion time calculated when data packet p enters the first node 0, A 0 (p) represents the actual time or virtual time when the data packet p enters the head node 0, L(p) represents the length of the data packet p, and r represents the service rate allocated to the flow to which the data packet p belongs.

[0064] Meanwhile, in the embodiment of the present invention, the completion time calculated at the head node is referred to as the global completion time.

[0065] In addition, the packet processing unit 110 calculates the completion time F at the core node h by the following formula 3:h (p).

[0066] [Formula 3] F h (p) = F h-1 (p)+d h (p)

[0067] Here, d h (p) represents a function of core node h and data packet, and is defined as the completion time increment calculated based on core node h. Core node h and core node h-1 refer to the nodes that data packet p passes through, and data packet p passes through core node h immediately after passing through core node h-1. At the same time, d h (p) must be a non-decreasing function between adjacent data packets to prevent the service order between adjacent data packets from changing. h The value of (p) can be realized by measuring the maximum delay from the beginning of the node's busy period to the entry time of packet p.

[0068] When using the above formula 2 and formula 3 to calculate the completion time at the core node h, only F is needed h-1 (p) and d h (p), so there is no need to store and manage the state information of each flow at the core node. h-1 (p) and d h The information of (p) can be stored in the form of metadata in the header of the data packet to be transmitted. Although the update is required at all nodes, it is allowed to be updated at an appropriate time between the arrival time and departure time of the data packet, thus giving sufficient time.

[0069] Furthermore, according to an embodiment, d h (p) According to W h-1 (p) and U h-1 (p) is determined by the rules defined between h-1 (p) represents the lower bound of the delay of data packet p at the core node h-1, and U h-1 (p) represents the maximum delay that data packet p can experience at core node h-1.

[0070] The core node h-1 is the node that the data packet p passes through just before reaching the node h.

[0071] According to the embodiment, d h (p) is determined according to the following formula 4.

[0072] [Formula 4]

[0073] here, represents the maximum packet length of all flows in core node h-1, R h-1 represents the link capacity in the core node h-1. In addition, L i represents the maximum packet length of flow i, and r i represents the service rate assigned to flow i to which packet p belongs.

[0074] According to another embodiment, d h (p) According to W h-1 (p) and U h-1 (p) is determined by the rules defined between h-1 represents the lower bound of the delay of any data packet in the core node h-1, and U h-1 It represents the maximum delay that any data packet can experience in the core node h-1.

[0076] Compared to Formula 2, for each node, the completion time calculated as described above is calculated without increasing L(p) / r. This reflects the fact that it is fair to treat the entire network as one node and keep the set initial distance between packets.

[0077] According to the above embodiments of the present invention, the change of the service order between data packets in the same path in the middle of the path can be reduced to a minimum. In addition, the scheduling based on completion time, which originally requires a general complex sorting algorithm, can be easily implemented by simply comparing the HoQ of multiple FIFO queues or using a priority queue to find the data packet with the minimum completion time among all stored data packets. The scalability of the core node can be obtained through this simple scheduler configuration.

[0078] At the same time, h (p) can be determined as U h-1 , and the completion time F h (p) can be calculated based on d h The value of affects the fairness between flows that reach the core node through different paths. Therefore, in order to solve this problem, F 0 (p)≥V(A 0 In other words, fairness among flows is ensured by associating the completion time F of the flows with the virtual time t.

[0079] In the embodiment of the present invention, d h (p) must be equal to the guaranteed delay U for packet p at node h-1 h-1 In other words, the upper limit of the delay satisfies the following formula 5.

[0080] [Formula 5] A h(p)+U h (p)≥A h+1 (p)

[0081] As described above, when the calculation of the completion time is completed, the packet processing unit 110 stores the calculated completion time in the header of the packet in the form of metadata.

[0082] When step S220 is completed, the scheduling unit 120 extracts the data packet with the smallest completion time using the completion time stored in the header of the data packet, and preferentially schedules the extracted data packet (S230).

[0083] In other words, the scheduling unit 120 uses the priority information and the minimum completion time of the packet traffic to adjust the output order of the data packets.

[0085] Finally, the data packet output unit 130 adjusts the interval between the transmitted data packets and outputs the data packets through the output port (S240).

[0086] In other words, the data packet output unit 130 receives the data packet output from the scheduling queue according to the scheduling result in the scheduling unit 120 , and outputs the data packet to the output port.

[0087] In the following, reference will be made to Figure 4 and Figure 5 The performance comparison results of the network delay guarantee system according to the present invention are described in more detail.

[0088] Figure 4 is a diagram showing a network topology for performance comparison, and Figure 5 is the maximum end-to-end delay of class C flow and d h 3D surface plot of the relationship between load and utilization.

[0089] First, if Figure 4 As shown, nodes 1, 2, 3, 13, 14, and 15 are ingress nodes directly connected to the source, and the other nodes are core nodes. Since the output port of the core node has a queue for each input port, the output port includes two queues. Unless otherwise specified, this queue is based on PIFO operation. Packets are sorted according to their completion time when they are added to the queue. The ingress node maintains the state of each flow and the FIFO queue for each flow. The link capacity of all links in the topology is 1 Gbps. Figure 4 In the figure, arrows indicate the flow direction.

[0090] The source generates a flow for each destination, so 36 flows are generated for all networks. Tables 1 and 2 below show the characteristics of three different flow types used for simulation. The flow type is determined by the destination of the flow. For example, all flows with node 1 as the destination are class A. Each destination has 6 flows, and each type has 12 flows. Here, the flow type can be classified according to at least one of the maximum burst size, the maximum packet size, and the allocated service rate. Specifically, flows that have at least one of the maximum burst size, the maximum packet size, and the average service rate that is the same within a preset error range can be classified. The flow generates packets of various lengths from 1Kbit to 10Kbit in units of 1Kbit.

[0091] [Table 1] Stream Type Maximum burst size Packet size destination A 200Kbit lK-10Kbit 1、6 B 200Kbit lK.10Kbit 3、4 C 20Kbit lK, 2Kbit 2、5

[0093] [Table 2]

[0094] The maximum end-to-end delay observed from each flow type is compared to each other. The flow with the longest path in the same flow type is worth noting. Table 3 shows the flow paths with the longest path for each flow type. The number of hops of the longest path is the same for all flow types. The utilization of each path can vary from link to link.

[0095] [Table 3]

[0096] In this simulation, d h is the same for all nodes. In Figure 6, each plane represents the maximum, average, and minimum values ​​of the observed maximum end-to-end delay. h As d increases, the maximum end-to-end delay also increases. This is because among flows of the same type, the flow with the largest number of hops determines the maximum end-to-end delay. h is a fixed value, so the completion time of packets with more hops increases, resulting in larger delays regardless of the path. Moreover, this effect can be observed between different types of flows. h The phenomenon will be more obviously observed with the increase of .

[0097] When h When d is 0, h When is the minimum value, the first generated packet has the advantage. Packets that travel more hops have a more advantageous completion time than packets that travel fewer hops.

[0098] In this way, the delay guarantee system according to the present invention records the completion time obtained from the head node as metadata in the data packet, and based on this, the completion time can be updated in a simple manner in the lower-level nodes without managing the status information of each flow, so that the data packet scheduling order can be determined.

[0099] Furthermore, the delay guarantee system according to the present invention is characterized in that the scheduler operates by adding only node-specific status information to the completion time at the head node, and therefore, the service order of data packets of all flows passing through the same path can be kept unchanged, and the data packet with the minimum completion time can be easily found by only comparing the HoQs of multiple FIFO queues or using a priority queue.

[0100] The present invention is described with reference to the embodiments shown in the accompanying drawings, but these embodiments are merely examples, and those skilled in the art will appreciate that various modifications and other equivalent embodiments can be obtained therefrom. Therefore, the true technical protection scope of the present invention should be determined by the technical ideas of the attached patent claims.

Claims

1. A network delay guarantee system based on global completion time maintaining service order, the network delay guarantee system include: a packet processing unit configured to generate new metadata by storing the arriving packets and calculating the completion time of the stored packets; a scheduling unit configured to extract a data packet having a minimum completion time by comparing the completion times of the stored data packets; as well as The data packet output unit is configured to output the scheduled data packets through an output port.

2. The network delay guarantee system according to claim 1, in, The scheduling unit extracts a data packet with a minimum completion time by storing a plurality of flow sets classified according to a predefined standard in separate FIFO (First In First Out) queues and by comparing completion times of Head of Queue (HoQ) of the plurality of FIFO queues with each other.

3. The network delay guarantee system according to claim 2, in, In the predefined criteria, flows having at least one of a maximum burst size, a maximum data packet size and an average service rate that is the same within a preset error range are classified.

4. The network delay guarantee system according to claim 1, in, The scheduling unit uses a priority queue to extract the data packet with the smallest completion time.

5. The network delay guarantee system according to claim 1, in, The data packet processing unit uses the fair queue method to calculate the completion time of the first node using the following formula: F 0 (p)=max{F 0 (p-1),A 0 (p)}+L(p) / r, Among them, F 0 (p) is the completion time calculated when data packet p enters the first node 0, A 0 (p) represents the actual time or virtual time when the data packet p enters the head node 0, L(p) represents the length of the data packet p, and r represents the service rate allocated to the flow to which the data packet p belongs.

6. The network delay guarantee system according to claim 1, in, The packet processing unit calculates the completion time at the core node h using the following formula: h (p)): F h (p)=F h-1 (p)+d h (p), Among them, d h (p) represents a function of a core node h and a data packet and is defined as a completion time increment calculated based on the core node h, wherein the core node h and the core node h-1 refer to nodes passed by the data packet p, and the data packet p passes through the core node h immediately after passing through the core node h-1.

7. The network delay guarantee system according to claim 6, in, d h (p) According to the value W h-1 (p) and U h-1 (p) is determined by the rules defined between h-1 (p) represents the minimum delay of the data packet p at the core node h-1, and U h-1 (p) represents the maximum delay that the data packet p can experience at the core node h-1.

8. The network delay guarantee system according to claim 7, in, d h (p) is determined according to the following formula: in, represents the maximum packet length of all flows in the core node h-1, R h-1 represents the link capacity in the core node h-1, L i represents the maximum packet length of flow i, and r i Represents the service rate allocated to the flow i to which the data packet p belongs.

9. The network delay guarantee system according to claim 6, in, d h (p) According to the value W h-1 and U h-1 The rules defined between them determine, W h-1 represents the minimum delay of any data packet in the core node h-1, and U h-1 It represents the maximum delay that any data packet can experience in the core node h-1.

10. A network delay guarantee method using a network delay guarantee system, the network delay guarantee method include: a step of generating new metadata by storing arriving packets and calculating a completion time (FT) of the stored packets; A step of extracting a data packet having a minimum completion time by comparing the completion times of the stored data packets with each other; as well as The step of outputting the scheduled data packets through an output port.

Images