JPH1051472A - Method and device for transmitting and receiving packet - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reserve irreducible bandwidth when connection is established and perform transmission and reception, and improve the efficiency of a network by scheduling received packets at a packet network node under control over an absolute time according to a speed corresponding to the reserved bandwidth. SOLUTION: A dedicated minimum service connection is expected with a queue 101 specified by a search mechanism 100, the connection queue is scheduled by a shaper unit 111 according to the speed corresponding to the reserved bandwidth by controlling the absolute time 190, and a queue identifier is stored. Then the identifier is read out and the starting packet of the corresponding queue is transmitted. While there is a free cell space, a priority mechanism function 178 performs ready identifier queue resetting out of a queue group 174, also preforms queue resetting from a minimum service connection queue 15 corresponding to the identifier, and fills a cell space 170 with ready cells.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates generally to network communication systems and, more particularly, to an ATM network node that supports services that guarantee a minimum bandwidth for user connections.

[0002]

2. Description of the Related Art Asynchronous transfer mode (ATM) has become the basic technology of the next generation of high-speed networks. High-speed networks have different traffic requirements and quality of service (Q
oS (Quality of Service) supports a wide variety of requirements. Such diversity requires different control flow policies. For example, certain applications, such as multimedia and time-critical applications, require guaranteed levels of delay and throughput, but can tolerate losses while LA
Other applications, such as N traffic, can tolerate delay and throughput variations, but are very sensitive to loss.

[0003] ATM Reserved Bandwidth (RB)
idth) service requires that the user establish a traffic contract with the network during call setup before transmitting data. The contract includes a specification of the desired QoS class and a set of traffic descriptors. The network either gives the desired QoS to the ATM connection or rejects the call depending on the resource allocation. Bandwidth allocated between source and destination is below peak speed to take advantage of statistical multiplexing gain, but requires complex software, and causes traffic congestion and data loss Sometimes.

[0004] In a LAN environment, the source is bursty and unpredictable. Traffic has very high variability on time scales over several magnitudes. For such unpredictable sources, a peak rate can be allocated to avoid excessive losses in the network. However, because bandwidth is reserved even during idle periods,
Use of the network is inefficient.

One way to increase the utilization of links to the network is to use non-reserved bandwidth (NRB).
ed Bandwidth) by adding a service class. In this class of service, no bandwidth is reserved and the source can get as much of the remaining bandwidth as possible to make the best possible transmission and impact RB traffic. Has no effect. Therefore, resources are not pre-allocated and NRB services require a control flow approach to control the source. The quality of the flow control mechanism limits the efficiency of NRB service support in terms of call loss rates. The hopping congestion flow control mechanism disclosed in European Patent Application No. EP94480125, implemented at a network node on the connection path, is an improved flow control mechanism. Selective (per connection) or global backpressure commands sent from neighboring nodes cause the incoming traffic to be selectively or globally resumed or throttled at each hop of the connection path.

However, the NRB connection requires the ATM Forum Technical Committee in the Traffic Man
agement Working Group).
There is no guarantee that services from the TM network will be available. The need for new service categories has been identified, and the available bit rate (ABR: Availa
ble Bit Rate) is provided between the reserved bandwidth service and the non-reserved bandwidth service. In the case of an ABR connection, the terminating system will specify both the maximum required bandwidth (Peak Cell Rate (PCR)) and the minimum required bandwidth (MCR: Minimum Cell Rate) for the network. . The ABR service defined by the ATM Forum also includes a flow control protocol between source and destination. The terminating system will adapt the traffic according to the feedback received from the flow control mechanism. The expected characteristics of the ABR service are guaranteed minimum bandwidth, reasonable sharing of available bandwidth,
And, in the case of the flow control protocol, the cell loss rate is low. It is not intended to support real-time applications, as this service cannot control cell transfer delays. This is the same as the NRB service, but with guaranteed QoS, critical data transfer (such as defense information), supercomputer applications, and good delay behavior (distributed file transfer, remote procedure call, computer Applies to applications where any data telecommunications (such as process swap / paging, etc.) are required (for NRB services, e-mail, telex, fax, banking transactions, LAN interconnects, etc.).

[0007] The cell loss rate of the ABR service depends on the quality of the flow control, but the ABR service with respect to the guaranteed minimum bandwidth per connection and the proper sharing of the available bandwidth.
The expected characteristics of the service depend on the quality of the scheduling method. In the industry of ATM network nodes where this ABR service is beginning to emerge, there is a need to provide a scheduling technique that guarantees minimum bandwidth and can support a reasonable sharing of available bandwidth. This so-called "minimum service" support requires a minimum bandwidth during connection setup and ensures that each accepting connection (MCR) has this minimum bandwidth. Depending on the scheduling method implemented at the network node of the connection path, a minimum bandwidth can be guaranteed on the path bandwidth-wise or constantly and at each node of the network node. Both methods of ensuring minimum bandwidth are tailored to the MCR parameters. Second of minimum service support
Is a reasonable sharing of the "remaining" bandwidth. Proper sharing of bandwidth includes bandwidth that is not reserved for minimum service traffic and some of the bandwidth that is reserved but not used by minimum service traffic. In fact, if the scheduling approach that supports the minimum service can be combined with a network flow control mechanism such as ABR flow control, the resulting support will improve the call loss rate. The better the efficiency of network flow control, the lower the call loss rate.

[0008]

SUMMARY OF THE INVENTION It is a first object of the present invention to provide a scheduling scheme that supports a minimum service connection such as an ABR connection, the minimum available bandwidth for each connection and the number of connections between these connections. The goal is to ensure proper sharing of the remaining bandwidth.

A second object of the invention is that the minimum bandwidth of each minimum service connection is actually provided, if needed, independently of the termination rate of each other connection supported by that node. Is to be done. More specifically, the minimum bandwidth allocated to each minimum service connection is
It is also guaranteed if the aggregated incoming traffic exceeds the output capability.

A third object of the present invention is to provide RB (CBR, VBR in the case of an ATM forum) or NRB (ATM
UBR in the case of a forum) or any mixed traffic including a minimum service connection such as an ABR connection defined in the ATM forum,
Maintaining the same characteristics or QoS as expected for each type of service, as if they were being serviced alone, especially maintaining a low CLR for ABR services.

[0011]

SUMMARY OF THE INVENTION The present invention is a method for transmitting and receiving packets for a connection having the minimum bandwidth required at the time of establishing a connection in a packet network node. Receiving; scheduling the connection queue under control of an absolute time according to a rate corresponding to the reserved bandwidth; storing a queue identifier of the scheduled connection queue; Reading the connection queue identifier and transmitting the first packet in the connection queue corresponding to the read identifier if any identifier is found; and if no identifier is stored, Schedule a connection queue and send the first packet if any packets are found The method comprises the steps disclosed.

In particular, the method of the present invention can be used for ATM networks where the packets are ATM cells.

[0013] The present invention also discloses an apparatus for performing the steps of the above method, which may be used for an adapter of a network node.

According to the solution of the present invention, dual scheduling is proposed. The first scheduler guarantees a minimum bandwidth when required by a minimum service connection,
Complementary scheduling shares the remaining bandwidth between connections and does not affect scheduling for minimum bandwidth. Further, the first scheduling can handle the RB traffic according to the QoS required by the ATM Forum. In parallel, complementary scheduling can also handle NRB traffic, and if the scheduler is a round robin scheduler, it will reasonably share the bandwidth between the NRB connection and the minimum service connection. Combined with the ABR feedback mechanism, the scheduling scheme of the present invention supports ATM Forum ABR service. Complementary schedulers can be submitted to network control flow commands such as prior art hopping mechanisms to provide lossless service to NRB and minimum service support.

In the case of the inventive scheduling, the minimum bandwidth is always taken into account at each node point of the connection path. Regarding the minimum bandwidth required by the minimum service connection when establishing a connection, the minimum bandwidth is a "deterministic" aspect in the network, but the "statistical"
Is not given in a manner.

[0016]

FIG. 1 shows the support of two ATM services, RB service and NRB service. These services are supported by independent scheduling techniques, with RB service scheduling taking precedence over NRB service, NRB service scheduling is activated when RB service is not active, and various NRB services are This results in sharing of the remaining bandwidth between the connections. Returning to the embodiment of FIG. 1, the ATM label of the incoming cell is first searched to determine which service the cell belongs to (1
00). The set of RB connection queues such as 140 and NRB connection queues such as 155 are managed by independent scheduling techniques. The cells of the NRB connection are queued in an intermediate queue (101) to allow for different processing times of the NRB traffic. As an example,
Only if the NRB queues 155 are not empty,
It may be readable by a second of the two independent scheduling techniques. While this latter operation improves the efficiency of the second scheduling approach,
Increase the queuing time when queuing cells to empty queue 155. If the media speed cannot be maintained due to additional processing, an intermediate queue 101 is added. This embodiment is chosen as the preferred embodiment. The queuing process in the queue 140 for RB traffic and the queue 101 for NRB traffic is performed at the medium speed. If the intermediate queue 101 does not exist, the queuing process in the queues 140 and 155 is performed at the medium speed. The RB scheduling scheme has priority and N
You can interrupt the RB's. The scheduling method for the RB service shown in FIG. 1 operates as follows. An optional policer (110) checks the incoming cell for a traffic contract established during call setup. Non-conforming cells are discarded as needed. An optional shaping function (111) reschedules the departure of the cells according to the traffic contract and reduces any CDV of the incoming cell stream accepted by the policer according to the Cell Delay Variation (CDV) tolerance setting. . The policing function is triggered by the arrival of cells, while the shaping function (based on a leaky bucket algorithm or departure calendar) uses absolute timing intercell spacing (Tics).
Triggered by 190. When a connection is formed, a corresponding connection queue (140) is scheduled by the shaper. If the queue is not empty, the head cell of the queue is dequeued from the connection queue 140 and enqueued to the RB service queue 160;
Transmitted on the output line. If the connection is not formed, the cells of this connection are queued directly to the RB service queue (160). Queue 160 only serves as a pipeline unit between the shaper and the output line. The cell transfer function (120) empties the RB service queue 160 and sends the cell on the output line. If the RB cell is empty, the cell transmission function transmits a cell for NRB service if such a cell is ready in NRB cell space 170. The scheduling approach for mixed NRB and RB service support is also shown in FIG. The scheduler of the NRB service (130) selects a non-empty connection queue (a round robin scheduler is an example of scheduling) and moves the first cell of the connection queue to the NRB cell space (170), Transmit. As with the processing of NRB traffic, the intermediate queue 101 allows for variable processing time for the minimum service and maintains the media speed at an average value.
The processing of the NRB service can be immediately stopped by the operation by the priority operation (147) serving the RB traffic (145). Optionally, a flow control mechanism such as that of the prior art is used for NRB traffic (19
5) with a selective backpressure command (some NRB connections) or a global pack pressure command (all NRB connections) applied to it. The selective backpressure command selects or deselects one of the round robin scheduler connection queues, and the global backpressure command stops or restarts the round robin scheduler (130).

Using the optional shaping and policing functions of prioritized scheduling of RB traffic in combination with call control rather than efficiency α, which is not part of the present invention, the quality of service of the RB traffic can be very low (10 ^{−). It} features a maximum cell delay transfer of less than ²⁰ ) and, if relevant, a guaranteed maximum peak-to-peak CDV for the traffic. N
The characteristic of the RB service is the scheduler type (130).
Is determined by Round robin scheduling follows a well-known "maximum-minimum algorithm" criterion to ensure a reasonable sharing of the remaining bandwidth between all NRB connections. FIF with simple NRB scheduler
In the case of O, compliance is not observed. In particular, as already mentioned, the cell loss rate is determined by its quality when using a flow control mechanism. Prior art hopping flow control guarantees lossless NRB service.

Using two prioritized scheduling mechanisms to support the mixed traffic generated by the RB and NRB connections, the dual scheduling mechanism shown in FIG. 2 provides a network node along the connection path and a minimum at each node. Enable service connection support. As identified by the ATM search mechanism (100), the cell with the least service connection is placed in the intermediate queue (10
If 1) is implemented, it is queued to this and then to the dedicated minimum service connection queue (150). In accordance with the dual scheduling mechanism of the present invention, the connection queue of least service connections is first scheduled by the shaper unit (111), which ensures that each connection gets at least at its minimum bandwidth. Any type of shaper, such as a leaky bucket or calendar, can be used. Complementary scheduling techniques share the available bandwidth remaining during the connection. This residual bandwidth is the bandwidth not served by the shaper. Cell dequeuing of the minimum service traffic is avoided by transmitting the minimum bandwidth traffic and the remaining shared bandwidth traffic in the same manner, ie, in cell space 170.
The cell space 170 of the preferred embodiment is a flip-flop, a limited space that smooths the scheduled output queue. Shared assets (minimum service connection queues such as 150 and cell space 17
The problem of the connection accessing 0) is that the least service connection scheduled minimum service connection queue ensures minimum service traffic of the least service connection if the least service connection queue (150) is not empty. A queue group (1) in which an identifier is queued by the shaper
74) is solved by adding a queue.
In a preferred embodiment, the shaper uses absolute time (1
90), the complementary scheduling method operates whenever the shaper has access to resources shared with the shaper (storage device containing the connection control block). Complementary scheduling is a shaper,
It is momentarily stopped (145) by operations from areas of priority operation, such as ATM cell search (100), cell transmission (120) and queuing of incoming traffic (101, 140). Each time complementary scheduling is activated, and at the same time that cell space (170) is not full, the priority mechanism function (178) will return if the queue of the queue group is not empty (minimum traffic is being serviced). Next, the ready identifier is dequeued from the queue of the queue group 174. Lady Cell then
Minimum service connection queue 150 corresponding to this identifier
And the cell space 170 is filled with ready cells. If the queue of the queue group is empty, the least service connection selected by complementary scheduling is the round robin scheduler (13
0). The scheduler of the preferred embodiment of the present invention provides for a graceful sharing of the available bandwidth between queues 150. However,
According to the object of the present invention, instead of round robin α,
Any scheduler can be used.

FIG. 3 shows a mixed environment including minimum service connections and support for RB and NRB connections according to the solution of the present invention. ATM label search function (1
If the cell identified by (00) is the cell with the least service connection, it is buffered in the intermediate queue (101) and then queued in the connection queue (150) reachable by the scheduler (130) and the shaper (111). Be transformed into
The minimum service queue is then scheduled by the same components as described with respect to FIG. If the cell is identified by the ATM label search function, depending on whether the RB connection is shaped (the shaper is optional for the RB connection), whether the RB cell is queued in the connection queue 140 , Or directly to the RB service queue 160. RB traffic is scheduled as described in FIG. The total reserved bandwidth of RB traffic and the total minimum bandwidth of minimum service traffic are controlled by the connection admission control so that it remains lower than the transmission link capacity. If the cell identified by the cell label search function (100) is an NRB-connected cell, the scheduler (130) that is buffered in the intermediate queue (101) and then used for complementary scheduling of the minimum service is used. It is queued in a reachable connection queue (155). NRB connection (155)
Is scheduled by the scheduler (130) when it is active and the queue of the least service connection queue group (174) is empty (no service of least traffic). Is done. The scheduler (130) does not know what type of queue this is processing. Thus, the queue for the least service (150, or 155 in the case of NRB) is scheduled with the same priority by the scheduler (130).

In another embodiment, the complementary scheduling scheme for minimum service connections and the scheduling of NRB connections are performed by the two types of schedulers themselves, which are controlled by certain rules, such as alternating processing. An obvious characteristic of this embodiment is that it controls the sharing of available bandwidth according to the type of service. In an example where the two schedulers are operating alternately, each service, the minimum service and the NRB service, is
If necessary, get half of the available bandwidth. In this embodiment, if a service requires less bandwidth than is actually available, other services can take advantage of this bandwidth. In the above example, N
One of the available bandwidth for what the RB service needs
If not more than / 4, the minimum service can utilize 3/4 of the available bandwidth.

The scheduler 130 of the complementary scheduling scheme is optionally submitted to a flow control mechanism, the advantage of which is sharing of minimum services in the remaining bandwidth and improving the cell loss rate in NRB services. FIG.
As shown in (2), the cell loss rate is improved for the NRB service. As shown in FIG. 3, the NRB scheduler 130 is submitted to a traffic flow control command (195) when network traffic flow control is being performed. In the case of the flow control cited in the prior art, the global back pressure shuts down the scheduler (130), which results in the scheduling of only the least service connection by the shaper.
The least service connection gets only its minimum bandwidth. Similarly, if an optional "stop" backpressure is received, the corresponding connection queue (15 for minimal service)
0, 155) for NRB service connection is no longer NRB
The least service connection that is no longer selected and selectively shut down by the scheduler 130 gets only its minimum bandwidth. In summary, there is no loss in minimum service when using the prior art hopping traffic control flow mechanism with the scheduling approach of the present invention.

FIG. 4 shows a flow of the operation of the shaper related to the embodiment of the present invention when the RB connection is formed.
This figure shows the scheduling of the RB connection and the minimum bandwidth of the minimum service connection. Each cell time (20
5) If one queue is selected by the shaper (right branch of decision 208) and the selected queue belongs to the RB service (left branch of decision 200), the selected queue 140 is examined. If this is empty, wait for a new cell time. Queue 14
If 0 is not empty, the corresponding cell in queue 140 is dequeued (202) and queue 1 for transmission.
It is queued at 60. If the selected queue belongs to the minimum service (right branch of decision 200), a dedicated process for minimum scheduling of the minimum service is started. If queue 150 is empty, wait for a new cell time (right branch of decision 212). Queue 150
Is not empty (left branch of decision 212), a minimum service is performed. The queue-in-queue bit (QiQ) read in the control block of the connection is examined (214) to detect an error. If the QiQ bit is not set, the identifier of the selected queue is queued in the queue (174) of the queue group (216) and the QiQ bit is set (218). Qi
If the Q bit is already set in the control block of the least service connection, this means that the corresponding queue has been enqueued in the queue of the queue group (174). In this case, in the preferred embodiment of the present invention, the element cannot be enqueued twice,
Operation must be aborted. However, the probability of occurrence of this error event, which indicates that the minimum service is greatly delayed, is very low (similar to the cell loss rate of the RB service, ie about 10 ^-20 ). Queue group (1
The queue of 74) can be replaced by an array list, which allows further processing of queues already belonging to it. However, in such an embodiment, when the list is exhausted,
You will have to face the same problem.

FIG. 5 shows the flow of operation of complementary scheduling according to the preferred embodiment of the present invention. This figure shows the operation flow of scheduling the NRB service and the minimum service. If the queue 174 is not empty (left branch of the decision 220), the corresponding minimum service connection queue (150) is dequeued from the queue of the queue group (174) (222) and the state "minimum service". Is set (224). The result of the judgment 220 is Ye
If s, the queue of queue group (174) is empty and there are no minimum service connections waiting to be serviced according to the minimum guaranteed bandwidth. This means that the scheduler 130 (round robin in the preferred embodiment) selects one queue from the minimum service connection queue (150) or the NRB connection queue (155) which has the same priority as the scheduler (155). 22
1). If queues 150 and 155 are empty,
The scheduler has no work to do ("no work" branch) and the priority mechanism is reactivated (step 220)
Proceed to). Otherwise, at step 223, cell space 170 is examined, and if not full, the selected queue is dequeued and stored in cell space (170), if applicable. Sets the cell space to full. If "minimum service state" is set (decision 22
5) The minimum service indicator is reset (226). That is, the QiQ bit and the “minimum service state” of the selected minimum service queue are reset.

FIG. 8 shows the minimum service and NR as in FIG.
13 shows NRB cell processing for providing service B. Unlike FIG. 5, there are two scheduling approaches to ensure the sharing of available bandwidth between the minimum service connection and the NRB connection according to the preferred embodiment suggested in the description of FIG. The algorithm supporting this embodiment is different from that described in step 221 of FIG. Step 821 replaces step 221 of activating the NRB scheduler. In step 821, one of the two schedulers is selected according to a control rule, eg, a rule that each scheduler operates alternately. The minimum service to service the minimum service connection, depending on the selected scheduler
The scheduler is activated (step 822) or the NRB scheduler serving the NRB connection is activated (step 823). Once activated,
There is work to be performed on the selected scheduler (queue 1
50 or 155 is not empty), the next step 223 is performed, or there is no work to be performed on the selected scheduler (queue is empty) and step 220 is performed to activate the priority mechanism. I do.

FIG. 6 illustrates the use of ingress and egress traffic scheduling according to the present invention on an adapter card of an ATM network element 600. This network element includes end user 602 and end user 60
AT between 4, such as RB, NRB or minimum service
Switch M traffic. Traffic enters network element 600 and enters outbound adapter card 620 via ATM link 632. The transmit adapter card first processes the ATM physical layer of component 670, and then the data is queued to DATA STORE component 660. A device that implements the method of the present invention, the so-called "ATM process" (650), schedules cells for transmission to switch 610 according to the present invention via switch interface 640. The cell reaches the switch 610 via the internal link 630. When the network switching process occurs at the switch, the cells are sent from the switch via the internal link 630 to the receiving adapter (621) for transmission. The cell is the switch interface 640
And the data is transferred to the scheduling adapter (650) of the present invention via the data storage unit 660. After being scheduled according to the present invention, the cells pass through the ATM physical layer, which is processed in component 670. The cell then outputs output A from network element 600.
It is sent over TM link 632 to destination 604.

FIG. 7 illustrates one embodiment of a network element of the scheduling scheme of the present invention that supports ABR connections. To support ABR services,
The scheduling method is combined with a per-connection measurement mechanism to provide EFCI (Explicit Forward Congestion Infor
mation: explicit forward congestion information) Provide feedback information to the source using either cell marking or relative rate marking specified in the ATM Forum specification. This combination is implemented at the receiving and transmitting adapters of all network elements of the ABR connection path. In FIG. 7, a transmission source (710) includes a data cell and a control cell (Resource Management (RM)).
ent) cell) to the destination (720). Traffic traverses two network elements (600) in FIG. The transmission adapter (620) and the reception adapter (62)
When the scheduling method of the present invention is applied to the ATM processing layer (650) of 1), the destination (720) returns the RM cell to the source (710) and closes the ABR control loop (730). The congestion measurement of each adapter (741) using the queuing status of the ATM processing (650) is based on the RM cell marking process (740, 7) of the reverse control cell flow.
Trigger 42). Marking process 740 is R
Congestion indication bit (CI: congestion indic) of M control cell
ation) and a No Increase (NI) bit field, the process 742 of the network element located at the network element at the terminating end of the network closest to the source does not use the CI bit. This indicates in the RM control cell an explicit rate equal to the minimum cell rate, directly indicating to the source that its traffic should be reduced.

[0027] Optionally, a hopping backpressure mechanism (described in the prior art) can be added to each node of the ABR path. The backpressure command (750) is returned from one ATM processing technique (650) to the upstream one and is triggered by the same per-connection measurement mechanism that also triggers the ABR feedback process for the source. The result of the back pressure is the ATM processing unit (6
50) sets the transmission rate of the back-pressured connection to
At most, down to that MCR, this bandwidth represents the rate that the ATM processing unit 650 guarantees. This immediately avoids an increase in congestion. When combined with the terminating network RM cell marking process (742) that returns the backpressure command at the last hop back to the source, the backpressure mechanism provides lossless ABR service.

In summary, the following matters are disclosed regarding the configuration of the present invention.

(1) In a method for transmitting and receiving a packet of a connection in which a minimum bandwidth is reserved at the time of establishing a connection at a packet network node, receiving the packet in a dedicated connection queue and controlling an absolute time. Under
Scheduling a connection queue according to a rate corresponding to the reserved bandwidth; storing a queue identifier of the scheduled connection queue;
Reading the stored connection queue identifier and, if any identifier is found, if any packet is found in the connection queue corresponding to the read identifier, transmitting the first packet; and storing the identifier. If not, scheduling a connection queue, and if any packets are found in the scheduled queue, transmitting the first packet. (2) The above (1), which further supports a second type of connection in which a certain bandwidth is reserved when the connection is established.
The method according to claim 1, wherein after the first scheduling step, if the scheduled connection belongs to a second type of connection, the first packet if any packet is found in the scheduled queue. Transmitting the message. (3) The method according to the above (1) or (2), further comprising a third type of connection in which no bandwidth is reserved when the connection is established, wherein the receiving step and the second scheduling step are connected. The method is performed on a connection where no bandwidth is reserved when the connection is established. (4) sent by the source end system to the destination end system for use in a network node located on a path between the source end system and the destination end system; (1) to (1) further supporting a first type of connection between the end systems establishing a control loop (730) containing control packets returned by the system to the source end system. 3) The method according to any one of the preceding claims, wherein the queuing level (74
5) measuring, and writing a bit corresponding to the exceeded threshold value into a control packet on a control loop (730) returning to the source when the specified threshold value is passed. (5) a network adjacent to the source, further comprising a step of writing, in a dedicated field of the control packet, a minimum bandwidth reserved for the first type of connection at the time of connection when a specific threshold is exceeded. -The method according to (4) above, which is used in a node. (6) The method according to any one of the above (1) to (5), wherein the second scheduling step (130) is controlled by a network traffic flow control command. (7) The method according to any one of (1) to (6) above, wherein the packet network is an ATM network and the packet is an ATM cell. (8) An apparatus for transmitting and receiving traffic cells, for which a minimum bandwidth is reserved at the time of establishing a connection, to and from an asynchronous transfer mode (ATM) network node, reads an incoming cell, and stores the cell (100) in a scheduler (130).
Means for queuing into a specific connection queue (150) readable by the user and scheduling the connection queue (111) according to a rate corresponding to the reserved bandwidth under control of an absolute time. Means for storing the connection queue identifier in the connection queue identifier queue (174) each time the queue is scheduled; and reading the stored connection queue identifier (1).
78) If any identifier is found, and if any packet is found in the connection queue corresponding to the read identifier, the first packet is transmitted. If no identifier is stored, the connection queue is sent. Means for transmitting the first packet if any packet is found in the scheduled connection queue. (9) receiving a cell from an output link of a connection established between the source (602) and the destination (604) of the ATM network and having a minimum bandwidth reserved at the time of connection establishment, and converting the cell to an internal link; At the receiving network adapter (620) transmitting to (630), the source (60
2) means (670) for processing the physical layer of the data receiving cells via the input network line (632), means for queuing the cells processed by the physical layer components (660), and establishment of a connection. An apparatus (650) according to (8), wherein said cell queued by said queuing means is transmitted and received according to a minimum bandwidth reserved at times.
Means (64) for sending said cell to an output link (630).
0) with the receiving network adapter (62)
0). (10) Cells received from the internal link (630) on the output link of the connection established between the source (602) and the destination (604) of the ATM network and having a minimum bandwidth reserved at the time of connection establishment. In the transmission network adapter (620) for transmitting the
Means (640) for receiving a cell from 0) and transmitting and receiving said cell queued to a data storage component in a queuing means according to the minimum bandwidth reserved at the time of connection establishment (8). (650), means for storing cells (660), and processing of the physical layer of the cells before the cells are output to an output network line (632).
A transmission network adapter (620) comprising means (670) for sending to the destination (604) via the network. (11) A network element (600) that connects a line (632) and supports a connection between the source (602) and the destination (604) of the ATM network where a minimum bandwidth is reserved when the connection is established. ) Receiving at least one of the data transmitted by the connection source 602 via the input network line (632) and transmitting it to the output internal link (630).
One adapter, means (610) for switching data received from the receiving means (620) toward one of the internal links (630), and data received from the switching means via the internal input link (630) Connection destination (60
A network element (600) comprising at least one adapter according to (10) for transmitting to an output network line (632) towards 4).

[Brief description of the drawings]

FIG. 1 shows a solution for scheduling mixed RB and NRB traffic in a network.

FIG. 2 illustrates the scheduling of minimum service traffic at a network node according to the method of the invention;

FIG. 3 illustrates the scheduling of mixed traffic of RB, NRB and minimum service at a network node according to the method of the invention.

FIG. 4 is a flowchart illustrating a basic operation of scheduling a minimum service RB and a minimum bandwidth according to the present invention;

FIG. 5 is a flowchart illustrating a basic operation of servicing minimum bandwidth traffic of a minimum service connection, NRB traffic of a minimum service connection, and residual bandwidth traffic according to the present invention;

FIG. 6 is a diagram illustrating one embodiment of the present invention in a network node.

FIG. 7A with optional network flow control
FIG. 3 shows the support of the minimum service according to the invention, used for BR services.

FIG. 8 is a flow diagram that describes the basic operation of serving a minimum service and an NRB service using two scheduling techniques for complementary scheduling.

[Explanation of symbols]

 100 ATM Label Search 101 Intermediate Queue 110 Policing Function 111 Shaping Function 120 Cell Transmission Function 130 NRB Service 140 RB Connection Queue 145 RB Traffic 147 Priority Operation 155 NRB Connection Queue 160 RB Service Queue 170 NRB Cell Space 190 Absolute Timing cell spacing 195 NRB traffic

Continued on the front page (72) Inventor Daniel Orsatti France Le Dauphin Bray Are des Boulaux 11 (72) Inventor Fabrice Felplanken France 06200 Nice Lut de Saint Laurent 1706

Claims (11)

[Claims] 1. A method for transmitting and receiving packets of a connection at a packet network node, wherein a minimum bandwidth is reserved when the connection is established, comprising the steps of: receiving a packet in a dedicated connection queue; Scheduling a connection queue according to a rate corresponding to the reserved bandwidth; storing a queue identifier of the scheduled connection queue; reading the stored connection queue identifier; If the packet is found and any packet is found in the connection queue corresponding to the read identifier, transmitting the first packet, and if no identifier is stored, schedule the connection queue. And if any packet is found in the scheduled queue , Methods and sending the first packet. 2. The method according to claim 1, further comprising supporting a second type of connection in which a certain bandwidth is reserved when the connection is established.
After the step, if the scheduled connection belongs to the second type of connection, further comprising transmitting the first packet if any packet is found in the scheduled queue. Features method. 3. The method of claim 3, wherein a bandwidth is not reserved when the connection is established.
3. The method according to claim 1, further comprising the step of: receiving the second scheduling step on a connection for which no bandwidth is reserved when the connection is established. Method. 4. The destination end system sent by the source end system to the destination end system for use in a network node located on a path between the source end system and the destination end system. 4. The method of claim 1, further comprising a first type of connection between said end systems establishing a control loop containing control packets returned by said end system to said source end system. 3. The method according to any one of the preceding claims, wherein for a given connection of a first type, measuring a queuing level (745) at said node;
The method further comprising writing a bit corresponding to the exceeded threshold to a control packet on a control loop (730) returning to the source if the defined threshold is passed. 5. The method according to claim 5, further comprising the step of writing, in a dedicated field of the control packet, a minimum bandwidth reserved for the connection of the first type at the time of connection if a certain threshold is exceeded. 5. The method of claim 4, wherein the method is used at a serving network node. 6. A second scheduling step (13).
6. The method according to claim 1, wherein 0) is controlled by a network traffic flow control command. 7. The method according to claim 1, wherein the packet network is an ATM network and the packets are ATM cells. 8. An apparatus for transmitting and receiving traffic cells in an asynchronous transfer mode (ATM) network node for which a minimum bandwidth is reserved at the time of establishing a connection. 130) can read a particular connection queue (1
50) means for queuing the connection queue (111) under control of an absolute time according to a rate corresponding to the reserved bandwidth; The connection queue identifier is stored in the connection queue identifier queue (174).
Reading the stored connection queue identifier (178);
If any identifier is found and any packet is found in the connection queue corresponding to the read identifier, the first packet is transmitted.If no identifier is stored, the connection queue is scheduled. Means for transmitting the first packet if any packet is found in the scheduled connection queue. 9. Receiving a cell from an output link of a connection established between a source (602) and a destination (604) of the ATM network, the connection having a minimum bandwidth reserved at the time of connection establishment. At the receiving network adapter (620) transmitting to the internal link (630), from the source (602) to the input network line (632)
Means (67) for processing the physical layer of the data receiving cell via
0), means for queuing the cells processed by the physical layer component (660), and the cells queued by the queuing means according to the minimum bandwidth reserved at the time of establishing the connection. An apparatus (650) according to claim 8, for transmitting and receiving, and means (640) for sending said cells to an output link (630).
A receiving network adapter (620) comprising: 10. A source of an ATM network (602).
A transmission network adapter (620) transmitting cells received from an internal link (630) on the output link of the connection for which a minimum bandwidth has been reserved at the time of connection establishment, established between the connection and the destination (604). Means for receiving cells from the internal link (630) (64
Device (650) according to claim 8, wherein said cell queued to a data storage component is transmitted to and received from a data storage component according to the minimum bandwidth reserved at the time of connection establishment.
Means for storing cells (660), and means for processing the physical layer of the cells and sending these cells to the destination (604) via the output network line (632) (670). Transmit Network Adapter (620). 11. A network element for connecting a line (632) and supporting a connection between a source (602) and a destination (604) of an ATM network, wherein a minimum bandwidth is reserved when the connection is established. At (600), the connection source 6 is connected via the input network line (632).
02, and at least one adapter according to claim 9 for receiving data transmitted by said internal link (630) and transmitting it to an output internal link (630). Means (61)
0) and the data received from the switching means are transferred to the internal input link (63
A network element (600) comprising at least one adapter according to claim 10 for transmission to an output network line (632) towards a connection destination (604) via a connection destination (604).