JP2001519973A - Prioritized access to shared buffers - Google Patents

Abstract

(57) [Summary] In a link-level flow control system, there is provided a method and apparatus having a function of partitioning a buffer resource shared by a plurality of connections into a priority buffer subset using at least one threshold. This allows for different categories of service levels for delay limits. The disclosed link level flow control system achieves cell-free. The shared buffer resource is divided into N priority pools defined by N-1 threshold levels. Each priority pool can be assigned to each service category. Link level counters and registers located at the transmitting element, along with priority level indicators associated with each connection, are used to implement shared buffer resources.

Description

DETAILED DESCRIPTION OF THE INVENTION Prioritized access to shared buffers Field of the invention The present application relates to a communication method and apparatus in a distributed switching architecture, and more particularly to a method and apparatus for performing prioritized access to a shared buffer resource in a flow control virtual connection. Background of the Invention Flow control virtual connection (FCVC) protocols used in a distributed switching architecture are currently known in the art and are described below with reference to FIG. This protocol communicates the state (buffer allocation and current state) of each connection, such as virtual channel connection and virtual path connection, between upstream and downstream network elements in order to provide "cell-free" guarantee. I need. A cell is a unit of data to be transmitted. Each cell needs a buffer to be stored. Currently known flow control mechanisms can also be applied to connection-level flow control between a plurality of connections each having equal access to the shared buffer pool of the receiver. Regardless of the time sensitivity of one or more data cell sequences, no buffer allocation is distinguished within a service category. Summary of the Invention In a link-level flow control system, a function is provided for partitioning buffer resources shared by a plurality of connections into a plurality of priority buffer subsets defined by at least one threshold. Thus, different categories of service levels are possible in terms of delay limits. The link level flow control system according to the present invention reduces a cell flow from a transmitter element to a receiver element based on a link level counter and a register in a transmitter element updated by feedback information from the receiver element. , Realizes no cell loss. A link can be defined as a physical link or as a logical group of logical connections. Access to the partitioned buffer subsets is based on the priority assigned to each subset and the priority associated with each connection having cells to be buffered. The buffer subset is defined by at least one threshold. In one embodiment, the threshold is dynamically adjustable. Link level counters and registers in the transmitter element maintain threshold levels, monitor buffer pool utilization of the receiver element, and control cell traffic from the transmitter element to the receiver element. As in the case of link level flow control, feedback is provided from the receiver element reflecting the current state of the shared buffer resource. BRIEF DESCRIPTION OF THE FIGURES The above and further advantages will be more fully understood with reference to the following description and the accompanying drawings. FIG. 1 is a block diagram of a connection level flow control device known in the art. FIG. 2 is a block diagram of the link level flow control device according to the present invention. 3A and 3B are flowcharts illustrating counter initialization and preparation for cell transmission in the flow control method according to the present invention. FIG. 4 is a flowchart showing cell transmission in the flow control method according to the present invention. 5A and 5B are flowcharts illustrating preparation and transmission of an update cell in the flow control method according to the present invention. 6A and 6B are flowcharts illustrating another embodiment different from the preparation and transmission of the update cell in FIGS. 5A and 5B. 7A and 7B are flowcharts showing reception of an updated cell in the flow control method according to the present invention. 8A, 8B, and 8C are flowcharts illustrating the preparation, transmission, and reception of a test cell in the flow control method according to the present invention. 9A, 9B, and 9C are flow diagrams illustrating another embodiment of the preparation, transmission, and reception of the test cells of FIGS. 8A, 8B, and 8C. FIG. 10 shows a state in which the cell buffer pool according to the present invention is viewed from an upstream element. FIG. 11 is a block diagram of a link level flow control device of an upstream element that provides prioritized access to a shared buffer resource of a downstream element according to the present invention. FIGS. 12A and 12B are flowcharts showing initialization and preparation of a counter for cell transmission in the priority access method according to the present invention. 13A and 13B show another embodiment of the cell buffer pool according to the present invention as viewed from an upstream element. FIG. 14 is a block diagram of an upstream element flow control device that provides guaranteed minimum bandwidth and prioritized access to shared buffer resources for downstream elements according to the present invention. 15A and 15B are flowcharts illustrating counter initialization and preparation for cell transmission in the guaranteed minimum bandwidth mechanism using priority access according to the present invention. FIG. 16 is a block diagram illustrating a transmitter, a data link, and a receiver in which the disclosed joint flow control mechanism is implemented. FIG. 17 shows a data structure associated with the queue of the receiver of FIG. Detailed description FIG. 1 shows resources required for connection level flow control. As mentioned above, the configuration shown in FIG. 1 is currently known in the art. However, a brief description of the connection level flow control configuration will facilitate the description of the link level flow control method and apparatus disclosed herein. As shown, one link 10 provides an interface between an upstream transmitter element 12, also known as an UP subsystem, and a downstream receiver element 14, also known as a DP subsystem. Each element 12, 14 can operate as a switch between other network elements. For example, the upstream element 12 of FIG. 1 can receive data from a PC (not shown). This data is communicated via link 10 to downstream element 14. The downstream element 14 sends the data to a device such as a printer (not shown). Alternatively, the illustrated network elements 12, 14 themselves may be network terminal nodes. An important function of the arrangement disclosed herein is to transfer data cells from the upstream element 12 via the connection 20 in the link 10 to the downstream element 14 where the data cells are transferred to the cell buffer 28. Is to keep it temporarily. The cell format is known, and is referred to as “quantum flow control” (1st. 5. 1 edition, Issued June 27, 1995), and It is subsequently described in a new edition published by the Flow Control Consortium. In FIG. The block marked cell buffer 28 It represents a set of cell buffers allocated to each connection 20. The data cell is By sending to another link beyond the downstream element 14, Or With the use of cells in the downstream element 14, It is released from the buffer 28. If the downstream element 14 is a terminal node such as a workstation, Cell utilization in the downstream element 14 This may include building a data frame from individual data cells. Each of the upstream element 12 and the downstream element 14 The processor is controlled by UP (upstream processor) 16 and DP (downstream processor) 18. Processor 16, In each of the 18 A set of buffer counters for realizing connection level flow control is associated. These buffer counters Respectively, It is implemented as a rising counter / limit register set that is set to facilitate changes in resource usage. The counter of FIG. 1 described in detail below In the first embodiment, It is implemented in the UP internal RAM. Some of the counter names described and illustrated for the prior art include: The same counter names used for the flow control method and apparatus disclosed in this application are used. This means simply, It merely indicates that there is a similar function or element in the prior art to the newly disclosed elements such as counters and registers. In the first embodiment, inside the link 10 which is a copper conductor, A plurality of virtual connections 20 are provided. In another embodiment, The link 10 is a logical group of a plurality of virtual connections 20. The number of connections 20 realized inside the link 10 is Each network element 12, together with the bandwidth required per connection, Depends on 14 requirements. In FIG. For simplicity, Only one connection 20 and the corresponding counter are shown. First, With respect to the upstream element 12 of FIG. Two buffer state controllers, That is, A BS counter BS_Counter22 and a BS limit BS_Limit24 are provided. In the first embodiment, Each is Implemented as a 14-bit counter / register, 16, Connection to 383 buffers is possible. This number is For example, 139Mbps, 10, It supports 000km round trip service. Buffer status counter 22, 24 is It is used only when the flow control of the corresponding connection 20 is permitted. That is, One bit of each connection descriptor or queue descriptor of UP16 is This is set to indicate that the flow control of the connection 20 has been permitted. The BS counter BS_Counter22 is Each time a data cell is transferred from the upstream element 12 through the corresponding connection 20, Incremented by UP16. As described below, This counter 22 It is adjusted periodically during the updating process based on the information received from the downstream element 14. Therefore, The BS counter BS_Counter22 is Whether the connection 20 between the upstream element 12 and the downstream element 14 is currently being transmitted, Or It will indicate the number of data cells that have not yet been released from the buffer 28 of the downstream element 14. BS limit BS_Limit24 is At connection establishment time, The setting is made to reflect the number of buffers 28 available in the receiver 14 for this connection 20. For example, A BS counter BS_Counter 22 for this connection 20 indicates that 20 data cells have been transmitted, If the BS limit BS_Limit 24 indicates that this connection 20 is limited to 20 receiver buffers 28, UP16 is Until further buffer space 28 is notified by downstream element 14 that connection 20 is available for connection 20 Further transmission from the upstream element 12 will be prohibited. The transmission counter Tx_Counter 26 is It is used to count the total number of data cells transmitted by UP 16 through this connection 20. In the first embodiment, This counter is It is a 28-bit counter that circulates at 0xFFFFFFFF. As described below, The transmission counter Tx_Counter 26 is At the time of the inspection event, It is used to consider the error cell for this connection 20. In the downstream element 14, DP18 also A counter set is managed for each connection 20. The buffer limit register Buffer_Limit30 is In the downstream element 14, Perform administrative functions to protect against unauthorized transmitters. In particular, The buffer limit register Buffer_Limit30 is The maximum number of cell buffers 28 in the receiver 14 that can be used by this connection 20 is shown. In most cases, BS limit BS_Limit24 is It is equal to the buffer limit register Buffer_Limit30. However, It may be necessary to adjust the maximum number of cell buffers 28 for this connection 20 to the increasing side or the decreasing side. This feature Coordinated by network management software. In order to prevent "dropout" of data during transmission, The increase in buffers per connection is Before the BS limit BS_Limit 24, the data is first reflected in the buffer limit register Buffer_Limit 30. vice versa, The reduction in the number of receiver buffers per connection is First, it is reflected in BS limit BS_Limit24, afterwards, This is reflected in the buffer limit register Buffer_Limit30. The buffer counter Buffer_Counter32 is In the downstream element 14, Shows the number of buffers 28 currently in use for storing data cells. As explained below, This value is For the upstream element 12, Used to more accurately signal buffer availability at the downstream element 14. In the first embodiment, Both the buffer limit register Buffer_Limit 30 and the buffer counter Buffer_Counter 32 are 14 bits wide. The N2 limit N2_Limit34 is The frequency of connection flow rate communication to the upstream transmitter 12 is determined. A cell containing such flow rate information is Each time the receiver element 14 sends out a number of cells equal to the N2 limit N2_Limit 34 from the receiver element 14, it is sent upstream. About this update operation, Further explanation will be given. In the first embodiment, The N2 limit N2_Limit 34 is 6 bits wide. DP18 uses N2 counter N2_Counter36, Track the number of cells transmitted from the receiver element 14 since the N2 limit N2_Limit 30 was last reached. In the first embodiment, The N2 counter N2_Counter 36 is 6 bits wide. In the first embodiment, DP18 is Holding the transmission output counter Fwd_Counter38, Holds the current count of the total number of cells transmitted through the receiver element 14. this is, It includes a buffer that is released when a data cell is used for data frame construction at a terminal node. When the maximum count value of the counter 38 is reached, The counter value returns to zero and continues counting. The total number of cells received by the receiver element 14 is It can be obtained by adding the buffer counter Buffer_Counter 32 to the transmission output counter Fwd_Counter 38. As described below, The transmission output counter Fwd_Counter 38 is During the inspection event, Used for correcting the error cell of the transmitter element 12. In the first embodiment, The transmission output counter Fwd_Counter 38 is 28 bits wide. In the second embodiment, The DP 18 holds a reception counter Rx_Counter40. The reception counter Rx_Counter40 is The counter is incremented each time a data cell is received through the connection 20 corresponding to the downstream element 14. The value of this counter 40 is Depending on the test cell, as well as, In generating an update cell, Can be used directly. Both of these will be further described below. In the second embodiment, The reception counter Rx_Counter40 is It has a 28-bit width similarly to the transmission output counter Fwd_Counter38. Connection-level flow control protocols include: There are two events in addition to the steady state condition. That is, Renewal and inspection. In steady state, Data cells are transmitted from the transmitter element 12 to the receiver element 14. In update mode, To correct the counter value of the transmitter element 12, Buffer occupancy information is returned by the receiver element 14 to the upstream side. Inspection mode is It is used to check for cell loss or cell contamination due to transmission errors between the upstream transmitter element 12 and the downstream receiver element 14. In the attached drawings, By attaching "[i]" to the connection level counter, It shows the correspondence with one connection [i] of a plurality of connections that may exist. As shown in FIG. Before every action, The counters of the upstream element 12 and the downstream element 14 are initialized. For initialization, Zeroing the counter value, as well as, This includes assigning initial values to limit registers such as Link_BS_Limit and Link_Buffer_Limit. In FIG. 3A, Buffer_Limit [i] is It has been initialized to (RTT * BW) + N2. This value is It is the product of the round trip time and the virtual connection bandwidth plus a value to accommodate the delay in processing the updated cell. About Link_N2_Limit, "X" represents the buffer status update frequency for the link, Also, For N2_Limit [i], “Y” indicates the buffer status update frequency for each connection. In steady state operation, The UP 16 of the transmitter element 12 is Which virtual connection (VC) 20 Has a non-zero cell count value (ie, Have cells ready for transmission), Has a BS_Counter value smaller than BS_Limit, as well as, It is determined whether the VC has an indicator that it is to be sent next (FIGS. 3A and 3B). UP16 is If flow control is allowed, When the data cell is transmitted on each connection 20, the BS counter BS_Counter22 and the transmission counter Tx_Counter26 are always incremented (FIG. 4). DP 18 When receiving a data cell, It is checked whether the buffer counter Buffer_Counter32 is equal to or larger than the buffer limit register Buffer_Limit30. this is, This is identification information as to whether or not there is a buffer that can be used for receiving data cells. If Buffer_Counter ≧ Buffer_Limit holds, The data cell is discarded (FIG. 3B). Otherwise, As shown in FIG. The DP 18 increments a buffer counter Buffer_Counter 32 and a reception counter 40, The data cells are placed in buffer cells 28. When the transmission counter Tx_Counter 26 and the reception counter Rx_Counter 40 reach their maximum values, they return to zero. If flow control is not allowed, The disclosed function is not implemented. Connections that do not use flow control on the link Cannot coexist with connections using link flow control. Cells are sent from connections that are not flow controlled, If received, No flow control accounting is used. This flow control accounting is Includes both connection level accounting and link level accounting. This allows A flow control connection and a non-flow control connection can function simultaneously. When a data cell is transmitted from the receiver element 14, The buffer counter Buffer_Counter 32 is decremented. If the connection level flow control protocol is allowed, BS limit BS_Limit24 has been reduced, Unless the receiver element 14 has a sufficient number of cells to send to make the buffer counter Buffer_Counter 32 less than the buffer limit register Buffer_Limit30, The buffer counter Buffer_Counter32 must not exceed the buffer limit register Buffer_Limit30. Buffer status update This is performed when the receiver element 14 has transmitted a number of data cells equal to the N2 limit N2_Limit34. In the first embodiment in which the DP 18 holds the transmission output counter Fwd_Counter 38, As shown in FIG. 6A, If an update occurs, The transmission counter Fwd_Counter 38 is returned from the receiver element 14 to the transmitter element 12 by an update cell. In the embodiment using the reception counter 40 for the downstream element 14, As shown in FIG. 5A, A value obtained by subtracting the value of the buffer counter Buffer_Counter 32 from the value of the reception counter Rx_Counter 40 is transmitted to the update cell. As shown in FIG. 7A for two embodiments, The updated cell is transmitted at the transmitter 12 It is used to update the value of the BS counter BS_Counter22. Since the BS counter BS_Counter22 is independent of the buffer allocation information, Without affecting the performance of the above aspects of connection level flow control, Buffer allocation can be changed. Update cells require allocated bandwidth to guarantee a limited delay. This delay is As a component of the round trip time, It must be considered to determine the buffer allocation for each connection. The amount of bandwidth allocated to the update cell is It is a function of the maximum update counter (not shown) of the corresponding downstream transmitter element (not shown). This counter forces the scheduling of update cells and test cells. For this, This will be described below. Corresponding to the maximum update counter, Downstream transmitter elements include: There is a minimum update interval counter (not shown) that controls the space between update cells. Normal cells are Has seven records per cell, The minimum update interval counter is also set to 7. UP16 is Since we can only process one update record per cell time, For a full update cell received at UP16, Some records will be dropped. The update event is performed as follows (FIG. 1, FIG. 5A, And FIG. 5B). When the downstream element 14 sends (releases) the cell, The buffer counter Buffer_Counter32 is decremented, The N2 counter N2_Counter36 and the transmission output counter Fwd_Counter38 are incremented. If the N2 counter N2_Counter36 is equal to the N2 limit N2_Limit34, DP18 is Prepare an update cell to be returned to the upstream element 12, The N2 counter N2_Counter 36 is set to zero. The upstream element 12 receives from the downstream element 14 that sent the cell, Receiving the connection identification information, Identify which connection 20 should be updated. In the first embodiment, The DP 18 causes the value of the transmission output counter Fwd_Counter 38 to be inserted into the update record payload (FIG. 6A). In the second embodiment, DP18 is A value obtained by subtracting the value of the buffer counter Buffer_Counter32 from the value of the reception counter Rx_Counter40 is inserted into the update record payload (FIG. 5A). When the update cell is full of records, Or When the minimum bandwidth pacing interval is reached, The update cell is transmitted to the upstream element 12. When received on the upstream side, UP 16 receives the connection identification information from the update record, identifies the transmitter connection, From the update record, send / output counter Fwd_Counter 38 value, Or The value obtained by subtracting the value of the buffer counter Buffer_Counter32 from the value of the reception counter Rx_Counter40 is extracted. The BS counter BS_Counter22 is It is reset to a value obtained by subtracting the updated record value from the transmission counter 26 (FIG. 7A). If this connection has been prohibited from transmission because the BS counter BS_Counter 22 is greater than or equal to the BS limit BS_Limit 24, This state is reversed. in this case, The connection is allowed to send again. Summary, Update events are: For the sending element 12, Providing identification information on how many of the cells originally transmitted from the transmitting element 12 have been released from the buffer in the receiving element 14; This allows For the sending element 12, It provides more accurate identification information indicating the availability of the receiver element 14 buffer 28 for that connection 20. The check buffer status event serves two purposes. That is, 1) Provide a mechanism to calculate and compensate for cell loss or cell contamination due to transmission errors, 2) If the update cell is lost, Or Provides a mechanism to start (or restart) the flow if enough data cells are lost and the N2 limit N2_Limit34 is not reached. One timer (not shown) of the UP subsystem 16 is responsible for all connections. The connection is For each connection, Whether test cells should be sent from the upstream transmitter element 12 to the downstream receiver element 14 is allowed or prohibited. The inspection process at the transmitter element 12 includes: Search all connection descriptors, Inspection involves finding what was authorized (see FIGS. 8A and 9A). After the minimum pacing interval (inspection interval) has elapsed, The test cell is sent to the receiver element 14, The next test-permitted connection is identified. The test cell interval for the same connection is The number of active flow control connections, It is a function of the product of the specified cell spacing for all connections. The test cell has a higher priority than the update cell. The inspection event is performed as follows (see FIGS. 8A to 8C and FIGS. 9A to 9C). Each transmission element 12 connection 20 Inspection is performed after the timed inspection interval is reached. Connection flow control is allowed, And, If the connection is valid, The test event is scheduled for transmission to the receiver element 14. The buffer status check cell is In the test cell payload, Generated using the value of the transmission counter 26 for the connection 20, It is transmitted using the connection identification information from the corresponding connection descriptor (FIGS. 8A and 9A). In the first embodiment, The calculation of the error cell is At the receiver element 14, The sum of the transmission output counter Fwd_Counter 38 and the buffer counter Buffer_Counter 32 is calculated, The contents of the test cell record whose value was sent, That is, This is performed by subtracting from the value of the transmission counter Tx_Counter 26 (FIG. 9B). The value of the transmission output counter Fwd_Counter 38 is It is incremented by the error cell count value. And An update record having a new value of the transmission output counter Fwd_Counter 38 is generated. The updated send / output counter Fwd_Counter 38 value is: continue, Update the BS counter BS_Counter22 value of the transmitter element 12. In the second embodiment shown in FIG. 8B, The same function is This is realized by resetting the value of the reception counter Rx_Counter r40 to a value equal to the inspection cell payload value (transmission counter Tx_Counter26). The following update record It is established using the difference between the values of the reception counter Rx_Counter 40 and the buffer counter Buffer_Counter r32. in this way, Inspection events are Sent from the transmitter element 12 through the connection 20, It allows accounting for cells that are dropped or not received by the receiver element 14. The transmitter element 12 Have an up-to-date account of the number of buffers in the receiver element 14 available for receiving data cells; Also, Having the identification of when data cell transmission should be stopped due to the lack of available buffer 28 on the downstream side, “Cell-free” compensation is Enabled using buffer state accounting at the connection level. To enhance the receiver element buffer sharing mechanism in the above protocol, Link level flow control, also known as link level buffer state accounting, is added to connection level flow control. Such link level flow control, It can also be realized without connection level flow control. However, Without connection level flow control, Since there is no limit on the number of buffers that a single connection can consume, It is preferable to combine these two. For the following reasons: Buffer state accounting, It is desirable to execute at the link level in addition to the connection level. Link-level flow control While maintaining the “cell-free” compensation provided by connection-level flow control, Enables cell buffer sharing in receiver elements. By sharing the buffer, A limited number of buffers can be used with maximum efficiency. Since not all connections always require all buffers, Rather than providing a number of buffers equal to RTT times the bandwidth for each connection, Fewer buffers are made available at the receiver element 14. A further advantage of link-level buffer state accounting is that Without the need for increased inversion bandwidth for each connection A more accurate indication of the availability of the downstream buffer is given to each connection. About link level flow control Less than, This will be described with reference to FIG. In addition, In FIG. Elements similar to FIG. 1 include: The same reference numbers are indicated with dashes. Also in FIG. Only one virtual connection 20 'is shown on link 10'. However, Normally, Link 10 'accepts multiple virtual connections 20'. Also, In the first embodiment, Link 10 'is a physical link, In the second embodiment, A logical group of a plurality of virtual connections. The upstream transmitter element 12 '(FSPP subsystem) In part, Switch originating port processor (From Switch Port Processor; FSPP) 16 '. The FSPP processor 16 ' Two buffer status counters, That is, BS counter BS_Counter22 ', as well as, A BS limit BS_Limit 24 'is provided. These counters Respectively, It has the same function as the function for each connection described with reference to FIG. The embodiment of FIG. Furthermore, Added to the upstream element 12 'and the downstream element 14', Contains a set of resources that enable link-level buffer accounting. These resources are It has the same functions as those used for each connection, Works at link level. For example, The link BS counter Link_BS_Counter50 is Including cells being transmitted between receiver 12 'and transmitter 14' and cells stored in receiver 14 'buffer 28', Track cells traveling between the FSPP 16 'and elements downstream of the receiver element 14'. As with the update event described for connection-level buffer accounting, The link BS counter Link_BS_Counter 50 indicates during the link update event Link transmission output counter Link_Fwd_Counter 68 value, Or The difference between the link reception counter Link_Rx_Counter 70 and the link buffer counter Link_Buffer_Counter 62 is It is changed by subtracting from the link transmission counter Link_Tx_Counter 54 value. In the first embodiment, The link level counter is It is mounted on an external RAM associated with the FSPP 16 '. The link BS limit Link_BS_Limit 52 is Should be shared between all connections 20 for which flow control is allowed, Limit the number of shared downstream cell buffers 28 'of the receiver element 14'. In the first embodiment, The link BS counter Link_BS_Counter 50 and the link BS limit Link_BS_Limit 52 are both 20 bits wide. The link transmission counter Link_Tx_Counter 54 is Keep track of all cells transmitted on link 10 '. This counter is During the link level update event, It is used to calculate a new value of the link BS counter Link_BS_Counter50. In the first embodiment, The link transmission counter Link_Tx_Counter 54 is 28 bits wide. In the downstream element 14 ', To Switch Port Processor; TSPP) 18 'also As with the counters commonly shown in FIGS. 1 and 2, It manages a set of counters for each link 10 '. TSPP18 ' Furthermore, A link buffer limit register Buffer_Limit60 is included. The link buffer limit register Buffer_Limit60 is In the downstream element 14 ', By indicating the maximum number of cell buffers 28 'of the receiver 14' that all connections 10 'can use, The same function as the link BS limit Link_BS_Limit 52 of the upstream element 12 'is executed. In most cases, The link BS limit Link_BS_Limit 52 is It is equal to the link buffer limit register Buffer_Limit60. The effect of adjusting the number of available buffers 28 'up or down based on the link width is: This is the same as the effect described above with respect to adjusting the number of buffers 28 that can be used by a specific connection 20. In the first embodiment, The link buffer limit register Link_Buffer_Limit 60 is 20 bits wide. The link buffer counter Link_Buffer_Counter62 is Shows the number of buffers in the downstream element 14 'currently used for storing data cells by all connections. This value is In the inspection event, It is used to correct the link send / output counter Link_Fwd_Counter 68 (described below). In the first embodiment, The link buffer counter Link_Buffer_Counter 62 is 20 bits wide. In the first embodiment, the link N2 limit Link_N2_Limit 64 and the link N2 counter Link_N2_Counter 66 each having an 8-bit width are: Used to generate link update records. This link update record Mixed with connection level update record. The link N2 limit Link_N2_Limit64 is Used to establish the threshold number that triggers the generation of link level update records (FIGS. 5B and 6B); The link N2 counter Link_N2_Counter 66 and the link transmission output counter Link_F wd_Counter 68 are: Incremented each time a cell is released from a buffer cell in receiver element 14 '. In the first embodiment, The N2 limit N2_Limit34 'and the link N2 limit Link_N2_Limit64 are both A static variable that is initialized only once. However, In a further embodiment of the present invention, Each variable is It can be dynamically adjusted based on the measured bandwidth. For example, If the forward link bandwidth is relatively high, The link N2 limit Link_N2_Limit64 is The link level update record transmission is adjusted in a decreasing manner so that transmission occurs more frequently. The impact on forward bandwidth is considered to be minimal. If the forward bandwidth is relatively low, Since the unknown availability of the buffer 28 'of the downstream element 14' becomes less significant, The link N2 limit Link_N2_Limit64 can be raised. The link transmission output counter Link_Fwd_Counter 68 is It keeps track of all cells coming from the corresponding link 10 'and released from the buffer cells 28' of the receiver element 14 '. In the first embodiment, This counter is 28 bits wide, Used to recalculate the link BS counter Link_BS_Counter50 in the update event. In another embodiment, Instead of the link transmission output counter Link_Fwd_Counter 68, a link reception counter Link_Rx_Counter 70 is used. In this example, The link transmission output counter Link_Fwd_Counter 68 is also 28 bits wide, Track the number of cells received across all connections 20 'of link 10'. Referring to FIG. 2 and subsequent figures, The receiver element buffer sharing method will be described. As shown in FIG. 3B, Normal data transfer to the downstream element 14 'by the FSPP in the upstream element 12' As long as the link BS counter Link_BS_Counter50 is less than or equal to the link BS limit Link_BS_Limit52, This can be done via all connections 20 'of link 10'. By this determination, FS PP16 ' Transmission of a larger number of data cells than data cells determined to be usable by the downstream element 14 'is prevented. The accuracy of this decision is It is maintained by the update and inspection events described below. If neither buffer limit register Buffer_Limit 30 at the connection level nor the link level is exceeded, A data cell is received at the downstream element 14 '(FIG. 3B). If the limit is exceeded, The cell is discarded. As shown in FIGS. 5B and 6B, In the update event at the link level, When the value of the link N2 counter Link_N2_Counter 66 reaches (is equal to or exceeds) the value of the link N2 limit Link_N2_Limit64, A link update record is generated. In the first embodiment, The link N2 limit Link_N2_Limit64 is set to 40. In the embodiment of FIG. 6B, the link update record obtained from the link transmission output counter Link_Fwd_Counter 68 is: In the updated cell transferred to FSPP1 6 ', It is mixed with the update record for each connection (the value of the transmission output counter Fwd_Counter 38 '). In the embodiment of FIG. 5B, The value obtained by subtracting the value of the link buffer counter Link_Buffer_Counter 62 from the value of the link reception counter Link_Rx_Counter 70 is Mixed with update records per connection. When the upstream element 12 'receives an update cell with a link update record, The upstream element 12 ′ stores a link BS counter Link_BS_Counter 50, It is set to a value obtained by subtracting the value of the update record from the value of the link transmission counter Link_Tx_Counter 54 (FIG. 7B). And The link BS counter Link_BS_Counter 50 of the upstream element 12 ′ is: It is reset to reflect the number of data cells transmitted by the upstream element 12 'but not yet released in the downstream element 14. In the first embodiment, Transferring update records Provide a corresponding FSPP processor (not shown) for each TSPP subsystem 14 ' This is achieved by providing a corresponding TSPP processor (not shown) for each FSPP subsystem 12 '. Therefore, When an update record to be returned by the subsystem 14 'to the upstream FSPP subsystem 12' is prepared, TSPP 18 'communicates the updated record to the corresponding FSPP (not shown). The cell is From the corresponding FSPP, It is transmitted to a TSPP (not shown) corresponding to the upstream FSPP subsystem 12 '. The corresponding TSPP is: Extract the update record from the received update cell, The record is transmitted to the upstream FSPP subsystem 12 '. For link-level inspection events, A test cell having a link transmission counter Link_Tx_Counter 54 value is: By FSPP16, It is transmitted for every W test cells. In the first embodiment, W is equal to four. In the embodiment of FIG. 9C, At the receiver element 14 ', the TSPP 18' Link transmission output counter Link_Fwd_Counter68 The inspection record value, that is, the link transmission counter Link_Tx_Counter 54 is increased by a value obtained by subtracting the sum of the link buffer counter Link_Buffer_Counter 62 and the link transmission counter Link_Fwd_Counter 68 from the link transmission counter Link_Tx_Counter 54, The inspection function at the connection level described above is executed. In the embodiment of FIG. 8C, The link reception counter Link_Rx_Counter 70 is The value is changed to a value equal to the content of the inspection record (link transmission counter Link_Tx_Counter 54). this is, The error cell is considered based on the link width. And An update record having a value obtained from the updated link transmission output counter Link_Fwd_Counter 68 value or the link reception counter Link_Rx_Counter 70 value is generated (FIGS. 8C and 9C). In the event of a major transmission link failure, In order to quickly readjust the value of the link transmission output counter Link_Fwd_Counter 68 (FIG. 9C) or the value of the link reception counter Link_Rx_Counter 70 (FIG. 8C), It is necessary to perform inspection events at the link level in addition to the connection level. Referring again to FIG. An example of the initial value of the described counter in the embodiment having 100 connections on one link is shown below. BS_Limit (24 ') = 20 Buffer_Limit (30') = 20 N2_Limit (34 ') = 3 Link_BS_Limit (52) = 1000 Link_Buffer_Limit (60) = 1000 Link_N2_Counter (66) = 40 Both of connection and link The BS_Limit value is equal to the Buffer_Limit value. BS_Limit 24 'and Buffer_Limit 30' are both equal to 20, As reflected in Link_BS_Limit52 and Link_Buffer_Limit60, There are only 1000 buffers 28 'in the downstream element. this is, This is because buffer pool sharing is enabled by link level feedback. if needed, Link_BS_Counter; Link_N2_Counter; And incrementing Link_Buffer_Counter, By prohibiting link level test cell transfer, Link level flow control can be prohibited. in this case, No updates are made. The disclosed invention, Furthermore, This can be enhanced by the dynamic buffer allocation scheme as described above for N2 limit N2_Limit 34 and link N2 limit Link_N2_Limit64. This scheme is In addition to the N2 limit N2_Limit34 and the link N2 limit Link_N2_Limit64, BS limit BS_Limit24, Link BS limit Link_BS_Limit52, Buffer limit Buffer_Limit30, And a function of dynamically adjusting limit parameters such as a link buffer limit Link_Buffer_Limit60. Such adjustments In one embodiment, Done according to the measured characteristics of the individual connection or of the entire link, In another embodiment, Established according to the determined priority scheme. Therefore, According to dynamic buffer allocation, Given limited buffer resources, It is possible to prioritize one or more connections or links. The link N2 limit Link_N2_Limit is set according to the desired accuracy of the buffer accounting. If based on link width, Since accurate buffer accounting allows for greater buffer sharing between multiple connections, As the number of connections in a link increases, It is desirable to reduce the link N2 limit Link_N2_Limit. vice versa, If the number of connections in the link decreases, Since the importance of sharing limited resources between a relatively small number of connections is reduced, The link N2 limit Link_N2_Limit may be increased. To change the maximum bandwidth maintained for a connection, In addition to adjusting per-link limits, It is desirable to adjust the limit for each connection. The dynamic allocation scheme disclosed in this application is: Based on the performance goals described above, Executed during link operation. In the first embodiment of the present invention, All the logic for incrementing the counter is located in the FSPP processor 16 '. In this regard, Reset to zero, The counter described as counting up to the limit value, In a further embodiment, Starting from the limit value, It may be implemented as counting down to zero. The transmitter processor and the receiver processor Limit value, Interpreted as the starting point of each counter, Decrements when an appropriate event is detected. For example, If the buffer counter (or link buffer counter) is implemented as a decrement counter, Each time a data cell is allocated to a buffer in the receiver, The counter is decremented. When the data cell is released from the buffer, The counter is decremented. In this way, When the counter reaches zero, Serves to indicate that all available buffers have been allocated. With dynamic bandwidth allocation, Since dynamic adjustment of limit values must be considered at non-zero count values, It is relatively difficult to use such an implementation. Cell-free loss as described above, To further improve the link level flow control method, This includes providing a plurality of shared cell buffers 28 '' in the downstream element 14 ''. in this case, The cell buffer 28 '' N-1 threshold levels (threshold (1) 102, Threshold (2) 104, Threshold (3) 106) N priority cell buffer subsets (priority 0 108a, Priority 1 108b, Priority 2 108c, And priority 3 108d). In FIG. Such a cell buffer pool 28 '' is shown. In FIG. The four priorities indicated by priority 0 to priority 4 are: The state defined by three thresholds indicated by threshold (1) to threshold (3) is shown. With this priority buffer pool, "Hunger" low priority connections, Or While preventing cells from being transmitted downstream during link conflicts, It becomes possible to make a connection of a high priority transmit. The cell priority is identified for each connection. The policy to establish a threshold is In the first embodiment, it is determined according to a cell traffic prediction model, In another embodiment, it is adjusted dynamically. Such dynamic adjustments May be performed in response to cell traffic observed in the upstream transmitting element, Or, This may be done according to empirical cell traffic data observed in the prioritized buffer pool of the downstream element. For example, In one embodiment using dynamic threshold adjustment, If a fair amount of priority 0 traffic is detected, Reducing the number of buffers available for data cells having a priority less than priority 0, Or vice versa, It would be desirable to increase the number of buffers above threshold (3). The cell buffer pool 28 '' shown in FIG. This is obtained in view of the above-described modified version 12 ″ of the link-level flow control upstream element 12 ′. A pool 28 '' resides in the corresponding downstream element 14 ''. The modified upstream element 12 '' shown in FIG. As mentioned above, At least one link BS threshold (n) 100 established in association with a link BS counter Link_BS_Counter 50 ″ and a link BS limit Link_BS_Limit 52 ″; 102, 104, The cell buffer boule 28 '' of the downstream element 14 '' is characterized. These link BS thresholds 102, 104, 106 is It defines the number of cell buffers in the pool 28 '' that can be allocated to cells of a given priority. here, The priority is For each connection 20 '', It is identified by a register 108 associated with a BS counter BS_Counter 22 ″ and a BS limit BS_Limit 24 ″ register. The priority 108 a shown in FIG. 108b, 108, 108d is The highest priority 0 to priority 3 are identified. If there is no congestion, This is reflected by the fact that the link BS counter Link_BS_Counter 50 ″ becomes less than the link BS threshold (1) in FIGS. 10 and 11, and An arbitrary priority flow control connection can be transmitted. If congestion occurs, This means Indicated by an increase in the value of the link BS counter Link_BSCounter50 '', Access to downstream buffers of lower priority connections is denied, In effect, transmission of those cells is prohibited. In case of severe congestion, Only the highest priority cells are allowed to transmit. For example, Referring again to FIG. If the link level BS threshold (3) 106 has reached the downstream side, Only cells with priority 0 108a are allowed to transmit from upstream element 12 '' to downstream element 14 ''. Therefore, The higher the priority connection, Since the shared downstream buffer pool can be accessed preferentially, Hardly affected by the condition of the net. However, Connection-level flow control If the path for which the connection was intended is severely congested, Note that transmission of high priority connections may be prohibited. As mentioned above, The link BS counter Link_BS_Counter50 '' is It is updated periodically based on the value contained in the link level update record transmitted from the downstream element 14 ″ to the upstream element 12 ″. This periodic update Required to guarantee the correct functioning of the prioritized buffer access of the present invention. Threshold level 102, 104, In one embodiment of the present invention where 106 is dynamically changed, FSPP16 '' As a result of tracking the priority associated with the cell received at the upstream transmitter element, Or Based on the observed usage of the buffer of the downstream receiver element, Must have an accurate record of the state of cell buffer 28 ''. With multiple priority levels, Within a single quality of service, Various categories of services can be provided in terms of delay limits. Within each quality of service, The highest priority for shared buffers is Typically provided for connection / network management traffic. The connection / network management traffic is identified by the cell header. The second highest priority is Low bandwidth, Given to small burst connections, The third highest priority is Given to bursty traffic. According to the prioritization assigned as described above, Due to congestion in any of the service categories, It does not prevent the connection / management traffic from having the lowest cell delay. FIG. 12A shows the initialization process of the upstream element 12 ″ shown in FIG. The important thing is The same counters and registers as the upstream element 12 '' shown in FIG. Link BS Threshold 102, 104, Except that the 106 values are initialized to their respective buffer values T That is, it is set so as not to permit the prioritized access to the shared buffer resource. As mentioned above, These threshold buffer values can be preset and static, Or, It can also be adjusted dynamically based on empirical buffer utilization. FIG. Performed before sending the cell from the upstream element 12 '' to the downstream element 14 '', FIG. 4 shows a number of determination processes that are the same as the processes shown in FIG. 3B. However, 3B is different from the case of FIG. 3B in that a determination process for providing a priority access to the shared buffer resource is added. FS PP16 '' Using the priority value 108 associated with the cell to be transferred, A threshold value 102 at which transfer to the downstream element 14 '' of the cell is prohibited, 104, Determine the value of 106. And The link BS counter Link_BS_Counter50 ″ value is an appropriate threshold value 102, 104, It is determined whether it is equal to or greater than 106. If a positive determination is made, No data cells are transmitted. If a negative determination is made, As mentioned above, The cell is transmitted, Connection level collision determination is performed. In another embodiment, More or less than four priorities This is achieved with an appropriate number of thresholds. However, The minimum number of priorities is 2, The corresponding minimum number of thresholds is one. In general, For N priorities, There are N-1 thresholds. In yet another embodiment, Flow control is only performed at link level, Not performed at the connection level. However, For each connection, It is necessary to provide some form of priority indicator similar to the priority field 108 shown in FIG. In yet another embodiment, The link level flow control protocol described above is To ensure a guaranteed minimum cell rate per connection with no cell loss, Can be further enhanced. This minimum cell rate is Also referred to as guaranteed bandwidth. Flow control of connections below this minimum allocation rate This can be done only by the receiver element associated with the connection. Therefore, The minimum rate of one connection is not affected by congestion in other connections. The cell existing in the upstream element corresponding to FSPP116 is Should be transmitted using the allocated bandwidth from the upstream element, Or What is identified by whether it should be transmitted using dynamic bandwidth is This is a requirement of the disclosed mechanism. For example, To indicate which cells require allocated bandwidth, Cells may be provided in queues corresponding to a list named "priority". Similarly, To indicate a cell that requires dynamic bandwidth emission, The cells may be provided in queues corresponding to a list named "dynamic". In the frame relay configuration, The mechanism is used to monitor and limit both dynamic and allocated bandwidth. In settings involving only Internet traffic, Only the dynamic part of the mechanism is significant. In settings with only CBR flows, Only the allocated part of the mechanism is used. Therefore, According to the disclosed method and apparatus, From connections that require all allocated bandwidth to connections that require all dynamic bandwidth, And the use of mixed scheduling connections, including those in between. In this mechanism, A downstream cell buffer pool 128 similar to pool 28 'of FIG. It is logically divided into an allocation part 300 and a dynamic part 301. This allows Cells identified as receiving allocated bandwidth should be Buffered in the allocation part 300, Cells identified as receiving dynamic bandwidth are buffered in dynamic portion 301. In FIG. 13A, Two parts 300, Although 301 is shown as a separate entity, The allocated part is not a physically separate memory block, It represents the number of cell buffers located somewhere in the pool 128. In a further embodiment, The mechanism to guarantee the minimum bandwidth to be disclosed is As described with reference to FIGS. 10 and 11, It is applicable to mechanisms that provide prioritized access to downstream buffers. Referring to FIG. 13B, The downstream buffer pool 228 is logically divided into an allocation part 302 and a dynamic part 208. The dynamic part 208 is Threshold level 22, 204, 206 logically subdivides into cell buffer subsets 208 a-d with priority. As in the case of FIG. 13A, The division of the buffer pool 228 is a logical division, not a physical division. The elements required to implement this guaranteed minimum bandwidth mechanism are shown in FIG. In FIG. Elements similar to those in FIGS. 2 and 11 are given the same reference numbers plus 100 or 200. Note that no new components have been added to the downstream element. That is, This minimum bandwidth guarantee is transparent to downstream elements. A new aspect of the flow protocol is Seen at both the connection and link levels. First, Regarding addition and change of connection level, The D_BS counter D_BS_Counter 122 is By tracking the number of cells scheduled using the dynamic bandwidth transmitted downstream to the receiver 114, Specify resource consumption. This counter is Basically, It has the same function as the BS counter BS_Counter 22 'in FIG. However, In the BS_counter 22, Allocated and dynamically scheduled cell traffic were not distinguished. Similarly, The D_BS limit D_BS_Limit 124 used to limit the number of downstream buffers available to store cells from the transmitter 112 is: It has a function corresponding to the BS limit BS_Limit 24 'in FIG. As described above for link level flow control, Dynamic bandwidth can be statically shared. Dynamic cell traffic can over-allocate the actual number of available buffers. The amount of buffer "D" given to the connection is equal to the product of RTT and dynamic bandwidth plus N2. A_BS counter A_BS_Counter 222 and A_BS limit A_BS_Limit 224 are also By comparing the number sent and the available buffer limit, Track and limit the number of cells that a connection can transmit, respectively. However, These values are Applies only to layout-type cells. Assignable cells are: Cells identified as requiring allocated bandwidth (guaranteed minimum bandwidth) for transmission. The limit information is set in the connection initialization time, It may be increased or decreased depending on the change of the guaranteed minimum bandwidth. If the connection has no allocated components, The A_BS limit A_BS_Limit 224 becomes zero. A_BS counter A_BS_Counter 222, as well as, A_BS limit A_BS_Limit 224 is D_BS counter D_BS_Counter 122 described above, as well as, D_BS limit D_BS_Limit 124 is additionally provided. The amount of buffer "A" allocated to the connection is equal to the product of RTT and dynamic bandwidth plus N2. The actual number of buffers allocated to allocatable traffic cannot be over-allocated. This allows It is guaranteed that the collision on other connections does not affect the guaranteed minimum bandwidth. The connection has been enqueued to the cell, When the A_BS counter BS_Counter 222 and the A_BS counter BS_Counter 224 indicate that there are no A buffers, The connection is Loss or runs out of bandwidth allocated by the corresponding upstream switch. If the connection is flow controlled below its allotted rate, The connection is Until congestion eases, Part of the allocated area in the exchange will be lost. this is, Multiple sources, each with a minimum guaranteed rate on the same connection, This occurs in the case of multipoint-to-multipoint (M2P) switches merging at a single exit point that is less than the sum of the source rates. In one embodiment of the disclosed mechanism, wherein the transmitter element is part of a switch having a switch flow control function, When there are no more A buffer states, Further transmission of the connection-type cell traffic in the exchange is prohibited. The buffer return policy for each connection is First, Return the buffer to the allocation pool until the A_BS counter BS_Counter 222 equals zero. next, While decreasing the D_BS counter D_BS_Counter 122, Returns a buffer to the dynamic pool. As described above for connection level flow control and priority access, A transmission counter 126 and a priority 208 are provided. At the link level, To enable a guaranteed minimum cell rate for each connection, The following elements are added: A link A_BS counter Link_A_BS_Counter 250 is added to the FSPP 116. This counter is Identified as requiring allocated bandwidth, TSPP118 cell buffer 128, All cells "running" between the FSPP 116 and the downstream switch body, including the cells in 228, are tracked. When the connection level update function (described below) is executed, The counter 250 is decremented for each connection by the same amount as the A_BS counter. The link BS limit Link_BS_Limit 152 is Reflects the total number of buffers available only to dynamic cells that do not include allocated buffers. But, A link BS counter Link_BS_Counter 150 reflects the total number of allocated and dynamic cells transmitted. Therefore, If the link BS counter Link_BS_Counter 150 (all running cells buffered or in the downstream switch body) is greater than the link BS limit Link_BS_Limit 152 (the maximum number of available dynamic buffers), The connection cannot use its dynamic bandwidth. This means Necessary to ensure that congestion does not affect the allocated bandwidth. In one embodiment, To eliminate the possibility of outdated (ie, less frequent) connection level updates, Sum of each A_BS limit A-BS_Limit 224 value, That is, Cell buffer space 300 allocated for each connection, The sum of 302 is Less than the allocated cells that are actually allocated. Update and check events are also implemented in the disclosed allocation / dynamic flow control mechanism. The downstream element 114 is The priority list and the VBR priority 0 list become empty, If the update queue is full, Or When the maximum update interval (not shown) is reached, Send the connection level update cell. In the upstream terminal 112, The update cells are analyzed to identify the appropriate queue. The FSPP 116 A_BS counter A_BS_Counter 222 for the queue, as well as, Adjust the D_BS counter D_BS_Counter 122. in this case, Since FSPP 116 cannot distinguish between allocated and dynamic buffers, As mentioned above, Return up to A cell buffers first, Then return up to D items. The number A of buffers returned for each connection is subtracted from the link A_BS counter Link_A_BS_Counter250. Other link-level elements, such as link transmission counter 154, having similar functions as described above in link-level flow control, Used in connection with the minimum guaranteed bandwidth mechanism. Also, As mentioned above, A further embodiment of the mechanism is Threshold 202, 204, It works with link-level flow control that introduces prioritized access to downstream buffer resources 228 using 206. The function of these elements has been described above. Less than, 4 shows an example of a typical initialization in a flow control link according to the present invention. The downstream element has 3000 buffers; Links are short paths, Thus the RTT * bandwidth is equal to one cell; 100 allocated connections each requiring 7 (A) buffers, consuming a total of 700 buffers; 3000-700 = 2300 (D) buffers are shared by 512 connections; Link BS limit = 2300. If D_BS_Counter ≧ D_BS_Counter, The queue is Indicating that there is a cell ready for transmission is prevented. In the above embodiment, wherein the upstream element is a switch having a composite bandwidth, Such a situation The queue is removed from the dynamic list, This is caused by preventing queues from being transmitted using dynamic bandwidth. For layout cells, As each cell is enqueued, With that cell, With other enqueued cells, It is determined whether the sum with A_BS_Counter is a number greater than A_BS_Limit. If a negative decision is made, The cell is enqueued, The queue is added to the priority list. If a positive determination is made, Connections are prohibited from transmitting further cells through the upstream element 12 switch body. FIG. 15 shows an initialization process of the upstream element 112 shown in FIG. Basically, The counter and register set of the upstream element 12 ′ in FIG. 3A (when priority access to the shared buffer resource is not permitted) and the upstream element 12 ″ (when priority access is permitted) in FIG. 12A Is the same. However, Link A_BS counter Link_A_BS_Counter 250 is initialized to zero; Connection level allocation type and dynamic BS counter 122, 222 is set to zero; Connection level allocation type and dynamic BS_limit 124, 224 is N _A And N _B The exception is that it is set to Similarly, at the downstream terminal at the connection level, the allocation and dynamic buffer limit Buffer_Limit and the buffer counter Buffer_Counter are set such that the buffer limit Buffer_Limit has the bandwidth value for each traffic type (ie, BW _A = Allocated cell bandwidth, BW _D = Dynamic cell bandwidth). . Further, each cell to be transmitted, when received from the switch body, is identified as an allocatable cell or one that requires dynamic bandwidth. FIG. 15B shows a number of the same discrimination processes as those shown in FIGS. 3B and 12B performed before transmitting a cell from the upstream element 112 to the downstream element 114. However, the difference is that the over-allocation of the buffer state for each connection is checked only for dynamic traffic, performed by subtracting Link_A_BS_Counter from Link_BS_Counter and comparing the result with Link_BS_Limit. Link width based over-allocation is calculated by comparing Link_BS_Limit (which tracks both allocated and dynamic cell traffic) and Link_A_BS_Counter with Link_BS_Limit. Similarly, over-allocation in downstream elements is checked at the connection level for both allocated and dynamic traffic. As mentioned above, the mechanism that provides the guaranteed minimum bandwidth can be used with or without the prioritized access mechanism, but FIGS. 15A and 15B show that The latter aspect is shown for completeness. As mentioned above, connection level flow control known in the art relies on separate controls for individual connections. In particular, between network elements, such as a transmitting element and a receiving element, control is from the transmitter queue to the receiver queue. Thus, a single queue Q at the transmitter element _A Are the four queues Q associated with a single receiver processor _W , Q _X , Q _Y , And Q _Z Even in the situation shown in FIG. 16, which is the source of the data cells for, the prior art does not provide a mechanism to handle such situations. In FIG. 16, the transmitter element 10 is provided in an FSPP element having an associated FSPP 11. The receiver element 12 is provided in a TSPP element having an associated TSPP13. The FSPP 11 and TSPP 13 used in FIG. 16 have the same functions as those described above, such as link-level flow control, priority access to the shared downstream buffer, and the guaranteed minimum cell rate at the connection level, in addition to the connection-level flow control mechanism. Selectively provide programmable features. Whether one or more of these enhancements is used with connection-level flow control is at the discretion of the system designer. Yet another function provided by the FSPP and TSPP according to the present invention is a function of uniformly handling a group of receiver queues for connection level flow control. In FIG. 16, instead of using four parallel connections, the mechanism of the present invention uses one connection 16 in link 14. Connection 16 has four queues Q _W , Q _X , Q _Y , And Q _Z , But for connection level flow control, these four queues are essentially treated as a single union. This is required because if N2 is set to a small value of 10 or less, some network elements require the use of flow control services but cannot handle the updated cell processing bandwidth. (See above for connection level flow control). If N2 is set to a large value such as 30 for a large number of connections, a large amount of downstream buffers is required. This is due to buffer or phaning, where the buffer is not busy but is considered to be busy upstream due to infrequent update events. This mechanism is also useful for terminating a virtual channel connection (VCC) within a VPC when applying flow control to a virtual path connection (VPC). This function of grouping the receiver queues _W , Q _X , Q _Y , And Q _Z By manipulating the queue descriptor associated with. Referring to FIG. 17, a queue descriptor for the receiver queue is shown. As shown, the queue Q _W , Q _X , And Q _Y Is provided on the left side and has almost the same characteristics. One of the first fields according to the present disclosure is a bit denoted by a symbol J. This bit, when set, indicates that the associated queue is treated as part of the receiver's joint connection. Instead of holding all connection level flow control information in each queue descriptor of each queue in the group, a particular flow control element is held in only one of the group's queue descriptors. In the case shown, this one queue is queue Q _Z It is. Queue Q _W , Q _X , And Q _Y In each of the descriptors for _Z Is the offset or pointer to the set of flow control elements in the descriptor for This pointer field may serve another function if the J bit is not set. The buffer limit register (indicated by reference numeral Buff_Limit in FIG. 17) and the N2 limit N2_Limit are locally held in each descriptor. The joint buffer counter (Jt_Buff_cntr), the joint N2 counter (Jt_N2_Cntr), and the joint transmission output counter (Jt_Fwd_Cntr) for all the queues in the group include the queue Q _Z Is kept in the descriptor for Queue Q _W , Q _X , And Q _Y The same counter in the descriptor for is not used. The joint counter performs the same function as each counter, such as the function at the connection level shown in FIG. 2, but is incremented or decremented as appropriate by the operation performed for each queue. Thus, for example, when a buffer cell corresponding to any one of the queues receives a data cell or releases a data cell, the joint buffer counter is updated. The same is true for the joint N2 counter and the joint delivery counter. In another embodiment of the flow control mechanism described above, each send counter is replaced by a receive counter. Similarly, in another embodiment of the mechanism of the present disclosure, the joint send counter is replaced by a joint receive counter, depending on where the queues are held. Only embodiments that include a delivery counter and a joint delivery counter are illustrated. Not all descriptor elements per queue are superseded by the common descriptor function. A buffer limit (indicated by reference numeral Buff_Limit in FIG. 17) is set and referred to for each queue. Therefore, the joint buffer counter is compared with the buffer limit register of each queue. Further, the buffer limit may be set as a joint buffer limit and a common limit value may be held instead of holding individual limit values. The strategy is to set the same buffer limit for all TSPP queues associated with a single joint buffer counter. As described above, the update event is triggered when the joint N2 counter reaches the queue level N2 limit. The policy is to set all N2 limits for all queues associated with a single joint flow control connection to the same value. When the test cell of the connection is received, an attempt to change the reception counter associated with the reception queue results in a change in the joint reception counter. Therefore, the indirect level obtained by the number of joints is applicable to both data cells and test cells. At the transmitter element 10, only one set of upstream flow control elements is retained. At connection establishment time, as far as the upstream element is concerned, the joint connection is established as a single point-to-point connection. Thus, in the embodiment of FIG. 16, instead of keeping four sets of upstream elements, the mechanism of the present disclosure requires one set of elements (transmit counter, BS counter, BS limit each having the functions described above). It is only. Once the joint flow control is established, another TSPP queue for additional connections is added. To do this, each new queue must have the same N2 limit and buffer limit. The queue for the additional connection will refer to the common joint N2 counter and either the joint send counter or the joint receive counter. As described above, when J = 1, the joint number (Joint Number) field is used as an offset into the group descriptor. Queue Q _Z 17, the joint number of the group descriptor indicates itself. This is also true for point-to-point connections (VCC-to-VCC instead of VPC-to-VCC as shown in FIG. 16). In this case, each joint number points to its own descriptor. Thus, the implementation of point-to-point and the point-to-multipoint connection described is simplified. While the preferred embodiment of the invention has been described, other embodiments using this concept will be apparent to those skilled in the art. The above examples of the present invention are merely illustrative, and the actual scope of the present invention should be determined by the following claims.

[Procedure amendment] Article 184-8, Paragraph 1 of the Patent Act [Date of submission] February 7, 1997 (1997.2.7) [Content of amendment] Furthermore, each cell to be transmitted must be transmitted from the exchange itself When received, it is identified as requiring allocated cells or dynamic bandwidth. FIG. 15B shows a number of the same discrimination processes as those shown in FIGS. 3B and 12B performed before transmitting a cell from the upstream element 112 to the downstream element 114. However, the difference is that the buffer state over-allocation based on link width is checked only for dynamic traffic, done by subtracting Link_A_BS_Counter from Link_BS_Counter and comparing the result with Link_BS_Limit. The per-connection over-allocation is calculated by comparing D_BS_Counter with D_BS_Limit and checking for programming or other faults per connection by comparing A_BS_Counter with A_BS_Limit. Similarly, over-allocation in downstream elements is checked at the connection level for both allocated and dynamic traffic. As mentioned above, the mechanism that provides the guaranteed minimum bandwidth can be used with or without the prioritized access mechanism, but FIGS. 15A and 15B show that The latter aspect is shown for completeness. As mentioned above, connection level flow control known in the art relies on separate controls for individual connections. In particular, between network elements, such as a transmitting element and a receiving element, control is from the transmitter queue to the receiver queue. Thus, a single queue Q _A of the transmitter element is the source of data cells for the four queues Q _W , Q _X , Q _Y , and Q _Z associated with a single receiver processor, FIG. Even in the situations shown in Figure 1, the prior art does not provide a mechanism to handle such situations. In FIG. 16, the transmitter element 10 is provided in an FSPP element having an associated FSPP 11. The receiver element 12 is provided in a TSPP element with an associated TSPP 13 [FIG. [Procedure of Amendment] Article 184-8, Paragraph 1 of the Patent Act [Date of Submission] September 15, 1997 (September 15, 1997) [Details of Amendment] Claims 1. A method for providing prioritized access to shared buffer resources of a receiver device by data cells transmitted from a transmitter device to a receiver device across a link, the method comprising: Determining at the transmitter device a first count indicating the number of data cells transmitted to the receiver device across the link for storage in the shared buffer resource; Storing at the transmitter device at least one buffer threshold that is the maximum number of buffers that can be allocated to data cells of a priority level, wherein the at least one buffer threshold, if present, The transmitter device identifies whether there is less than or equal to a count, and If the at least one buffer threshold thresholding to respond is less than the first count, prohibits the transmission of the data to be the transmission by the transmitter device, the method consisting of the steps. 2. The method of claim 1, wherein determining the priority level further comprises determining the priority level of the data cell to be transmitted based on a connection of the data cell. 3. The method of claim 1, wherein N-1 buffer thresholds are stored for each of the determined N priority levels. 4. The method of claim 4, further comprising modifying the first count when a data cell is transmitted by the transmitter device across the link to the receiver device. 5. Modifying the first count further comprises receiving a value from the receiver device at the transmitter device indicating a number of buffers allocated to data cells in the shared buffer resource, and responsive to the value. 2. The method of claim 1, comprising modifying the first count. 6. Storing the at least one buffer threshold further comprises dynamically adjusting the at least one buffer threshold in response to an empirical analysis of data cells to be transmitted by the transmitter device. The method of claim 1, comprising: 7. Dynamically adjusting the at least one buffer threshold further comprises: responsive to an empirical analysis at the transmitter device of a priority of a data cell to be transmitted by the transmitter device. 7. The method of claim 6, comprising dynamically adjusting the buffer threshold. 8. Storing the at least one buffer threshold comprises dynamically adjusting the at least one buffer threshold in response to an empirical analysis by the transmitter device of the utilization of the shared buffer resource. The method of claim 1 comprising: 9. The method of claim 1, further comprising establishing at the transmitter device a maximum number of data cells that can be stored in the shared buffer resource in the receiver device. 10. The method of claim 9, wherein the step of inhibiting transmission further comprises the step of inhibiting transmission of the data cells when the first count is greater than or equal to the maximum number of storable data cells. 11. The method of claim 1, wherein said data cells are ATM cells. 12. The method of claim 1, wherein said data cells are of a predetermined fixed length. 13. A method for providing prioritized access via a communication resource to a shared buffer resource of a receiver element by a data cell of a transmitter element having a corresponding memory resource, the method comprising: Determining a maximum number of available priority allocation buffers for each of the plurality of data cell priorities and storing each of the maximum numbers in the memory resource; Storing in the memory resource a value indicating the number of data cells transmitted to the receiver element via the communication resource for storage, and modifying the value when the data cell is transmitted to the shared buffer resource. Identifying, in said transmitter element, a priority assigned to a first data cell, wherein said value is a priority assignment buffer; Transmitting the first data cell to the receiver element via the communication resource by the transmitter element for storage in the shared buffer resource if less than a respective maximum number of fas. . 14． 14. The method of claim 13, further comprising dynamically adjusting a maximum number of the priority allocation buffers in response to empirical analysis of data cells to be transmitted by the transmitter element. 15. The step of dynamically adjusting the maximum number of the priority allocation buffers further comprises: an empirical analysis at the transmitter element of the priority of the data cells of the transmitter element. 15. The method according to claim 14, comprising the step of dynamically adjusting according to: 16. 14. The method of claim 13, further comprising dynamically adjusting a maximum number of the priority allocation buffers in response to empirical analysis at the transmitter element of usage of the shared buffer resources. 17． 14. The method of claim 13, wherein identifying the priority further comprises identifying the priority assigned to the first data cell based on a connection of the first data cell. 18. The step of defining the maximum number includes, for the N data cell priorities, further defining a maximum number of one of the priority assignment data cells equivalent to the buffer resource capacity, and equivalent to the data cell priority threshold. A method comprising the steps of defining a maximum number of N-1 priority priority cells. 19. Modifying the value further comprises receiving, at the transmitter element, from the receiver element, a value indicating a number of buffers allocated to data cells in the shared buffer resource, and responsive to the received value, 14. The method of claim 13 comprising modifying. 20. 14. The method of claim 13, wherein said data cells are ATM cells. 21. 14. The method of claim 13, wherein said data cells are of a predetermined fixed length. 22. A communication medium having a transmitter terminal and a receiver terminal; a transmitter network element provided at the transmitter terminal of the communication medium for transmitting a data cell from an output port to the communication medium; and the communication of the communication medium. And a receiver network element for receiving the data cell at an input port, the receiver network element having an associated priority shared buffer resource for storing the data cell, the transmitter network element being provided with the data cell. Analyzing the cell to determine a priority associated with the data cell, wherein the transmitter network element, when a buffer of the shared buffer resource is available for storing data cells of the same priority, A prioritized buffer sharing device for transmitting the data cell to the receiver network element via the communication medium for storage in the shared buffer resource. 23. A communication medium having a transmitter terminal and a receiver terminal; a transmitter provided at the transmitter terminal of the communication medium, for transmitting a data cell across the communication medium; and a communication terminal of the communication medium. And a receiver for receiving the data cell, the receiver having an associated priority shared buffer resource for storing the data cell, wherein the transmitter analyzes the data cell, and Determining an associated priority, wherein the transmitter stores the data cell in the shared buffer resource when a buffer of the shared buffer resource is available for storing a data cell of the same priority. Transmitting to the receiver via the communication medium, the transmitter further comprising a processing element and a memory element, wherein the memory element includes A first value indicating the maximum number of data cells that can be stored in the shared buffer resource with priority; and a second value indicating the number of buffers transmitted to the shared buffer resource with priority for storage there. A priority-shared buffer device that retains at least one threshold value that indicates a maximum number of data cells of each priority that can be stored in the shared buffer resource with priority. 24. 24. The shared buffer with priority according to claim 23, wherein the processing element prohibits transmission of the data cell from the transmitter to the receiver when the second value is not smaller than the first value. apparatus. 25. 24. The priority-shared buffer sharing device according to claim 23, wherein the processing element inhibits transmission of the data cell when the second value is not smaller than each threshold value for the priority of the data cell. 26. 24. The prioritized buffer sharing device of claim 23, wherein the at least one threshold value is dynamically adjustable by the processing element based on empirical analysis of data cells to be transmitted to the shared buffer resource. 27. 27. The prioritized system of claim 26, wherein the at least one threshold value is dynamically adjustable by the processing element based on empirical analysis at the transmitter of data cells to be transmitted to the shared buffer resource. Buffer sharing device. 28. 24. The prioritized buffer sharing device of claim 23, wherein the at least one threshold value is dynamically adjustable by the processing element based on empirical analysis of the state of use of the shared buffer resource. 29. 24. The buffer sharing apparatus with priority according to claim 23, wherein there are N possible data cell priorities and N-1 threshold values. 30. The second value is updated by the processing element in response to data received at the transmitter from the receiver that reflects a number of data cells currently stored in the shared buffer source. 23. The buffer sharing device with priority according to 23. 31. 23. The shared buffer with priority according to claim 22, wherein the data cell is an ATM cell. 32. 23. The shared buffer with priority according to claim 22, wherein the data cell has a predetermined fixed length.

────────────────────────────────────────────────── ─── Continuation of front page    (81) Designated countries EP (AT, BE, CH, DE, DK, ES, FI, FR, GB, GR, IE, IT, L U, MC, NL, PT, SE), OA (BF, BJ, CF) , CG, CI, CM, GA, GN, ML, MR, NE, SN, TD, TG), AP (KE, LS, MW, SD, S Z, UG), EA (AM, AZ, BY, KG, KZ, MD , RU, TJ, TM), AL, AM, AT, AU, AZ , BB, BG, BR, BY, CA, CH, CN, CZ, DE, DK, EE, ES, FI, GB, GE, HU, I L, IS, JP, KE, KG, KP, KR, KZ, LK , LR, LS, LT, LU, LV, MD, MG, MK, MN, MW, MX, NO, NZ, PL, PT, RO, R U, SD, SE, SG, SI, SK, TJ, TM, TR , TT, UA, UG, US, UZ, VN (72) Inventor Caldara, Stephen A             Massachusetts, United States             01776, Sudbury, Horse Pond Low             C 220 (72) Inventors Manning, Thomas A             Massachusetts, United States             01532, Northborough, Summer Street               26th

Claims (1)

[Claims] 1. By data cells transmitted from the transmitter device to the receiver device across the link , Method for providing prioritized access to a shared buffer resource of a receiver device So,   Determining the priority level of the data cells to be transmitted by the transmitter device;   The receiver device across the link for storage in the shared buffer resource Generating a first count at the transmitter device indicating the number of data cells transmitted to And   At least the maximum number of buffers that can be allocated to data cells of each priority level Storing one buffer threshold in said transmitter device;   If the at least one buffer threshold is present, the first Whether or not there is less than the count by the transmitter device,   The at least corresponding to the priority level of the data cell to be transmitted If one buffer threshold is less than or equal to the first count, the The transmission of data to be transmitted is prohibited by the transmitter device, the method comprising: Law. 2. The step of determining the priority level further comprises the step of transmitting the data set to be transmitted. Determining the priority level of the data cell based on the connection of the data cell The method of claim 1 comprising: 3. N-1 buffer thresholds for each of the N determined priority levels A method according to claim 1, wherein 4. Data cells across the link by the transmitter device Modifying the first count upon transmission to a receiver device. The method of claim 4. 5. Modifying the first count may further comprise modifying the first count at the transmitter device. Allocated from the receiver device to a data cell in the shared buffer resource. Receiving a value indicating the number of buffers and modifying the first count according to the value; The method of claim 1, comprising floors. 6. Storing the at least one buffer threshold further comprises: Depending on an empirical analysis of the data cells to be transmitted by the transmitter device, 2. The method of claim 1, further comprising the step of dynamically adjusting at least one buffer threshold. The described method. 7. The step of dynamically adjusting the at least one buffer threshold comprises: Further, the transmitter device has a priority of a data cell to be transmitted by the transmitter device. At least one buffer threshold depending on empirical analysis in the 7. The method of claim 6, further comprising the step of: dynamically adjusting. 8. Storing the at least one buffer threshold comprises: storing the at least one buffer threshold. At least one buffer threshold is used for the shared buffer resource Dynamically adjusting according to an empirical analysis by the transmitter device of the state Item 7. The method according to Item 1. 9. Data cells that can be stored in the shared buffer resource in the receiver device. The method of claim 1, further comprising establishing a maximum number at the transmitter device. . 10. The step of prohibiting the transmission further comprises the step of storing the first count in the storage state. Prohibiting the transmission of the data cells when the number of data cells is equal to or greater than the maximum number. 10. The method of claim 9, comprising: 11. The method of claim 1, wherein said data cells are ATM cells. 12. The method of claim 1, wherein said data cells are of a predetermined fixed length. 13. Reception by data cells of the transmitter element with corresponding memory resources Access to the shared buffer resources of the A method of providing,   Priority available in the shared buffer resource by the transmitter element Determining a maximum number of allocated buffers for each of the plurality of data cell priorities; Storing each of the maximum number in said memory resource;   The communication resource for storage in the shared buffer resource by the transmitter element. A value indicating the number of data cells transmitted to the receiver element via the Stored in resources,   Modifying the value when a data cell is transmitted to the shared buffer resource,   At the transmitter element, identifying a priority assigned to the first data cell;   If the value is less than the respective maximum number of priority allocation buffers, the first data The cell is stored by the transmitter element for storage in the shared buffer resource. Transmitting to the receiver element via a communication resource. 14． The maximum number of the priority allocation buffers is transmitted by the transmitter element. Dynamically adjust according to empirical analysis of data cells to be 14. The method of claim 13, further comprising the step of: 15. The step of dynamically adjusting the maximum number of the priority allocation buffers further comprises: , The maximum number of priority allocation buffers is determined by the priority of the data cells of the transmitter element. Dynamically adjusting according to previous empirical analysis at said transmitter element The method according to claim 14. 16. The maximum number of the priority allocation buffers is determined by the Dynamically adjusting responsive to empirical analysis at the transmitter element for use. 14. The method of claim 13, comprising providing. 17． Identifying the priority may further comprise assigning the priority to the first data cell. Identifying the determined priority based on the connection of the first data cell. 14. The method of claim 13, comprising floors. 18. The step of defining the maximum number further comprises, for the N data cell priorities, ,   Defines the maximum number of one priority-assigned data cell equivalent to buffer resource capacity And   N-1 maximum of priority assigned cells equivalent to data cell priority threshold A method consisting of steps, defining a number. 19. Modifying the value may further comprise modifying the receiver element at the transmitter element. The number of buffers allocated to data cells in the shared buffer resource 14. The method of claim 13, further comprising the step of receiving the indicated value and modifying said value according to said received value. The described method. 20. 14. The method of claim 13, wherein said data cells are ATM cells. 21. 14. The method of claim 13, wherein said data cells are of a predetermined fixed length. 22. A communication medium having a transmitter terminal and a receiver terminal;   A data cell is provided at the transmitter terminal of the communication medium for passing data cells across the communication medium. A transmitter for transmitting   An association provided in the communication terminal of the communication medium for storing the data cell A receiver having a shared buffer resource with priority   The transmitter analyzes the data cell and determines the superior associated with the data cell. Determine your priorities,   The transmitter is configured so that the buffer of the shared buffer resource has the same priority The data cell, if available for storage of the shared buffer resource. Priority buffer to be transmitted to the receiver via the communication medium for storage in the storage medium. Sharing device. 23. The transmitter further comprises a processing element and a memory element, wherein the memory element Is the processing element   Indicates the maximum number of data cells that can be stored in the shared buffer resource with priority A first value;   The buffer transmitted to the prioritized shared buffer resource for storage there. A second value representing the number of fas;   The maximum number of data cells of each priority that can be stored in the shared buffer resource with priority 3. The method of claim 2, wherein at least one threshold value each representing a large number is held. 2. The buffer sharing device with priority according to 2. 24. The processing element may determine if the second value is not less than the first value. 24. Prohibiting transmission of said data cells from said transmitter to said receiver. Priority-based buffer sharing device. 25. The processing element may determine that the second value is a different value for the priority of the data cell. A request to prohibit transmission of the data cell if the value is not smaller than a threshold value. Item 24. The buffer sharing device with priority according to Item 23. 26. The at least one threshold value is the shared buffer resource Based on the empirical analysis of the data cells to be transmitted to the 24. The buffer sharing apparatus with priority according to claim 23, wherein the apparatus is adjustable. 27. The at least one threshold value is the shared buffer resource Based on empirical analysis at the transmitter of data cells to be transmitted to the 27. The prioritized buffer sharing of claim 26, wherein the buffering is dynamically adjustable by processing elements. apparatus. 28. The at least one threshold value is the shared buffer resource Can be dynamically adjusted by the processing element based on empirical analysis of usage of A buffer sharing device with priority according to claim 23. 29. There are N possible data cell priorities and N-1 threshold values. 24. The shared buffer with priority according to claim 23, wherein 30. The second value received at the transmitter from the receiver; For data that reflects the number of data cells currently stored in the shared buffer source 24. The shared buffer with priority according to claim 23, which is updated by the processing element in response to the request. apparatus. 31. 23. The priority buffer according to claim 22, wherein the data cells are ATM cells. Key sharing device. 32. 23. The priority buffer according to claim 22, wherein the data cell has a predetermined fixed length. Sharing device.