CA2159528A1 - Implementation of selective pushout for space priorities in a shared memory asynchronous transfer mode switch - Google Patents

Implementation of selective pushout for space priorities in a shared memory asynchronous transfer mode switch

Info

Publication number
CA2159528A1
CA2159528A1 CA002159528A CA2159528A CA2159528A1 CA 2159528 A1 CA2159528 A1 CA 2159528A1 CA 002159528 A CA002159528 A CA 002159528A CA 2159528 A CA2159528 A CA 2159528A CA 2159528 A1 CA2159528 A1 CA 2159528A1
Authority
CA
Canada
Prior art keywords
queue
packet
packets
priority
queues
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Abandoned
Application number
CA002159528A
Other languages
French (fr)
Inventor
Abhijit Kumar Choudhury
Ellen Louise Hahne
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
AT&T Corp
Original Assignee
AT&T Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by AT&T Corp filed Critical AT&T Corp
Publication of CA2159528A1 publication Critical patent/CA2159528A1/en
Abandoned legal-status Critical Current

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • H04L12/54Store-and-forward switching systems 
    • H04L12/56Packet switching systems
    • H04L12/5601Transfer mode dependent, e.g. ATM
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/10Packet switching elements characterised by the switching fabric construction
    • H04L49/104Asynchronous transfer mode [ATM] switching fabrics
    • H04L49/105ATM switching elements
    • H04L49/108ATM switching elements using shared central buffer
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • H04L12/54Store-and-forward switching systems 
    • H04L12/56Packet switching systems
    • H04L12/5601Transfer mode dependent, e.g. ATM
    • H04L2012/5638Services, e.g. multimedia, GOS, QOS
    • H04L2012/5646Cell characteristics, e.g. loss, delay, jitter, sequence integrity
    • H04L2012/5651Priority, marking, classes
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • H04L12/54Store-and-forward switching systems 
    • H04L12/56Packet switching systems
    • H04L12/5601Transfer mode dependent, e.g. ATM
    • H04L2012/5678Traffic aspects, e.g. arbitration, load balancing, smoothing, buffer management
    • H04L2012/5681Buffer or queue management

Abstract

A system for queueing and selective pushout and method are disclosed for a packet communications module such as a shared memory asynchronous transfer mode (ATM) switch. The shared memory stores packets in queues, each packet having a field and at most two pointers. Within each queue, the packets having respective space priorities are stored in sub-queues each having the respective space priorities. The packets are stored in these priority sub-queues using a first pointer pointing to the next packet of the same space priority in the queue. The second pointer associated with a stored packet points to the previous packet of greater than or equal space priority in the FIFO order in the queue. The field of a packet is used to store the priority value corresponding to the next packet in FIFO order in the queue, and this field is used by a processor to decide priority sub-queues to serve next. The packets are stored in the queues in a FIFO order using the two pointers and the fields of the packets. Theprocessor controls the selective pushout to push out a packet and uses the two pointers and the fields of the packets to restore the FIFO order. A method is also disclosed including the steps of storing packets in a queue, with each of the queued packets associated with the two pointers and a field; serving the queue; pushing out packets from the queue; and maintaining queue-lengths and a state information table.

Description

NEW IMPE;EMENTATION OF SELECTIVE PUSHOUT FOR SPACE
PRIORlTIES IN A SHARED MEMORY ASYNCHRONOUS
TRANSFER MODE SWITCH

BACKGROUND OF THE rNVENTION
5 l. Field of the Invention This disclosure relates to packet-switched communications systems. More particularly, this disclosure is directed to a system for q~le~ ng and selective pushout and method for use in packet communication processing modules.
2. D~3_~ivlion of the Related Art In the field of switched broadband integrated services digital networks (BISDN), packet processing such as asynchronous transfer mode (ATM) cell switching may be implemented using a shared-buffer-memory-based archile.,lule, where incoming paclcets such as ATM cells destined for dirre-enl output ports are buffered in queues sharing a common memory with the queues corresponding to those ports. Such packet 15 processing is employed to support a wide spectrum of information services with very different bandwidth and quality of service (QoS) requirements.
A flexible way to provide different QoS is to use a priority me~.h~ni~m inside a packet communications module. Priorities may be of two dirre-ellt types: delay (or time) priorities and loss (or space) priorities. Delay priorities provide prererenlial 20 service to some classes of traffic in order to control their end-to-end delay and delay variation (jitter). Loss (or space) priorities provide plc:rerelllial access to buffer space.
Pushout systems and methods allow the entire buffer space to be shared flexibly among the lirrerent queues as well as among the various space priority classes. As long as the buffer is not completely full, arriving packets of any priority are admitted.
25 When the buffer is full, pushout systems provide space for the arriving packet by pushing out (i.e. overwriting) another packet that is already in buffer. In selective pushout, a packet that arrives to find the buffer full is allowed to enter by pushing out a packet of lower priority. For a packet communications module using first-in-first-out (FIFO) queues, while the physical memory location of the lower priority packet is usurped, the higher priority packet does not take over its position in its logical FIFO
queue, i.e., the pushing and the pushed packets may be d~stine~ to different outputs - and hence belong to d,rre~ t logical FIFO's. The higher priority packet that has arrived joins the tail of the logical FIFO queue for its own output.
Generally, the arriving packet has a choice of lower priority victims. In that case, the arriving packet should consider only those output queues cont~inin~ packets of the lowest priority, and among these, the longest output queue should be selected.
This bias against long queues allows shorter queues to l.ongth~n, thereby creating a degree of fairness in the sharing of buffer space among output queues; it also tends to keep more memory outputs busy, which increases the efficiency of the system. Thus, pushout may be used to equalize queue lengths as well as to enforce space priorities (also referred to herein as loss priorities). If this longest queue has more than one lowest priority packet, the lowest priority packet nearest the head of the queue is selected for pushout. If a packet arrives to a full buffer when there are no lowest priority packets in the shared memory, then a packet of the second-lowest priority is pushed out. If there are no such packets, then the third-lowest priority is considered, and so on. If there are no packets in the buffer of priority lower than the arriving packet, then the arriving packet may push out a packet of the same priority as itself, from the longest queue co~ ing such packets. While this does not immediately 2 0 improve the packet loss rate for that priority class, it does contribute to queue length equalization. If all the packets in the buffer have higher priority than the arriving packet, then the arriving packet is dlu~ped.
In a first implementation of selective pushout, the packets in the queue are in a FIFO linked-list arrangement, with the order of the FIFO queue provided by packet pûinters~ i.e., pointtl~ for pointing from one packet to another. In addition to the QUEUE_LENGTH (QL) counter and the FIRST_PACKET ~FP) and LAST_PACKET
(LP) pointers, the output queue also m~int~inS FIRST_PACKET_OF_CLASS_i (FP[i]) pointers to the first packet of each priority class i. Whenever a pushout is needed, one of these pointers will indicate the packet to be pushed out. Then the queue must be 3 0 scanned to update that pointer. In order to mend the FIFO linked-list after a pushout, 2159S2~

the output queue also maintains another set of pointers PACKET_BEFORE_FIRST_PACKET_OF_CLASS_i (PBFP[i]) to the packet - immediately prece~ling the first packet of each priority class i. This implementation of selective pushout is relatively simple and requires minim~l state information, but in s a worst case situation, the entire shared memory may need to be sc~nned in order to update the FP[i] pointer. For this reason, this first implementation is impractical for large buffer sizes.
In a second implementation of selective pushout shown in FIGS. lA-lB, each output port maintains its FIFO queue as a doubly-linked-list, so that the FIFO order 10 may be re-established whenever a packet is pushed out from the middle of the queue.
In addition, each space priority class has its own singly-linked chain through the FIFO
queue. These priority chains allow the ATM switch to quickly find the appropriate packet for a pushout.
As shown in FIG. lA, every packet needs two poinle~ related to the FIFO
5 order. In an exemplary embodiment for ~ritching fixed length ATM packets (or cells), the two pointers are: a NEXT_PACKET (NP) pointer and a PREVIOUS_PACKET_IN_FIFO_ORDER (PP) pointer. Each packet also needs one pointer related to its priority chain, which, in the above example, is the NEXT_PACKET_OF SA~E_CLASS (NS) pointer. In addition to the 2 o QUEUE_LENGTH (QL) counter and the FIRST_PACKET (FP) and LAST_PACKET
(LP) pointers, each output queue also needs two other pointers for each space priority class i = l, 2, ...,P, in order to m~intain the priority chains: a FIRST_PACKET_OF CLASS_i (FP~i]) and a LAST_PACKET_OF_CLASS_i (LP[i]) pointer. As shown in FIG. lB, when the packet occul,ying location X2 is pushed out 25 by a newly arrived packet, the new packet joins the tail of the queue. The various pointers are then updated to re-establish the FIFO order after the pushout as shown in FIG. lB. The second implementation of selective pushout thus uses more state information than the first implementation to provide for generally faster processing 215952~

speeds, but the additional state information requires more space in the shared memory.
SUMMARY
A system.is disclosed for queueing and selective pushout of packets such as asynchronous transfer mode (ATM) cells in a packet communications module. In an s exemplary embodiment, the packet communications module is an ATM switch, including a processor with stored programs, a shared memory, and a plurality of input and output ports. The ATM switch receives a plurality of packets on its input ports, stores them temporarily in the shared memory, and transmits each packet on an appropriate output port. The memory is configured for storing the packets in a 10 plurality of queues, each queue being associated with packets received at a colle~onding one of the output ports. All the output port queues share a common memory space. The processor is configured for admitting received packets to the shared memory by creating space if needed using selective pushout on packets that are already queued therein.
Each packet is queued using a first and second pointer and a field. Within each output port queue, the packets having respective space priorities are stored in a plurality of sub-queues each associated with the respective space priorities. The packets are stored in these priority sub-queues using the first pointer which points to the next packet of the same space priority in the output queue. The second pointer 2 o associated with a stored packet points to the previous packet of space priority greater than or equal to itself in the FIFO order in the output queue. The field associated with a packet is used to store the priority value corresponding to the next packet in FIFO
order in the output queue, and this field is used by the processor to decide which priority sub-queue to serve next. The packets are stored in the output queue in a FIFO
2 5 order using the first and second pointers and the fields associated with the packets. The processor controls the selective pushout to push out a first packet and uses the first and second pointers and the fields of the rem~ining packets to re-establish the FIFO order.
A method is also disclosed for queueing and selective pushout of packets in the shared memory of the packet communications module. The method includes the stepsof storing packets in the at least one queue, and within the at least one queue, storing the packets in a plurality of priority sub-queues corresponding to the space priorities of the packets, where the at least one queue and the sub-queues have some data 5 structures associated with them and the packets are associated with a first pointer and second pointers and a field; serving the output queue; pushing out packets from the output queue; and maintaining queue-lengths and a state information table. The data structures associated with the at least one queue are a queue-length counter and a field indicating which priority is first in the FIFO order; and the data structures associated lo with each priority sub-queue are a first and a second pointer pointing to the first and last packets respectively in the sub-queue, and a field indicating the priority of the previous cell in the FIFO order whose priority is greater than or equal to that of the sub-queue under consideration.
The step of storing a packet in the at least one output queue includes the steps15 of receiving an incoming packet having an ~e.~oçi~te~ space priority; using selective pushout to create space if needed; storing the packet in the priority sub-queue corresponding to the space priority of the packet using the first pointer; storing in the field associated with each packet the priority value corresponding to the next packet in FIFO order; m~int~ining a FIFO order among the packets queued in the output 20 queue using the field and the second pointer that points to the previous packet of greater or equal priority in the FIFO order; and incremçntin~ the queue-length counter and updating the other data structures associated with the at least one queue and the priority sub-queues contained therein.
The step of serving the at least one output queue includes the step of 2S transmitting the first packet in the FIFO order in the at least one output queue;
updating the second pointers; and decrem~nting the queue-length counter and updahng the other data structures associated with the at least one queue and the priority sub-queues contained therein.

21S952g The step of pushing out the first packet includes the steps of determining a longest queue of the at least one output queue containing a packet of the lowestpriority lower than or equal to that of the incoming packet; pushing out the first packet from the lowest priority sub-queue within the longest queue; decrementing the queue-length counter and updating the other data structures associated with the at least one queue and the priority sub-queues contained therein; and restoring the FIFO order of the at least one output queue using the first and second pointers and the fields of the queued packets.
The method includes the step of m~int~ining the queue-length of the at least 0 one output queue and a state inforrnation table including identifying information for the at least one output queue and information about the priority of packets contained therein; and the step of determining from the state information table a longest queue containing a packet of the lowest priority lower than or equal to that of the incoming packet.
BKIEF DESCRIPTION OF THE DRAWINGS
The fealules of the disclosed implementation of the selective pushout system and method will become more readily app~"ll and may be better understood by referring to the following detailed description of an illustrative embodiment of the present invention, taken in conjunction with the accompanying drawings, where:
FIGS. lA-lB illustrate an example of an implementation of selective pushout using three packet pointers;
FIG. 2 illustrates a block diagram of the components of the packet communications module using the implementation of selective pushout system and method using t~vo packet poinle,~, and the field disclosed herein;
FIG. 3 illustrates a flow diagram of the operation of the disclosed system for queueing and selective pushout and method;
FIG. 4 illustrates a flow diagram illustTating the storage of packets;
FIG. 5 illustrates a flow diagram illustrating selective pushout of packets;

21~952~

FIG. 6 illustrates a flow diagram illustra~ng the tr~ncmi~cion of packets;
FIGS. 7-8 illustrates an example of the implementation of the selective pushout system and method disclosed herein; and FIG. 9 illustrates a state information table.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Referring now in specific detail to the drawings, with like reference numerals identifying similar or identical elements, as shown in FIG. 2, the present disclosure describes a shared-buffer-memory-based packet communications module l 0; for example, an ATM switch module. The shared-buffer-memory-based packet communications module l0 disclosed herein processes packets, which are herein defined as units of data that are addressed with routing information. In the context of packet processing systems, the packets are allowed to be of any length, or may be of a length not greater than a predetermined mL~imulll.
In the exemplary embodiment described herein, the disclosed implementation of the system for queueing and selective pushout and method are illustrated for a packet switch with multiple input ports and multiple output ports, such as an asynchronous transfer mode (ATM) switch, using fixed length packets. For example, the packets may be fixed to be exactly 53 bytes long, where such 53 byte long packets are termed "cells". Alternatively, the packets in a higher protocol layer may have a longer length and are termed "messages" which may be subdivided to generate a plurality of cells for ATM switching.
It is understood that other packet leng~s, including packets of variable length, may be processed by the disclosed implçm~nt~tion of the selective pushout system and method.
2 5 The present disclosure describes an exemplary embodiment of a shared memory ATM switch implementing the system for queueing and selective pushout and method.
In alternative embodiments, the disclosed system for que~leing and selective pushout and method may also be used in conjunction with multiple-input, single-output packet ~15952~

systems (i.e., multiplexers) and single-input, multiple-output packet systems (i.e., demultiplexers). For multiplexers, the selective pushout would be used to allocate memory among packets of different space priorities received from competing inputports.
In the exemplary embodiment, the disclosed packet co~nml~nications module 10, as a shared memory ATM switch, is connected to one or more input ports 12 over which the communications module 10 receives incoming parl~çtC, with individual packets having a priority level. The packet communications module 10 is also connected to one or more output ports 14 for transmitting the switched packets. Input 0 ports are connected to input and output (I/O) CiLCuilly 16 which couples input ports to a processor 18 and a memory 20.
ATM switches are well known in the art. In the exemplary embodiment of the packet communications module 10 as an ATM switch, the I/O cilcuilly 16 may include such meçh~nicm.c as phase-alignment circuitry, serial-to-parallel and parallel-to-serial shift registers. Copies of header portions of incoming packets received from input ports are sent by I/O circuitry 16 to processor 18, and the packets are sent by I/O
ci,~,uil~y 16 to memory 20 for storage. The shared memory 20 shown in FIG. 2 maybe embodied as one or more RAMs.
In an exemplary embodiment, the associated memory 22 in FIG. 2 preferably 2 0 implements logical queues in the form of linked lists, but in alternative embodiments, the queues may be implemented as physical queues or some other data structures in the memory 20 and/or the associated memory 22.
Referring again to FIG. 2, processor 18 controls storage of incoming packets in memory 20 and tr~ncmission of stored packets at the output ports. In the exemplary embodiment disclosed herein, the processor 18 includes associated memory 22 and control logic to m~int~in an input port counter 24, an output port counter 26, queue-length counters 28, a state information table 30, and queues 32 including queue pointers 34, linked lists 36 with packet pointers and fields, and a list 38 of free or 21~9528 g available locations in memory 20, such that for every individual one of the output ports a separate sub-queue in memory buffers for each one of the packet priorities.
The associated memory 22 stores the pointers to the packets in queues 32 which comprise the queue pointers 34 which point to packets in the queue, such as a first 5 packet, as well as the linked-lists 36 using the packet pointers (i.e., which are defined in the present disclosure as pointers for pointing from one packet to another) and fields which indicate a priority value as disclosed herein. In the system for queueing and selective pushout and method disclosed herein, there are two packet pointers NS and PG, as disclosed herein. Also in queues 32 is a free list 38 comprising free list 0 pointers which point to free locations in memory available for insertion of received packets A state information table 30 may be provided and regularly updated to indicate the presence of packets of each priority level in each output queue. The queue-length counters 28 also keeps track of the lengths of the dirre~ output queues.
The input port counter 24 determines which input port to read and when to read 5 the received packets therefrom. Based on the header of an incoming packet, received from I/O circuitry 16, processor 18 determines the priority and the ~ ~in~ion output port of the received packet. Then the processor 18 causes the information in thepacket to be stored in a memory location in memory 20, puts a pointer to that location in the corresponding queue in the associative memory, and updates the queue-lengths 20 in queue-length counters 28 and the state information table 30. Processor 18 also periodically acc~sses the output port that is identified by the present count of the output-port counter, retrieves there~loll, a packet that is at the head of the queue, updates the queue-lengths in queue-length counters 28 and the state information table 30, and causes I/O circuitry 16 to transmit the retrieved packet on the corresponding 25 one of the output ports.
Throughout this disclosure, numerical labels are used for the space priority classes. For consistency, the convention used herein assigns higher numbers to classes with higher priority. For example, when there are two classes, Class 2 packets belong 21~952~
-to the loss-sensitive class, while Class 1 packets can tolerate higher packet loss rates.
Previous implementations of selective pushout described above require either one or three packet pointers per packet, where packet pointers are herein defined as pointers associated with packets which point from one packet to another. In the 5 exemplary embodiment of the system for ql~euein~ and selective pushout and method disclosed herein, only two packet pointers and a field per packet are used. While previous implementations, such as the three pointer implementation, provided redundant information between the FIFO chain traced by the NP and PP pointers and the priority chains traced by the NS pointers, the exemplary implementation of the 0 system for queueing and selective pushout and method di~closed herein reduces the redundancy in information carried by the packet pointers.
OPERATION OF THE DISCLOSED SELECTIVE PUSHOUT IMPLEMENTATION
As shown in FIG. 3, the method rlisclosed herein performs queuing and selective pushout of packets in the shared memory of the packet communications 5 module 10. The method includes the steps of be~...~...~ the packet switching in the packet communications module 10 using selective pushout of packets in the bufferspace in step 40, receiving an arriving packet in step 42, storing a plurality of packets in at least one output queue including a first and second sub-queue associated with a first and second priority, respectively, in step 44, with each of the queued packets 20 associated with first and second pointers and a field; reading stored packets out of memory for de~ e in step 46; mA;nl-A;~ing and updating the data structures associated with queues and packets in step 48; and continuing to receive packets in step 50.
As shown in FIG. 4, the step of storing the arriving packet in step 44 includes 25 the steps of starting to process packets for arrival in step 52, obtaining a memory location for the arriving packet in step 54, de~ hling if more space is required in step 56, getting a free memory location from a free list in step 58 if no additional space is required, and otherwise performing pushout to free a memory location in step 60 if additional space is required. After the memory space is obtained after steps 56-60, the method includes the steps of reading the co~ of the arriving packet to the free memory location in step 62, adding the arriving packet to the sub-queue correspondlng to the pnorlty class of the arnvmg packet m lts tiloshn~hon queue m step 5 64, updating the PLG array in step 66, and incrementing the queue length counter in step 68.
As shown in FIG. 5, the performance of selective pushout in step 60 includes the steps of starting to perform selective pushout in step 70, and selecting priority level and output queue for pushing out a packet in step 72 which includes choosing a 0 longest queue for pushout of the at least one output queue COlll~ g a packet of the lowest priority lower than or equal to that of the incoming packet from a state information table 30 in step 74. Then the performance of selective pushout includes the steps of locating a particular buffer to be freed by selective pushout in step 76, m~int~ining the FIFO order of the queue from which a packet is being pushed out in 15 step 78, updating the PG pointers of packets which are pointing to the packet being pushed out in step 80, and updating the first packet pointer in a sub-queue in step 82, decrementing the queue length counter in step 84.
As shown in FIG. 6, the step of reading out stored packets in step 46 in FIG.
3 includes the steps of starting reading stored packets out for departure in step 86, 2 o verifying that a given queue has a packet for departure in step 88, and serving a packet at the head of the sub-queue collei,~,onding to space priority PF in step 90 which includes determining a packet to depart in step 92, and transmitting the packet over an output port in step 94. Then the reading out of stored packets for departure includes the steps of nulling out PG pointers of packets which are pointing to the packet being 2 5 served in step 96, updating queue length counter, FP pointer, and PF field in step 98, and adding the freed memory location of the departed packet to the end of a free list in step lO0. '5 As shown in FIG. 7-8, the primary data structures of the disclosed selective ~1 j9~2~

pushout system and method are separate logical queues for each space priority class, as shown in an exemplary embodiment as queue 102 for priority class 1 and queue 104 or priority class 2. Secondary state variables are used to establish the FIFO ordering between the packets. For the example herein disclosed for ATM packets, a singly-5 linked chain is maintained for each priority class using theNEXT_PACKET_OF_SAME_CLASS (NS) pointer associated with each ATM packet.
In cor~ to the second implementation described above which uses a NEXT_PACKET pointer, the present selective pushout system and method maintains the overall FIFO ordering using a PRIORITY OF NEXT_PACKET IN FIFO (PN) 0 field to indicate the priority queue that is to be served next in order to establish the overall FIFO order. In the exemplary embodiment in FIGS. 7-8, pointers are indicated by solid arrows and fields are indicated by dashed arrows. The fields used in the disclosed implement~ion of the system for q~leueing and selective pushout and method utilizes less storage than a full pointer does. In order to restore the PIFO order after 5 a pushout, each ATM packet of the present exemplary disclosure uses a PREVIOUS PACKET_OF_GREATER_OR_EQUAL_PRIORITY (PG) pointer to point to the previous packet of priority greater than or equal to itself in the FIFO order.
The reason for using a PG pointer in the present implemPnt~tion of the selectivepushout system and method is illustrated as follows: if a PP pointer is used to point 20 to the previous packet in FIFO order instead of the PG pointer, and there are three space priority classes, a situation may occur that a sequence of a high prionty packet, followed by a low priority packet, followed by a medium priority packet is positioned deep within the FIFO order of some output queue. If the low priority packet is pushed out, establishing the FIFO order in the forward direction is simple. The high priority 2 5 packet is located using the PP pointer of the low priority packet. The PN field of the high priority packet which had the value "low" is replaced with the PN field of the low priority packet, which has the value "medium". Establishing the FIFO order in the reverse direction is more difficult since one cannot copy the PP pointer of the low ~l~9~2~

priority packet into the PP pointer of the medium priority packet since the medium priority packet cannot be located. Although the packet following the low priority packet in FIFO order is known to be a medium priority packet, it may not necessarily be determined which medium priority packet it is. Thus it is more advantageous to use 5 a PG pointer instead of a PP pointer.
The PG pointer is sufficient, since the reverse FIFO order may be used after a pushout. When a particular packet X is vulnerable to pushout (i.e., packet X is nearest to the head of the queue for its priority class, and has a priority class that is the lowest priority present in the output queue), the PG pointer of packet X is 0 guaranteed to point to the packet pleceding it in FIFO order, permitting establishment of the forward FIFO order. When trying to establish the reverse PG pointers, one must realize that by the time a packet X is vulnerable to pushout, the only packet whose PG
pointer could still be pointing to packet X is the next packet in the priority class of packet X, which can be located using the NS pointer of packet X.
Also, the PG pointer points to the previous packet of greater than or equal priority, rather than the previous packet of strictly greater priority. The reason for this is to limit the number of PG pointers that must be nulled out when the target packet of those pointers reaches the head of the FIFO order and gets served.
The required data structures for the r1i~closed selective pushout system and 2 o method are given below in categories (A)-(E).
A) For each packet:
NS = packet pointer to next packet of same priority;
PG = packet pointer to previous packet of greater or equal priority in FIFO
order; and PN = field indicating priority of next packet in FIFO order.
B) For each output queue:
QL = counter indicati~g length of output queue; and PF = field indicating which priority is first in overall FIFO order.

~15952~

C) For each output queue, for each space priority class i = l, 2, ..., P:
FP[i] = pointer to first packet of class i;
LP[i] = pointer to last packet of class i;
PLG[i] = field indicating priority of latest packet whose priority is >= i.
5 D) For the list of free locations in the shared memory:
FF = pointer to first packet of a free list; and LF = pointer to last packet of the free list.
FIGS. 7-8 illustrate an example of method of selective pushout disclosed herein using two packet pointers and a field for each packet, with each packet associated one 10 of two space priorities, where the pushing packet and pushed packet belong to the same FIFO queue. It is understood that the disclosed selective pushout system and method may be applied to a packet communications module such as an ATM switch for switching packets having two or more space priorities.
The packet communications module l0 tracks the QUEUE_LENGTH (QL) of 5 each queue, so pushout is performed on dpp~o~liate queues. As shown in FIGS. 7-8, gray shaded packets have a class l priority and non-gray packets have a class 2 priority which is of greater priority than the class l packets. For example, packet Xl is of class 2 and has an NS pointer to the next packet in the queue of the same priority; i.e. packet X3. Each output port m~int~inc an output queue using a first-in-20 first-out (FIFO) linked-list order. As shown in FIGS. 7-8, to m~int~in the priority chains, each output queue has two pointers for each priority level i = l, 2, ... P:
FIRST_PACKET_OF_CLASS_i (FP[i]) and LAST PACKET OF CLASS i (LP[i]).
Each output queue also has a PRIORITY OF FIRST PACKET (PF) field to indicate which priority is first in overall FIFO order. When a pushout is to be performed, one 2 5 of the pointers FP[i] indicates the packet to be pushed out. For example, as shown in FIG. 7 prior to the pushout, packet X2 is to be pushed out. As shown in FIG. 8, the new packet of class 2 has beén added to the queue and is stored in memory location X2, so the LP[2] pointer is updated to point to X2 and the PN field of packet Xl0 is 215952~

updated to indicate that X2 is next in FIFO order. It is understood that after the pushout, X2 has the newly queued packet which may be of a di~lent priority than the pushed-out packe~.
In order to set the PG pointer for arriving packets, each output queue needs a 5 field PLG[i], for each priority level i = l, 2, , P, to indicate the priority of the latest packet having a priority greater than or equal to priority i. Note that PLG[l] gives the priority of the last packet in overall FIFO order, analogous to PF. The implementation of the system for queueing and selective pushout and method disclosed herein mayoperate faster with only a relatively small amount of additional storage if PF and 0 PLG[i] are pre-infl~te~l, i.e. converted into the pointers FP[PF] and LP[PLG[i~] and are stored as such pointers in memory.
The amount of state information used by the disclosed selective pushout system and method is less than the amount used by the second previous implementation ofselective pushout, coll~,onding to the three packet pointer implenn~nt~tion. Thedisclosed selective pushout system and method is also faster in speed than the first previous implementation of selective pushout cGllesl,onding to the one packet pointer implementation.
The follo~,ving codes in steps (1)-(6) collc;~,uollding to the steps- 70-84 in FIG.
5 are used to illustrated the pushout of packets in the disclosed selective pushout 2 0 system and method:
l) Select the priority j and the output queue PQ for pushout:
LET i BE THE PRIORITY OF THE PACKET INSTIGATING THE
PUSHOUT.
FOR j = l TO i {
2 5 ARE 1 ~RE OUTPUT QUEUES WITH PACKET(S) OF PRIORII Y
i (FP~j] != NULL) ?
IF SO, CHOOSE LONGEST QUEUE AND BREAK

~159~2~

IF NOT, CONTINUE.
-2) Locate the particular buffer location NEWPACKET to be free:
NEWPACKET = FP [j]

The state variables mentioned below are those associated with output queue PQ.
3) Maintain the FIFO order as follows: since NEWPACKET was selected for pushout,there are no packets of priority less than j in output queue PQ, and there are no packets of priority j ahead of NEWPACKET in the FIFO order of output queue PQ.
Therefore the packet identified by the PG pointer of NEWPACKET is immediately ahead of NEWPACKET in FIFO order and adopts the PN value of NEWPACKET as its own:
IF (NEWPACKET -> PG != NULL) NEWPACKET -> PG -> PN = NEWPACKET -> PN
IF (PF = j) PF = NEWPACKET -> PN
4) If the NEXT_PACKET_OF_CLASSj had its PG pointer pointing to the pushed-out location NEWPACKET, then this PG pointer is updated to match the PG pointer of NEWPACKET. This is the only packet having its PG pointer pointing to NEWPACKET, because the priority sub-queues of PQ with priority less than j are empty.
2 0 IF ((NEWPACKET -> NS != NULL) AND (NEWPACKET -> NS -> PG ==
NEWPACKET)) NEWPACKET -> NS -> PG = NEWPACKET -> PG
5) Since the pushed-out packet is located at the head of one of the logical sub-queues, the order in the logical sub-queue is m~int~ined by updating the first packet pointer to 2 5 the new head of the sub-queue:
FP[j] = NEWPACKET -> NS
IF (FPtj] e NULL) {

21~9S2~

LP~j] = NULL
IF (j c P) SET PLG[l], PLG[2], ..., PLG~j] TO PLG~j + l]
ELSE SET PLG[l], PLG[2], ..., PLG[P] TO 0 }
6) The queue length counter ls then decremented:
QL = QL - l.

The system for queueing and selective pushout and method disclosed herein uses the associated memory 22 to store state information relating to the packets and the output queues. Generally, for ATM switches, the overwhelming majority of this 0 use of memory is required by the packet pointers belonging to the individual packets.
The memory space required for fields of packets is negligible, and ~e queue poinle~
and counters dedicated to individual output queues conC~lme a relatively small amount of memory. However, the packet pointers can concl~me a relatively large amount of memory if the number of packets in the memory is large.
The present implementation of the system for queueing and selective pushout and method using only two packet pointers and one field per packet reduces the nurnber of memory-consuming pointers. For an ATM switch capable of queu~ing a great number of packets, the memory savings of the present selective pushout system and method may be substantial.
2 o The following exemplary codes are used to illustrate the departure and arrival of packets in the disclosed system for quç~lPing and selective pushout and method.
DEPARTIJRE OF PACKETS
The departure of packets is illustrated in code in steps (1)-(4~ below corresponding to step 86-l00 in FIG. 6.
l) Verify that the queue has a packet to serve:
IF (FP - NU~L) THEN EXIT.
2) Serve packet X of space priority PF at the head of the queue:

5 2 ~

` - 18 -X = FP [PF];
TRANSMIT PACKET AT MEMORY LOCATION X OVER A GIVEN
OUTPUT PORT.
3) All the packets whose PG pointers point to the packet being served have these PG
5 pointers nulled out. If there are any such packets, they can only be located at the heads of the queues with prionty less than PF or just after the packet being served in the logical queue for class PF.
FORj = l TOPF-l {
IF ((FP[j] != NULL) AND (FP~j] -> PG = X)) FP~] -> PG= NULL
}

IF ((X -> NS != NULL) AND (X -> NS -> PG = X)) X -> NS -> PG= NULL;
5) The queue length counter, the FP[PF] pointer, and the PF field are updated:
QL= QL - l FP [PF] = X -~ NS
IF (FP[PF] == NULL {
LP [PF] = NULL
FOR (INITIALIZE j = PF; W~LE ((j > 0) AND (PLG~]==PF));
DECR j) PLG~j] = 0 }

PF = X -> PN.
4) Then add the freed memory location to tail of free list:
IF (FF = NULL) FF = X
ELSE LF -> NP = X
LF = X.

21~9S2~

ARRIVAL OF PACKETS
The arrival of packets is illustrated in code in steps (1)-(4) below corresponding to the steps of FIG. 4 1) Store new packet of priority i in memory location NEWPACKET, either fetched 5 from the free list or determined by pushout:
IF (FF = NULL) DO A PUSHOUT TO GET LOCATION OF NEWPACKET
ELSE {
NEWPACKET = FF
FF = FF -> NC
IF (FF -- NULL) LF = NULL
}

READ THE CONTENTS OF THE ARRrVING PACKET INTO MEMORY
LOCATION NEWPACKET
2) Add a newly arrived packet of Class i:
IF (PLG[i] = 0) NEWPACKET -> PG= NULL
ELSE NEWPACKET -> PG= LP[PLG[i]] I F
(PLG[1)--0) PF = i ELSE LP[PLG[1]] -> PN = i 2 0 IF (FP[i]--NULL) FP[i] = NEWPACKET
ELSE LP[i] -> NS = NEWPACKET
LP[i] = NEWPACKET, 3) The PLG array is updated:
PLG~j] = i FOR j = 1, 2, ..., i 25 4) The queue length counter is then incremented:
QL = QL + 1.
FINDING THE LONGEST OUEUE FOR SELECTIVE PUSHOUT
By m~int~inin~ a state information table 30, the search for the longest 21S9~28 approp~late queue described above for step 74 in FIG. 5 can be performed more quickly without greatly increasing the size of the ~csoci~tp~d memory 22 ` . FIG. 9 illustrates an exemplary state information table 30, used by the system for queveing and selective pushout and method and stored in the associated memory 5 22, which contains identifying information for the queues and information about the presence of packets of each priority level in the queues. In an altemative embodiment, the system for queueing and selective pushout and method disclosed herein may equivalently use a hierarchical list, a set of arrays, a matrix, or equivalent data structures to store the state information table 30.
Referring to FIG. 9, each row except the bottom one col,~onds to one output port of the shared memory. The rows are sorted by output port queue length, with the longest queue at the top. Column 0 of row q gives the identification (ID) number of the q~-longest output queue. For j = 1, ..., P, column j of row q csnt~inc one bit indicating whether there are any packets of priority j ~ ly present in the q~-longest 15 output queue. The entries along each column j = l, ..., P are logically OR'd together to produce the entry in the bottom row; this entry indicates whether there are any packets of priority j currently present anywhere in the shared memory.
When a packet of priority i is to be stored by a pushout of another packet, the bottom row of the state information table 30 is read from left to right to find the 2 0 lowest priority j <= i for which packets are present somewhere in the shared memory.
Then column j is read from top to bottom to find the longest queue with packet(s) of priority j. This queue is thus determined from column 0 to be the queue from which a packet should be pushed.
In performing selective pushout, the ordering of the rows by queue length is not25 required to be absolutely up-to-date. A background process may be used to manage the time-concnming tasks of re-ordering the queues; this process may regularly rearrange the rows of the state information table 30 as npce~ y The contents of each row of the state information table 30 are regularly updated 215952~

to be current, so that pushouts are not a~ ed from queues that no longer have any packets of the desired priority. Therefore, in the exemplary embodiment, the relevant bit in the state information table 30 is updated after each packet arrival, service, and pushout For performing the updating of the state information table 30, each output 5 queue is associated with a field indicating its current ranking in the queue length ordering, i.e. its row in the state information table 30. The use of such a field allows the output queue to maintain its row of bits in the state information table 30 as packets join the queue, are served by the queue, or are pushed out from the queue.
While the disclosed implementation of the system for que-~eing and selective 0 pushout and method has been particularly shown and described with reference to the prefelled embodiments, it will be understood by those skilled in the art that various modifications in form and detail may be made therein without departing from the scope and spirit of the invention. Accordingly, modifications such as those suggested above, but not limited thereto, are to be considered within the scope of the invention.

Claims (17)

1. A system for queueing and selective pushout of packets comprising:
a packet processing circuit for processing packets as incoming and outgoing packets, for receiving an incoming packet on an input port and for transmitting an outgoing packet on an output port, with the packets belonging to at least one space priority class;
the shared memory configured for storing the processed packets in at least one queue, and within the at least one queue, storing the packets in a plurality of priority sub-queues corresponding to the space priorities of the packets;
each packet associated with at most a first and a second pointer and at most a field, the first pointer for pointing to the next packet of the same priority in the at least one queue, the second pointer for pointing to a previous packet of greater or equal space priority in the FIFO order in the at least one queue, and the field for storing the priority of the next packet in the FIFO order in the at least one queue; and the processor configured for storing received packets in the at least one queue, transmitting packets from the at least one queue and pushing packets out of the at least one queue, the received packets are stored in priority sub-queues within the at least one queue using the first pointers associated with the packets, and a FIFO order is maintained among all the packets in the queue using the fields and the secondpointers of the packets, after pushout, the FIFO order is restored using the first and second pointers and the fields associated with the packets.
2. The system of claim 1 wherein the packet processing circuit processes packets of variable lengths.
3. The system of claim 1 wherein the packet processing circuit processes packets of fixed lengths.
4. The system of claim 1 wherein the packets are asynchronous transfer mode (ATM) cells.
5. The system of claim 1 wherein the packet processing circuit is connected to a plurality of input ports; and the processor controls the packet processing circuit to perform packet multiplexing functions.
6. The system of claim 1 wherein the packet processing circuit is connected to a plurality of output ports; and the processor controls the packet processing circuit to perform packet demultiplexing functions.
7. The system of claim 1 wherein the packet processing circuit is connected to a plurality of input ports and a plurality of output ports; and the processor controls the packet processing circuit to perform switching functions.
8. The system of claim 1 wherein the queue of the incoming packet includes packets destined for the same output port as the incoming packet and belonging to at least one space priority class.
9. The system of claim 1 wherein the queue of the incoming packet includes packets destined for the same input port as the incoming packet and belonging to at least one space priority class.
10. The system of claim 1 wherein the processor determines a longest queue from the at least one output queue and performs selective pushout on the longestqueue.
11. The system of claim 1 wherein the shared memory stores the queue-length of the at least one queue and a state information table including identifying information for the at least one queue and the priority of packets contained therein; and the processor uses the queue-length of the at least one queue and state information table to determine the longest queue containing a packet of the lowest priority lower than or equal to that of the incoming packet.
12. A method for queueing and selective pushout of packets in the shared memory of a packet communications module, the method comprising the steps of:

storing packets in the at least one queue, and within the at least one queue, storing the packets in a plurality of priority sub-queues corresponding to the space priorities of the packets, where the at least one queue and the sub-queues have some data structures associated with them and the packets are associated with a first pointer for pointing to the next packet of the same priority in the at least one queue, a second pointer for pointing to a previous packet of greater or equal space priority in the FIFO order in the at least one queue, and a field for storing the priority of the next packet in the FIFO order in the at least one queue;
transmitting packets from the at least one output queue;
pushing out packets from the at least one output queue; and maintaining queue-lengths and a state information table.
13. The method of claim 12 wherein the data structures associated with the at least one queue are a queue-length counter and a field indicating which priority is first in the FIFO order, and the data structures associated with each priority sub-queue are a first and a second pointer pointing to the first and last packets respectively in the sub-queue, and a field indicating the priority of the previous packet in the FIFO order whose priority is greater than or equal to that of the sub-queue under consideration.
14. The method of claim 12 wherein the step of storing packets in at the least one queue includes the steps of:
receiving an incoming packet having an associated space priority;
using selective pushout to create space if needed;
storing the packet in the priority sub-queue corresponding to the space priority of the incoming packet using the first pointer;
storing in the field associated with the packet the priority value corresponding to the next packet in FIFO order;
maintaining a FIFO order among the packets queued in the output queue using the field and the second pointer; and incrementing the queue-length counter and updating the other data structures associated with the at least one queue and the priority sub-queues contained therein.
15. The method of claim 12 wherein the step of transmitting includes the steps of:
reading out the first packet in FIFO order of the at least one output queue;
updating the second pointers pointing to the packet being transmitted;
and decrementing the queue-length counter and updating the other data structures associated with the at least one queue and the priority sub-queues contained therein.
16. The method of claim 12 wherein the step of pushing out a packet includes the steps of:
determining a longest queue of the at least one output queue containing a packet of the lowest priority lower than or equal to that of the incoming packet;
pushing out the first packet from the corresponding priority sub-queue within the longest queue;
decrementing the queue-length counter and updating the other data structures associated with the at least one queue and the priority sub-queues contained therein; and restoring the FIFO order of the at least one output queue using the first and second pointers and the fields of the queued packets.
17. The method of claim 12 including the step of maintaining the queue-length of the at least one output queue and a state information table including identifying information for the at least one output queue and the priority levels of the packets contained therein; and the step of determining a longest queue containing a packet of the lowest priority lower than or equal to that of the incoming packet using the state information table.
CA002159528A 1994-12-02 1995-09-29 Implementation of selective pushout for space priorities in a shared memory asynchronous transfer mode switch Abandoned CA2159528A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US348,530 1994-12-02
US08/348,530 US5521916A (en) 1994-12-02 1994-12-02 Implementation of selective pushout for space priorities in a shared memory asynchronous transfer mode switch

Publications (1)

Publication Number Publication Date
CA2159528A1 true CA2159528A1 (en) 1996-06-03

Family

ID=23368434

Family Applications (1)

Application Number Title Priority Date Filing Date
CA002159528A Abandoned CA2159528A1 (en) 1994-12-02 1995-09-29 Implementation of selective pushout for space priorities in a shared memory asynchronous transfer mode switch

Country Status (6)

Country Link
US (1) US5521916A (en)
EP (1) EP0715436A3 (en)
JP (1) JPH08298522A (en)
KR (1) KR960027837A (en)
CA (1) CA2159528A1 (en)
MX (1) MX9504950A (en)

Families Citing this family (86)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5446726A (en) * 1993-10-20 1995-08-29 Lsi Logic Corporation Error detection and correction apparatus for an asynchronous transfer mode (ATM) network device
AUPM699394A0 (en) * 1994-07-25 1994-08-18 Curtin University Of Technology Link level controlled access to available asynchronous network service
US5949781A (en) * 1994-08-31 1999-09-07 Brooktree Corporation Controller for ATM segmentation and reassembly
AU6712496A (en) * 1995-07-19 1997-02-18 Fujitsu Limited Point-to-multipoint arbitration
EP0845181A4 (en) * 1995-07-19 2001-07-18 Fujitsu Network Communications Link scheduling
US6002667A (en) 1995-07-19 1999-12-14 Fujitsu Network Communications, Inc. Minimum guaranteed cell rate method and apparatus
JPH11512583A (en) 1995-09-14 1999-10-26 フジツウ ネットワーク コミュニケーションズ,インコーポレイテッド Transmitter-controlled flow control for buffer allocation in a wide area ATM network
JP3545110B2 (en) * 1995-09-26 2004-07-21 富士通株式会社 Communication service quality control method
US5757771A (en) * 1995-11-14 1998-05-26 Yurie Systems, Inc. Queue management to serve variable and constant bit rate traffic at multiple quality of service levels in a ATM switch
AU1697697A (en) 1996-01-16 1997-08-11 Fujitsu Limited A reliable and flexible multicast mechanism for atm networks
US5724358A (en) * 1996-02-23 1998-03-03 Zeitnet, Inc. High speed packet-switched digital switch and method
US5841772A (en) * 1996-03-07 1998-11-24 Lsi Logic Corporation ATM communication system interconnect/termination unit
US6535512B1 (en) 1996-03-07 2003-03-18 Lsi Logic Corporation ATM communication system interconnect/termination unit
US5982749A (en) * 1996-03-07 1999-11-09 Lsi Logic Corporation ATM communication system interconnect/termination unit
US5920561A (en) * 1996-03-07 1999-07-06 Lsi Logic Corporation ATM communication system interconnect/termination unit
US5848068A (en) * 1996-03-07 1998-12-08 Lsi Logic Corporation ATM communication system interconnect/termination unit
US5841973A (en) * 1996-03-13 1998-11-24 Cray Research, Inc. Messaging in distributed memory multiprocessing system having shell circuitry for atomic control of message storage queue's tail pointer structure in local memory
US5872938A (en) * 1996-06-28 1999-02-16 International Business Machines Corp. Service priority queue implemented with ordered sub-queues and sub-queue pointers pointing to last entries in respective sub-queues
US5748905A (en) 1996-08-30 1998-05-05 Fujitsu Network Communications, Inc. Frame classification using classification keys
US6473404B1 (en) * 1998-11-24 2002-10-29 Connect One, Inc. Multi-protocol telecommunications routing optimization
US6016307A (en) 1996-10-31 2000-01-18 Connect One, Inc. Multi-protocol telecommunications routing optimization
FI103310B (en) * 1996-11-15 1999-05-31 Nokia Telecommunications Oy Implementation of buffering in a packet switching data communication network
GB9626752D0 (en) * 1996-12-23 1997-02-12 Northern Telecom Ltd Management of data structures
US6061358A (en) * 1997-02-13 2000-05-09 Mcdata Corporation Data communication system utilizing a scalable, non-blocking, high bandwidth central memory controller and method
DE19705789A1 (en) * 1997-02-14 1998-09-03 Siemens Ag Method and circuit arrangement for transmitting message cells in the course of virtual connections of different priorities
SE9700714L (en) * 1997-02-27 1998-08-28 Ericsson Telefon Ab L M Method of controlling the data packet flow in a switch unit and a switch unit adapted to operate according to the method
KR100236036B1 (en) * 1997-03-31 1999-12-15 전주범 Method of discarding atm cells in an atm nic
US6128295A (en) * 1997-07-11 2000-10-03 Telefonaktiebolaget Lm Ericsson Buffering of point-to-point and/or point-to-multipoint ATM cells
US6018515A (en) * 1997-08-19 2000-01-25 Ericsson Messaging Systems Inc. Message buffering for prioritized message transmission and congestion management
US6324165B1 (en) 1997-09-05 2001-11-27 Nec Usa, Inc. Large capacity, multiclass core ATM switch architecture
US6408005B1 (en) 1997-09-05 2002-06-18 Nec Usa, Inc. Dynamic rate control scheduler for ATM networks
US6198724B1 (en) 1997-10-02 2001-03-06 Vertex Networks, Inc. ATM cell scheduling method and apparatus
TW494302B (en) * 1998-02-24 2002-07-11 Accton Technology Corp Quasi-pushout packet discarding method of packet communication system and its shared buffer
US6167041A (en) * 1998-03-17 2000-12-26 Afanador; J. Abraham Switch with flexible link list manager for handling ATM and STM traffic
US6563836B1 (en) 1998-03-19 2003-05-13 International Business Machines Corporation Algorithm for dynamic prioritization in a queuing environment
TW463092B (en) 1998-03-20 2001-11-11 Accton Technology Corp Quasi-pushout method with upper layer packet discarding discipline and packet communication system implementing the method
US6590901B1 (en) * 1998-04-01 2003-07-08 Mosaid Technologies, Inc. Method and apparatus for providing a packet buffer random access memory
US6445680B1 (en) * 1998-05-27 2002-09-03 3Com Corporation Linked list based least recently used arbiter
US6003101A (en) * 1998-07-15 1999-12-14 International Business Machines Corp. Efficient priority queue
US6216174B1 (en) 1998-09-29 2001-04-10 Silicon Graphics, Inc. System and method for fast barrier synchronization
US6490629B1 (en) * 1998-10-13 2002-12-03 Genuity Inc. System and method for scheduling the transmission of packet objects having quality of service requirements
US6304936B1 (en) * 1998-10-30 2001-10-16 Hewlett-Packard Company One-to-many bus bridge using independently and simultaneously selectable logical FIFOS
US6269413B1 (en) * 1998-10-30 2001-07-31 Hewlett Packard Company System with multiple dynamically-sized logical FIFOs sharing single memory and with read/write pointers independently selectable and simultaneously responsive to respective read/write FIFO selections
US6229813B1 (en) * 1998-11-25 2001-05-08 Alcatel Canada Inc. Pointer system for queue size control in a multi-task processing application
US6345324B1 (en) * 1999-02-19 2002-02-05 International Business Machines Corporation Apparatus for transferring data using an interface element and a queued direct input-output device
US6401145B1 (en) * 1999-02-19 2002-06-04 International Business Machines Corporation Method of transferring data using an interface element and a queued direct input-output device
US6246682B1 (en) * 1999-03-05 2001-06-12 Transwitch Corp. Method and apparatus for managing multiple ATM cell queues
US6633575B1 (en) * 1999-04-07 2003-10-14 Nokia Corporation Method and apparatus for avoiding packet reordering in multiple-class, multiple-priority networks using a queue
JP2001103058A (en) * 1999-09-27 2001-04-13 Toshiba Corp Device and method for controlling queue
KR100357628B1 (en) * 1999-12-28 2002-10-25 삼성전자 주식회사 Cell processing apparatus having priority queue and method thereof in an asynchronous transfer mode system
US7649901B2 (en) * 2000-02-08 2010-01-19 Mips Technologies, Inc. Method and apparatus for optimizing selection of available contexts for packet processing in multi-stream packet processing
US7155516B2 (en) 2000-02-08 2006-12-26 Mips Technologies, Inc. Method and apparatus for overflowing data packets to a software-controlled memory when they do not fit into a hardware-controlled memory
US7058065B2 (en) * 2000-02-08 2006-06-06 Mips Tech Inc Method and apparatus for preventing undesirable packet download with pending read/write operations in data packet processing
US7502876B1 (en) 2000-06-23 2009-03-10 Mips Technologies, Inc. Background memory manager that determines if data structures fits in memory with memory state transactions map
US20010052053A1 (en) * 2000-02-08 2001-12-13 Mario Nemirovsky Stream processing unit for a multi-streaming processor
US7058064B2 (en) * 2000-02-08 2006-06-06 Mips Technologies, Inc. Queueing system for processors in packet routing operations
US7139901B2 (en) * 2000-02-08 2006-11-21 Mips Technologies, Inc. Extended instruction set for packet processing applications
US7165257B2 (en) * 2000-02-08 2007-01-16 Mips Technologies, Inc. Context selection and activation mechanism for activating one of a group of inactive contexts in a processor core for servicing interrupts
US7065096B2 (en) 2000-06-23 2006-06-20 Mips Technologies, Inc. Method for allocating memory space for limited packet head and/or tail growth
US7032226B1 (en) * 2000-06-30 2006-04-18 Mips Technologies, Inc. Methods and apparatus for managing a buffer of events in the background
US7076630B2 (en) * 2000-02-08 2006-07-11 Mips Tech Inc Method and apparatus for allocating and de-allocating consecutive blocks of memory in background memo management
US7042887B2 (en) 2000-02-08 2006-05-09 Mips Technologies, Inc. Method and apparatus for non-speculative pre-fetch operation in data packet processing
US7082552B2 (en) 2000-02-08 2006-07-25 Mips Tech Inc Functional validation of a packet management unit
DE10110578B4 (en) * 2000-03-29 2004-06-03 International Business Machines Corporation Hierarchical priority filter with integrated serialization
US20020107974A1 (en) * 2000-10-06 2002-08-08 Janoska Mark William Data traffic manager
US6963571B1 (en) * 2001-03-05 2005-11-08 Advanced Micro Devices, Inc. Systems and methods for merging packet lookup results
FR2822319B1 (en) 2001-03-16 2003-05-30 Thomson Csf VARIABLE SIZE INFORMATION FRAME SWITCH FOR ON-BOARD SECURITY NETWORKS
JP2002281080A (en) * 2001-03-19 2002-09-27 Fujitsu Ltd Packet switch device and multicast transmitting method
US7509671B1 (en) * 2001-06-20 2009-03-24 Microstrategy Incorporated Systems and methods for assigning priority to jobs in a reporting system
KR20030027968A (en) * 2001-09-18 2003-04-08 (주) 이노텔리텍 Scheduling Method for Distribution of Wireless Resources of Local Wireless Communication System
US20030056073A1 (en) * 2001-09-18 2003-03-20 Terachip, Inc. Queue management method and system for a shared memory switch
US7349419B1 (en) * 2002-02-08 2008-03-25 Nortel Networks Limited Queue sizing for packet routing
US6987761B2 (en) 2002-02-13 2006-01-17 International Business Machines Corporation Inbound data stream controller with pre-recognition of frame sequence
KR100547891B1 (en) * 2003-02-08 2006-01-31 삼성전자주식회사 Data input / output device and method according to priority in asynchronous transmission mode switching system
US7689738B1 (en) 2003-10-01 2010-03-30 Advanced Micro Devices, Inc. Peripheral devices and methods for transferring incoming data status entries from a peripheral to a host
US7826614B1 (en) 2003-11-05 2010-11-02 Globalfoundries Inc. Methods and apparatus for passing initialization vector information from software to hardware to perform IPsec encryption operation
US7533154B1 (en) * 2004-02-04 2009-05-12 Advanced Micro Devices, Inc. Descriptor management systems and methods for transferring data of multiple priorities between a host and a network
JP2006301894A (en) * 2005-04-20 2006-11-02 Nec Electronics Corp Multiprocessor system and message transfer method for multiprocessor system
JP2006309512A (en) * 2005-04-28 2006-11-09 Nec Electronics Corp Multiprocessor system, and message transmission method for multiprocessor system
GB0606367D0 (en) * 2006-03-30 2006-05-10 Vodafone Plc Telecommunications networks
JP4688038B2 (en) * 2006-05-30 2011-05-25 日本電気株式会社 Wireless packet communication device, communication system, and communication method
DE102007034754A1 (en) * 2007-07-25 2009-01-29 Rohde & Schwarz Gmbh & Co. Kg Device and method for increasing the data throughput in radio networks
US8358687B2 (en) * 2007-09-07 2013-01-22 Qualcomm Incorporated Method and apparatus for receiving multiple simultaneous stream bursts with limited DVB receiver memory
CN102347891A (en) * 2010-08-06 2012-02-08 高通创锐讯通讯科技(上海)有限公司 Method for using shared cache
US8612649B2 (en) 2010-12-17 2013-12-17 At&T Intellectual Property I, L.P. Validation of priority queue processing
US9083563B2 (en) * 2012-06-29 2015-07-14 Avaya, Inc. Method for reducing processing latency in a multi-thread packet processor with at least one re-order queue

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5193090A (en) * 1988-07-15 1993-03-09 Janusz Filipiak Access protection and priority control in distributed queueing
JPH02117243A (en) * 1988-10-27 1990-05-01 Toshiba Corp Packet communication equipment
JP2860661B2 (en) * 1989-03-14 1999-02-24 国際電信電話 株式会社 ATM switch
US5072443A (en) * 1989-07-28 1991-12-10 At&T Bell Laboratories Communications system
US5014265A (en) * 1989-11-30 1991-05-07 At&T Bell Laboratories Method and apparatus for congestion control in a data network
US5163046A (en) * 1989-11-30 1992-11-10 At&T Bell Laboratories Dynamic window sizing in a data network
US5272697A (en) * 1990-02-23 1993-12-21 At&T Bell Laboratories Apparatus and method for time multiplexing a resource among a plurality of entities
US5136584A (en) * 1990-07-11 1992-08-04 At&T Bell Laboratories Hardware interface to a high-speed multiplexed link
US5115430A (en) * 1990-09-24 1992-05-19 At&T Bell Laboratories Fair access of multi-priority traffic to distributed-queue dual-bus networks
DE69130853T2 (en) * 1990-11-21 1999-07-22 At & T Corp Bandwidth management and congestion protection for access to broadband ISDN networks
US5278969A (en) * 1991-08-02 1994-01-11 At&T Bell Laboratories Queue-length monitoring arrangement for detecting consistency between duplicate memories
US5233606A (en) * 1991-08-02 1993-08-03 At&T Bell Laboratories Arrangement for controlling shared-buffer-memory overflow in a multi-priority environment
US5285441A (en) * 1992-03-17 1994-02-08 At&T Bell Laboratories Errorless line protection switching in asynchronous transer mode (ATM) communications systems
US5289303A (en) * 1992-09-30 1994-02-22 At&T Bell Laboratories Chuted, optical packet distribution network

Also Published As

Publication number Publication date
EP0715436A2 (en) 1996-06-05
JPH08298522A (en) 1996-11-12
EP0715436A3 (en) 1997-10-22
US5521916A (en) 1996-05-28
KR960027837A (en) 1996-07-22
MX9504950A (en) 1997-01-31

Similar Documents

Publication Publication Date Title
US5521916A (en) Implementation of selective pushout for space priorities in a shared memory asynchronous transfer mode switch
AU632840B2 (en) An atm switching element for transmitting variable length cells
EP0471344B1 (en) Traffic shaping method and circuit
US4926416A (en) Method and facilities for hybrid packet switching
JP2907886B2 (en) Switching system
EP0702500B1 (en) Method of multicasting and multicast system
CA2297650C (en) Networking systems
JP2618327B2 (en) Wideband input buffer ATM switch
US7773602B2 (en) CAM based system and method for re-sequencing data packets
EP1013132B1 (en) Switch arrangement
US5402415A (en) Multicast virtual circuit switch using cell recycling
EP0785699B1 (en) Multicast routing in multistage networks
US6542502B1 (en) Multicasting using a wormhole routing switching element
JP2788577B2 (en) Frame conversion method and apparatus
US6791992B1 (en) Earliest-deadline-first queuing cell switching architecture and method
KR19990087752A (en) Efficient Output Request Packet Switch and Method
JPH1070556A (en) Atm switch
JP2000349789A (en) Tdm switch system very wide in memory width
EP0612171B1 (en) Data queueing apparatus and ATM cell switch based on shifting and searching
US6445680B1 (en) Linked list based least recently used arbiter
CA2151180C (en) Method and apparatus for multicast of atm cells
EP1377106A2 (en) Method and apparatus for burst scheduling
EP0710047B1 (en) Method and apparatus for multicast of ATM cells
JP2002368785A (en) Scheduling device and cell communication unit
Choudhury et al. New implementation of multi-priority pushout for shared memory ATM switches

Legal Events

Date Code Title Description
EEER Examination request
FZDE Discontinued