REGISTER  |  SIGN IN
United States Patent5748905
Hauser , ; et al.May 5, 1998
Title
Frame classification using classification keys
Abstract
An improved method and apparatus for recognizing, classifying and processing frames received at a frame processor in a computer network is disclosed. Following receipt of a frame at an input port of a frame processor, source and destination addresses are parsed from the frame. A plurality of lookup tables are provided in a memory, each of which contains a search field and a classification key field. Source or destination addresses are stored in the respective search fields along with other information associated with the frame and a compact classification key is stored in the corresponding classification key field. Searches are performed of the respective search fields within the respective lookup tables to determine whether a match exists between the each of the destination and source addresses and other information and the search field within the lookup tables. In the event the searches yield a match, a classification key associated with each respective address is retrieved. Classification keys corresponding to addresses thus retrieved are concatenated and the concatenated value is employed as the basis for a further search of the search field of another one of the lookup tables to identify a classification key which is employed to specify the manner of further processing of the received frame.
Inventors:Hauser; Stephen A. (Burlington, MA), Williams; Jeffrey L.  (Hopkinton, MA)
Assignee:Fujitsu Network Communications, Inc. (Richardson, TX)
Fujitsu Limited (Kawasaki, JP)
Appl. No.:701335
Filed:August 30, 1996
Current U.S. Class:709/249 
Current International Class:H04L 29/06 (20060101)
Field of Search:364/DIG.1,DIG.2 395/200.3,200.5,200.53,200.59,200.66,200.68,200.76,200.79,200.8,280,285,286
U.S. Patent Documents
3804991April 1974Hammond et al.
3974343August 1976Cheney et al.
4069399January 1978Barrett et al.
4603382July 1986Cole et al.
4715030December 1987Koch et al.
4727537February 1988Nichols
4737953April 1988Koch et al.
4797881January 1989Ben-Artzi
4821034April 1989Anderson et al.
4837761June 1989Isono et al.
4849968July 1989Turner
4870641September 1989Pattavina
4872159October 1989Hemmady et al.
4872160October 1989Hemmady et al.
4878216October 1989Yunoki
4893302January 1990Hemmady et al.
4893307January 1990McKay et al.
4894824January 1990Hemmady et al.
4897841January 1990Gang, Jr.
4899333February 1990Roediger
4920531April 1990Isono et al.
4922503May 1990Leone
4933938June 1990Sheehy
4947390August 1990Sheehy
4953157August 1990Franklin et al.
4956839September 1990Torii et al.
4958341September 1990Hemmady et al.
4979100December 1990Makris et al.
4993018February 1991Hajikano et al.
5021949June 1991Morten et al.
5029164July 1991Goldstein et al.
5060228October 1991Tsutsui et al.
5067123November 1991Hyodo et al.
5070498December 1991Kakuma et al.
5083269January 1992Syobatake et al.
5084867January 1992Tachibana et al.
5084871January 1992Carn et al.
5090011February 1992Fukuta et al.
5090024February 1992Vander May et al.
5093912March 1992Dong et al.
5115429May 1992Hluchyj
5119369June 1992Tanabe et al.
5119372June 1992Verbeek
5128932July 1992Li
5130975July 1992Akata
5130982July 1992Ash et al.
5132966July 1992Hayano et al.
5146474September 1992Nagler et al.
5146560September 1992Goldberg et al.
5150358September 1992Punj et al.
5151897September 1992Suzuki
5157657October 1992Potter et al.
5163045November 1992Caram et al.
5163046November 1992Hahne et al.
5179556January 1993Turner
5179558January 1993Thacker et al.
5185743February 1993Murayama et al.
5191582March 1993Upp
5191652March 1993Dias et al.
5193151March 1993Jain
5197067March 1993Fujimoto et al.
5198808March 1993Kudo
5199027March 1993Barri
5239539August 1993Uchida et al.
5253247October 1993Hirose et al.
5253248October 1993Dravida et al.
5255264October 1993Cotton et al.
5255266October 1993Watanabe et al.
5257311October 1993Naito et al.
5258979November 1993Oomuro et al.
5265088November 1993Takigawa et al.
5267232November 1993Katsube et al.
5268897December 1993Komine et al.
5271010December 1993Miyake et al.
5272697December 1993Fraser et al.
5274641December 1993Shobatake et al.
5274768December 1993Traw et al.
5280469January 1994Taniguchi et al.
5280470January 1994Buhrke et al.
5282201January 1994Frank et al.
5283788February 1994Morita et al.
5285446February 1994Yonehara
5287349February 1994Hyodo et al.
5287535February 1994Sakagawa et al.
5289462February 1994Ahmadi et al.
5289463February 1994Mobasser
5289470February 1994Chang et al.
5291481March 1994Doshi et al.
5291482March 1994McHarg et al.
5295134March 1994Yoshimura et al.
5301055April 1994Bagchi et al.
5301184April 1994Uriu et al.
5301190April 1994Tsukuda et al.
5301193April 1994Toyofuku et al.
5303232April 1994Faulk, Jr.
5305311April 1994Lyles
5309431May 1994Tominaga et al.
5309438May 1994Nakajima
5311586May 1994Bogart et al.
5313454May 1994Bustini et al.
5313458May 1994Suzuki
5315586May 1994Charvillat
5319638June 1994Lin
5321695June 1994Proctor et al.
5323389June 1994Bitz et al.
5333131July 1994Tanabe et al.
5333134July 1994Ishibashi et al.
5335222August 1994Kamoi et al.
5335325August 1994Frank et al.
5339310August 1994Taniguchi
5339317August 1994Tanaka et al.
5339318August 1994Tanaka et al.
5341366August 1994Soumiya et al.
5341373August 1994Ishibashi et al.
5341376August 1994Yamashita
5345229September 1994Olnowich et al.
5350906September 1994Brody et al.
5355372October 1994Sengupta et al.
5357506October 1994Sugawara
5357507October 1994Hughes et al.
5357508October 1994Le Boudec et al.
5357510October 1994Norizuki et al.
5359600October 1994Ueda et al.
5361251November 1994Aihara et al.
5361372November 1994Rege et al.
5363433November 1994Isono
5365514November 1994Hershey et al.
5371893December 1994Price et al.
5373504December 1994Tanaka et al.
5375117December 1994Morita et al.
5377262December 1994Bales et al.
5377327December 1994Jain et al.
5379397January 1995Glover et al.
5379418January 1995Shimazaki et al.
5390170February 1995Sawant et al.
5390174February 1995Jugel
5390175February 1995Hiller et al.
5392280February 1995Zheng
5392402February 1995Robrock, II
5394396February 1995Yoshimura et al.
5398235March 1995Tsuzuki et al.
5400337March 1995Munter
5402415March 1995Turner
5412648May 1995Fan
5414703May 1995Sakaue et al.
5420858May 1995Marshall et al.
5420988May 1995Elliott
5422879June 1995Parsons et al.
5425021June 1995Derby et al.
5425026June 1995Mori
5432713July 1995Takeo et al.
5432784July 1995Ozveren
5432785July 1995Ahmed et al.
5432908July 1995Heddes et al.
5436886July 1995McGill
5436893July 1995Barnett
5440547August 1995Easki et al.
5444702August 1995Burnett et al.
5446733August 1995Tsuroka
5446737August 1995Cidon et al.
5446738August 1995Kim et al.
5448559September 1995Hayter et al.
5450406September 1995Esaki et al.
5452296September 1995Shimizu
5455820October 1995Yamada
5455825October 1995Lauer et al.
5457687October 1995Newman
5459743October 1995Fukuda et al.
5461611October 1995Drake, Jr. et al.
5465331November 1995Yang et al.
5475679December 1995Munter
5479401December 1995Bitz et al.
5479402December 1995Hata et al.
5483526January 1996Ben-Nun et al.
5485453January 1996Wahlman et al.
5485455January 1996Dobbins et al.
5487063January 1996Kakuma et al.
5488606January 1996Kakuma et al.
5491691February 1996Shtayer et al.
5491694February 1996Oliver et al.
5493566February 1996Ljungberg et al.
5497369March 1996Wainwright
5499238March 1996Shon
5504741April 1996Yamanaka et al.
5504742April 1996Kakuma et al.
5506834April 1996Sekihata et al.
5506839April 1996Hatta
5506956April 1996Cohen
5509001April 1996Tachibana et al.
5509007April 1996Takashima et al.
5513134April 1996Cooperman et al.
5513178April 1996Tanaka
5513180April 1996Miyake et al.
5515359May 1996Zheng
5519690May 1996Suzuka et al.
5521905May 1996Oda et al.
5521915May 1996Dieudonne et al.
5521916May 1996Choudhury et al.
5521917May 1996Watanabe et al.
5521923May 1996Willmann et al.
5523999June 1996Takano et al.
5524113June 1996Gaddis
5526344June 1996Diaz et al.
5528588June 1996Bennett et al.
5528590June 1996Iidaka et al.
5528591June 1996Lauer
5530695June 1996Digne et al.
5533009July 1996Chen
5533020July 1996Byrn et al.
5535196July 1996Aihara et al.
5535197July 1996Cotton
5537394July 1996Abe et al.
5541912July 1996Choudhury et al.
5544168August 1996Jeffrey et al.
5544169August 1996Norizuki et al.
5544170August 1996Kasahara
5546389August 1996Wippenbeck et al.
5546391August 1996Hochschild et al.
5546392August 1996Boal et al.
5550821August 1996Akiyoshi
5553057September 1996Nakayama
5553068September 1996Aso et al.
5555243September 1996Kakuma et al.
5555265September 1996Kakuma et al.
5557607September 1996Holden
5568479October 1996Watanabe et al.
5570361October 1996Norizuki et al.
5570362October 1996Nishimura
5572522November 1996Calamvokis et al.
5577032November 1996Sone et al.
5577035November 1996Hayter et al.
5583857December 1996Soumiya et al.
5583858December 1996Hanaoka
5583861December 1996Holden
5590132December 1996Ishibashi et al.
5602829February 1997Nie et al.
5610913March 1997Tomonaga et al.
5623405April 1997Isono
5625846April 1997Kobayakawa et al.
Foreign Patent Documents
484943Mar., 1992JP
Other References
An Ascom Timeplex White Paper, Meeting Critical Requirements with Scalable Enterprise Networking Solutions Based on a Unified ATM Foundation, pp. 1-12, Apr. 1994. .
Douglas H. Hunt, ATM Traffic Management--Another Perspective, Business Communications Review, Jul. 1994. .
Richard Bubenik et al., Leaf Initiated Join Extensions, Technical Committee, Signalling Subworking Group, ATM Forum/94-0325R1, Jul. 1, 1994. .
Douglas H. Hunt et al., Flow Controlled Virtual Connections Proposal for ATM Traffic Management (Revision R2), Traffic Management Subworking Group, ATM.sub.- Forum/94-0632R2, Aug. 1994. .
Flavio Bonomi et al., The Rate-Based Flow Control Framework for the Available Bit Rate ATM Service, IEEE Network, Mar./Apr. 1995, pp. 25-39. .
R. Jain, Myths About Congestion Management in High Speed Networks, Internetworking Research and Experience, vol. 3, 101-113 (1992). .
Douglas H. Hunt et al., Credit-Based FCVC Proposal for ATM Traffic Management (Revision R1), ATM Forum Technical Committee Traffic Management Subworking Group, ATM.sub.- Forum/94-0168R1, Apr. 28, 1994. .
Douglas H. Hunt et al., Action Item Status for Credit-Based FCVC Proposal, ATM Forum Technical Committee Traffic Management Subworking Group, ATM.sub.- Forum/94-0439, Apr. 28, 1994. .
Timothy P. Donahue et al., Arguments in Favor of Continuing Phase 1 as the Initial ATM Forum P-NNI Routing Protocol Implementation, ATM Forum Technical Committee, ATM Forum/94-0460, Apr. 28, 1994. .
Richard Bubenick et al., Leaf Initiated Join Extensions, Technical Committee, Signalling Subworking Group, ATM Forum/94-0325, Apr. 28, 1994. .
Rob Coltun et al., PRP: A P-NNI Routing Protocol Proposal, ATM Forum Technical Committee, ATM.sub.- Forum/94-0492, Apr. 28, 1994. .
Richard Bubenik et al., Leaf Initiated Join Extensions, ATM Forum Technical Committee, Signalling Subworking Group, ATM Forum 94-0325, Apr. 28, 1994. .
Richard Bubenik et al., Requirements For Phase 2 Signaling Protocol, ATM Forum Technical Committee, Signalling Subworking Group, ATM Forum 94-1078, Jan. 1, 1994. .
H.T. Kung and K. Chang, Receiver-Oriented Adaptive Buffer Allocation in Credit-Based Flow Control for ATM Networks, Proceedings of INFOCOM '95, Apr. 2-6, pp. 1-14. .
H.T. Kung et al., Credit-Based Flow Control for ATM Networks: Credit Update Protocol, Adaptive Credit Allocation, and Statistical Multiplexing, Proceedings of ACM SIGCOMM '94 Symposium on Communications Architectures, Protocols and Applications, Aug. 31-Sep. 2, 1994, pp. 1-14. .
SITA, ATM RFP: C-Overall Technical Requirements, Sep. 1994. .
Hosein F. Badran et al., Head of Line Arbitration in ATM Switches with Input-Output Buffering and Backpressure Control, Globecom, pp. 347-351, 1991. .
Head of Line Arbitration in ATM Switches with Input-Output Buffering and Backpressure Control. By Hosein F. Badran and H. T. Mouftah, Globecom '91, pp. 0347-0351..~
Primary Examiner: Harrell; Robert B.
Attorney, Agent or Firm:Weingarten, Schurgin, Gagnebin & Hayes LLP
Claims

What is claimed is:
1. A method for classifying received data having at least one predetermined format and containing a plurality of data values within said received data comprising the steps of:
in a first parsing step, parsing said received data to retrieve at least first and second data values of said plurality of data values;
in a first searching step, searching a first field of a first lookup table having said first field and a second associated field to locate an entry having a value associated with said first data value, and identifying a first classification value in the respective second associated field of said first lookup table;
in a second searching step, searching a first field of a second lookup table having said first field and a second associated field to locate an entry having a value associated with said second data value, and identifying a second classification value in the respective second associated field of said second lookup table;
concatenating said first and second classification values to generate a third data value;
in a third searching step, searching a first field of a third lookup table to locate an entry having a value associated with said third data value, and identifying a third classification value in the respective second associated field of said third lookup table; and
selecting at least one processing step to be performed upon said received data based upon the value of said third classification value.

2. The method of claim 1 wherein said first, second and third searching steps comprise the steps of performing first, second and third searches of respective first fields of said first, second and third lookup tables to locate first, second and third entries within the respective first fields having values which are equal to said first, second and third data values respectively.

3. The method of claim 2 wherein said first, second and third searching steps comprise the steps of performing first, second and third binary searches of the first fields of respective first, second and third lookup tables to locate first, second and third entries having values within said fields which are equal to said first, second and third data values respectively.

4. The method of claim 1 wherein said first and second searching steps are performed concurrently.

5. The method of claim 1 wherein said selecting step comprises the steps of:
associating said third classification key with said received data;
transmitting said received data and said third classification value to at least one first data processor for processing of said received data; and
identifying a routine for processing said received data based upon the value of said third classification value; and
processing said received data within said first data processor in accordance with the routine identified in said identifying step.

6. The method of claim 5 wherein said associating step comprises the step of encapsulating said third classification value within said received data prior to said transmitting step.

7. The method of claim 1 wherein said first, second and third lookup tables comprise first, second and third ordered lists within a single table.

8. The method of claim 7 wherein said first, second and third ordered lists are stored within a single memory array.

9. The method of claim 7 wherein each of said searching steps includes the steps of:
specifying one of said ordered lists based upon a selection field which specifies the ordered list to be searched;
searching the first field of the ordered list specified in the specifying step to locate an entry which contains a first data portion equal to the respective data value and a second data portion equal to a value which specifies an identification of a virtual local area network for the respective data value.

10. A method for classifying received data having at least one predetermined format and containing at least first, second, third and fourth data values within said received data, comprising the steps of:
in a first parsing step, parsing said received data to retrieve at least said first and second data values;
in a first searching step, searching a first field of a first lookup table having said first field and a second associated field to locate an entry having a value associated with said first data value, and identifying a first classification value in the respective second associated field of said first lookup table;
in a second searching step, searching a first field of a second lookup table having said first field and a second associated field to locate an entry having a value associated with said second data value, and identifying a second classification value in the respective second associated field of said second lookup table;
in a first concatenating step, concatenating said first and second classification values to generate a fifth data value;
in a third searching step, searching a first field of a third lookup table having said first field and a second associated field to locate an entry having a value associated with said fifth data value, and identifying a third classification value in the respective second associated field of said third lookup table;
in a second parsing step, retrieving from said received data at least said third and fourth data values;
in a fourth searching step, searching a first field of a fourth lookup table having said first field and a second associated field to locate an entry having a value associated with said third data value, and identifying a fourth classification value in the respective second associated field of said fourth lookup table;
in a fifth searching step, searching a first field of a fifth lookup table having said first field and a second associated field to locate an entry having a value associated with said fourth data value, and identifying a fifth classification value in the respective second associated field of said fifth lookup table;
in a second concatenating step, concatenating said fourth and fifth classification values to generate a sixth data value;
in a sixth searching step, searching a first field of a sixth lookup table having said first field and a second associated field to locate an entry having a value associated with said sixth data value, and identifying a sixth classification value in the respective second associated field of said sixth lookup table; and
selecting at least one processing step to be performed upon said received data based upon at least one of said third and sixth classification values.

11. The method of claim 10 wherein said first, second, third, fourth, fifth and sixth searching steps comprise the steps of performing first, second, third, fourth, fifth and sixth searches of the first field of respective first, second, third, fourth, fifth and sixth lookup tables to locate entries having values which are equal to the respective first, second, fifth, third, fourth and sixth data values respectively and identifying in the respective second associated field of the respective lookup tables first, second, third, fourth, fifth and sixth classification values.

12. The method of claim 11 wherein said first, second, third, fourth, fifth and sixth searching steps include the steps of performing binary searches of the first field of the respective lookup tables to locate entries having values which are equal to the first, second, fifth, third, fourth and sixth data values respectively and identifying in the second associated field of the respective lookup tables first, second, third, fourth, fifth and sixth classification values.

13. The method of claim 10 wherein selecting step includes the steps of:
associating said third and sixth classification values with said received data;
transmitting said received data and said third and sixth classification values to at least one data processor for processing of said received data; and
identifying a routine for processing said received data based upon the value of at least one of said third and sixth classification values; and
processing said received data within said data processor in accordance with said routine identified in said identifying step.

14. The method of claim 13 wherein said associating step includes the step of encapsulating said third and said sixth classification values within said received data prior to said transmitting step.

15. The method of claim 10 wherein said first, second, third, fourth, fifth and sixth lookup tables comprise ordered lists within a single table.

16. The method of claim 15 wherein said first, second, third, fourth, fifth and sixth lookup tables are stored within a single memory array.

17. The method of claim 15 wherein each of said searching steps includes the steps of:
specifying one of said ordered lists based upon a selection field which specifies the ordered list to be searched;
searching the first field of the ordered list specified in the specifying step to locate an entry which contains a first data portion equal to the respective data value and a second data portion equal to a value which specifies an identification of a virtual local area network for the respective data value.

18. The method of claim 10 wherein said first and second searching steps are performed concurrently and said fourth and first searching steps are performed concurrently.

19. A method for classifying a frame received over a communication link, wherein said received frame contains a Media Access Control (MAC) destination address and a MAC source address, comprising the steps of:
in a parsing step, retrieving said MAC destination and source addresses from said frame;
in a first searching step, searching a first field of a first lookup table having said first field and a second associated field to locate an entry having a value associated with said MAC destination address, and identifying a first classification value in the respective second associated field of said first lookup table;
in a second searching step, searching a first field of a second lookup table having said first field and a second associated field to locate an entry having a value associated with said MAC source address, and identifying a second classification value in the respective second associated field of said second lookup table;
concatenating said first and second classification values to generate a concatenated search value;
in a third searching step, searching a first field of a third lookup table having said first field and a second associated field to locate an entry having a value associated with said concatenated search value, and identifying a third classification value in the respective second associated field of said third lookup table; and
selecting at least one processing step to be performed upon said frame based upon the value of said third classification value.

20. The method of claim 19 wherein said first, second and third searching steps comprise the steps of performing first, second and third searches of said first field of the respective lookup table to locate first, second and third entries which are equal to said MAC destination address, said MAC source address and said concatenated search value respectively.

21. The method of claim 20 wherein said first, second and third searching steps comprise the steps of performing first, second and third binary searches of at least a portion of the first field of the respective lookup tables to locate first, second and third entries which are equal to said MAC destination address, MAC source address and concatenated search values respectively.

22. The method of claim 19 wherein said selecting step comprises the steps of:
associating said third classification key with said frame;
transmitting said received frame and said third classification value to at least one data processor for processing of said received frame; and
identifying a routine for processing said received frame based upon said third classification value; and
processing said frame within said first data processor in accordance with said routine identified in said identifying step.

23. The method of claim 22 wherein said associating step comprises the step of encapsulating said third classification value within said received frame prior to said transmitting step.

24. The method of claim 19 wherein said first, second and third lookup tables comprise first, second and third ordered lists within a single table.

25. The method of claim 24 wherein said first, second and third ordered lists are stored within a single memory array.

26. The method of claim 24 wherein each of said searching steps includes the steps of:
specifying one of said ordered lists based upon a selection field which specifies the ordered list to be searched;
searching the first field of the ordered list specified in the specifying step to locate an entry which contains a first data portion equal to the respective data value and a second data portion equal to a value which specifies an identification of a virtual local area network for the respective data value.

27. The method of claim 19 wherein said first and second searching steps include the step of performing said searches concurrently.

28. A method for classifying a frame over a communication link at a frame processor, wherein said frame includes a Media Access Control (MAC) destination address, a MAC source address, a Network Layer destination address and a Network Layer source address, comprising the steps of:
in a first parsing step, retrieving from said received data at least said MAC destination address and said MAC source address;
in a first searching step, searching a first field of a first lookup table having said first field and a second associated field to locate a first entry having a value associated with said MAC destination address and retrieving a first classification value from the respective second associated field of said first lookup table;
in a second searching step, searching a first field of a second lookup table having said first field and a second associated field to locate a second entry having a value associated with said MAC source address and retrieving a second classification value from the respective second associated field of said second lookup table;
in a first concatenating step, concatenating said first and second classification values to generate a first concatenated search value;
in a third searching step, searching a first field of a third lookup table having said first field and a second associated field to locate a third entry having a value associated with said first concatenated search value and retrieving a third classification value from the respective second associated field of said third lookup table;
in a second parsing step, retrieving from said received frame at least said Network Layer destination address and said Network Layer source address;
in a fourth searching step, searching a first field of a fourth lookup table having said first field and a second associated field to locate a fourth entry having a value associated with said Network layer destination address, and identifying a fourth classification value from the respective second associated field of said fourth lookup table;
in a fifth searching step, searching said a first field of a fifth lookup table having said first field and a second associated field to locate a fifth entry having a value associated with said Network Layer source address and retrieving a fifth classification value from the respective second associated field of said fifth lookup table;
in a second concatenating step, concatenating said fourth and fifth classification values to generate a second concatenated search value;
in a sixth searching step, searching a first field of a sixth lookup table having said first field and a second associated field to locate an entry having a value associated with said second concatenated search value and retrieving a sixth classification value from the respective second associated field of said sixth lookup table; and
selecting at least one processing step to be performed upon said received frame based upon at least one of said third and sixth classification values.

29. The method of claim 28 wherein said first, second, third, fourth, fifth and sixth searching steps include the steps of: performing first, second, third, fourth, fifth and sixth searches of a portion of the first fields of said respective first, second, third, fourth and fifth and sixth lookup tables to locate first, second, third, fourth, fifth and sixth entries having values equal to said MAC destination address, MAC source address, first concatenated search value, Network Layer destination address, Network Layer source address respectively.

30. The method of claim 29 wherein said first, second, third, fourth, fifth and sixth searching steps include the steps of performing binary searches of a portion of the first field of each of the respective lookup tables to locate first, second, third, fourth and fifth entries having values equal to the MAC destination address, MAC source address, first concatenated search value, Network Layer destination address, Network Layer source address and the second concatenated search value and retrieving from the second associated field of the respective lookup tables first, second, third, fourth, fifth and sixth classification values respectively.

31. The method of claim 30 wherein said first and second searching steps are performed concurrently and said fourth and fifth searching steps are performed concurrently.

32. The method of claim 28 wherein selecting step includes the steps of:
associating said third and sixth classification values with said received frame;
transmitting said received frame and said third and sixth classification values to at least one data processor for processing of said received frame; and
identifying a routine for processing said received frame based upon the value of at least one of said third and sixth classification values; and
processing said received data within said data processor in accordance with the routine identified in said identifying step.

33. The method of claim 32 wherein said associating step includes the step of encapsulating said third and said sixth classification values within said received frame prior to said transmitting step.

34. The method of claim 28 wherein said first, second, third, fourth, fifth and sixth lookup tables comprise ordered lists within a single table.

35. The method of claim 34 wherein said first, second, third, fourth, fifth and sixth ordered lists are contained in a single memory array.

36. The method of claim 28 wherein each of said searching steps includes the steps of:
specifying one of said ordered lists based upon a selection field which specifies the ordered list to be searched;
searching the first field of the ordered list specified in the specifying step to locate an entry which contains a first data portion equal to the respective data value and a second data portion equal to a value which specifies an identification of a virtual local area network for the respective data value.

37. Apparatus for classifying frames received over a computer network comprising:
a first classification processor having an input bus interface for receiving a frame having a Media Access Control source address and a Media access control destination address and an output bus interface;
a second classification processor having an input bus interface and an output bus interface, said first classification processor output bus interface being coupled to said second classification processor input bus interface;
a pattern memory containing a plurality of lookup tables wherein each of said lookup tables includes a search field having a plurality of search field entry values and a key field having a like plurality of key field entry values, each of said key field entry values being associated with one of said search field entry values, said pattern memory in electrical communication with said first classification processor, said search field containing at least one first group of search field entry values corresponding to selected MAC source addresses, at least one second group of search field entry values corresponding to selected MAC destination addresses, and a third group of entry values, said key field containing a compact key value associated with each of said first, second and third groups of entry values;
said first classification processor being operative to receive a frame at the input bus interface, parse said frame to extract said source and destination addresses from said received frame, perform first and second searches of said first and second groups of entry values respectively within predetermined ones of said lookup tables to locate a first value in said first group of entry values which matches said parsed source address and a second value in said second group of entry values which matches said parsed destination address, retrieve respective first and second key values associated with said first and second values in said first and second groups which match said parsed source and destination addresses respectively, concatenate said first and second key values to form a concatenated key value, search said third group of entry values to locate a value corresponding to said concatenated key value, extract from said predetermined ones of said lookup tables an output key value associated with said value corresponding to said concatenated key value and forward said output key value from said first classification processor to said second classification processor in association with said received frame.

38. The apparatus of claim 37 wherein said second classification processor includes a memory containing a search table, said search table having a key field containing a plurality of key field entry values and a vector ID field containing a like plurality of vector ID entry values, each of said vector ID entry values being associated with a key field entry value in said search table,
said second classification processor, upon receipt of said output key value from said first classification processor, being operative to search said key field within said search table to locate a value corresponding to said output key value and to retrieve a vector ID value from said vector ID field corresponding to said output key value;
said retrieved vector ID value being employed to select a processing routine to be employed in the processing of said received frame.

39. The apparatus of claim 38 further including a frame modification processor having an input bus interface, said second classification processor output bus interface coupled to said frame modification processor input bus interface,
said second classification processor, being operative to forward said retrieved vector ID to said frame classification processor in association with said received frame; and
said frame modification processor being operative to select said frame processing routine based upon the value of said vector ID.

40. The apparatus of claim 37 wherein said apparatus further includes first and second registers in electrical communication with said first classification processor, said classification processor being operative to store said parsed source address in said first register prior to said search of said first group of entry values and to store said parsed destination address in said second register in said second register prior to said search of said second group of entry values.

41. The apparatus of claim 37 wherein said first classification processor is operative to perform said first and second searches of said first and second groups of entry values substantially simultaneously.

42. The apparatus of claim 37 wherein said first and second searches performed by said first classification processor are binary searches and said first, second and third groups of entry values comprise first, second and third ordered lists of entry values respectively.

43. A method for classifying data comprising the steps of:
parsing said data to retrieve a first data portion;
dividing said first data portion into a first data segment and at least one additional data segment;
in a first searching step, searching a first field of a first lookup table constituting one of a plurality of lookup tables, each of said lookup tables having said first field and a second field, said search being performed to locate a value within said first field of said first lookup table which is associated with the value of said first data segment and retrieving from said second field of said first lookup table a first classification key; and
in a concatenating step, for each of said additional at least one data segments concatenating the last retrieved classification key with another one of said at least one data segments to form a concatenated search value and searching another predetermined one of said plurality of lookup tables to locate a value within said first field which is associated with the respective concatenated search value and following each such search, retrieving a classification key from the corresponding second field of the respective lookup table to generate a final classification key.

44. The method of claim 43 further comprising the steps of:
parsing said data to retrieve a second data portion;
dividing said second data portion into a first data segment and at least one additional data segments;
repeating said first searching step and said concatenating step to generate a second final classification key.

45. The method of claim 44 further comprising the steps of:
concatenating said first final classification key and said second final classification key to produce a final search value;
searching the first field of another one of said plurality of lookup tables to locate a value which equals said final search value and retrieving from the respective second field of said lookup table an output classification key.

46. The method of claim 45 further comprising the step of performing further processing of said data based upon the value of said output classification key.

47. The method of claim 46 wherein said method further comprises the step or receiving a frame at in input port of a network switch prior to said parsing step, wherein said received frame constitutes said data.

48. A frame processor comprising:
first, second, third and fourth frame processing engines, each of said engines having an input port and an output port, the input port of each successive frame processing engine being electrically coupled to the input port of the preceding frame processing engine; said first frame processing engine operative to:
receive a frame of a predetermined type having a predetermined header format at said first frame processing engine input port;
encapsulate within said frame header information identifying said frame type; and
said second frame processing engine operative to:
receive said frame including said encapsulated information identifying said frame type at said second frame processing engine input port from said first frame processing engine output port;
parse said frame header to retrieve a plurality of data values embedded therein; and
generate at least one classification value from said plurality of data values for further processing of said frame;
encapsulate said at least one classification value within said frame header;
said third frame processing engine operative to:
receive said frame including said encapsulated at least one classification value at said third frame processing engine input port from said second frame processing engine output port;
identify a vector in response to said at least one classification value contained within said frame header; and
encapsulate said vector within said frame header;
said fourth frame processing engine operative to:
receive said frame including said encapsulated vector at said fourth frame processing engine input port from said third frame processing engine output port; and
at least some of the time, said fourth frame processing engine modifies said frame in accordance with a routine specified by said vector;
a buffer memory having an input port operative to:
receive said frame at said buffer memory input port from said fourth processing engine output port and to store said frame following receipt thereof;
wherein said first, second, third and fourth processing engines and said memory are pipelined and said first, second, third and fourth frame processing engines process said frame in sequence prior to storage of said frame in said buffer memory.

49. The frame processor of claim 48 wherein each of said frame processing engines is capable of processing a different frame concurrently.

50. A method of classifying a frame received at an input port of a network switch comprising the steps of:
parsing said received frame to retrieve a plurality of data segments;
in a first searching step, for each of said plurality of data segments, searching a first field of a predetermined one of a plurality of lookup tables having said first field and an associated second field to locate a value within said first field associated with the respective data segment and retrieving from the respective second field a classification key;
concatenating said classification keys retrieved in said searching step to form a concatenated search value;
in a second searching step, searching the first field of another predetermined one of said plurality of lookup tables having said first field and a second associated field to locate a value within said first field associated with said concatenated search value and retrieving the respective second field an output classification key; and
performing further processing of said frame based, at least in part, upon the value of said output classification key.
Description
FIELD OF THE INVENTION

The present invention relates to networking products and techniques and more specifically, to apparatus and methods for high speed frame recognition, classification and processing.

BACKGROUND OF THE INVENTION

Over the past decade with the increase in the use of computers in the corporate environment, there has been a rapid increase in the utilization of networks to facilitate communication and transmittal of information between computers connected to such networks. Initially, most of the networks were stand-alone networks and each network operated in accordance with a particular specification or standard. Some of the most common data link layer standards include the IEEE 802.3 standard, the Ethernet version 2 standard, the IEEE 802.5 token ring standard and the Fiber Distributed Data Interface (FDDI) standard and more recently the Asynchronous Transfer Mode (ATM) standard. Some of the common Network Layer standards include TCP/IP, DECNET, IPX and Appletalk. No two of the standards within the same protocol level employ the same frame format and the protocols implemented in many of the standards operate at differing speeds.

In the last several years there has been an increasing need to interconnect the disparate networks to facilitate communication and information transfer among such networks. This function is performed by frame processors which serve, among other functions, to recognize and translate frames received from one network in one frame format for use in another network having another frame format.

Frame processors typically have a number of input and output ports for receipt and transmission of frames. Such processors must therefore be able to recognize the incoming frame format, categorize and translate the frame if the frame is to be transmitted over a network possessing a different frame format from that of the received frame, and identify the forwarding path including the physical output port to be employed. It is desirable to process and translate received frames rapidly so as to minimize latency in the frame processor.

Frame processors have historically performed the above described process in software. This approach requires that the frame first be stored in memory and adds undesirable delay to the frame conversion process. Additionally, frame recognition and translation via software driven processors, does not permit frames to be processed at the frame reception rates. Accordingly, substantial buffering must be provided and/or frames discarded during high traffic bursts.

While some fast processors exist, they tend to employ techniques which require substantial memory resources resulting in a product which is expensive and which remains subject to overload at high frame rates.

It would therefore be desirable to have a frame processor which would be capable of recognizing, categorizing and translating frames at frame reception rates while minimizing the hardware resources necessary to implement such functions and the likelihood of overload at high frame rates.

SUMMARY OF THE INVENTION

In accordance with the present invention an improved method and apparatus for recognizing, characterizing and processing frames received from a plurality of computer networks is disclosed. A received frame having an identifiable frame format is received at one of a plurality of ports of a Local Area Network (LAN) interface. The received frame, which includes Media Access Control (MAC) source and destination addresses, is forwarded to a Mapping Engine state machine after segmenting the frame into a plurality of framelets, the first of which includes frame header information. The Mapping Engine state machine comprises two state machines; namely a Parsing Engine state machine and a Binary Compare Engine state machine. The Parsing Engine state machine extracts the MAC source and destination addresses from the initial framelet which contains the frame header information and stores the respective addresses in first and second Binary Compare Registers along with other frame identifying information.

A plurality of lookup tables is set up and maintained in a high speed static memory by system software. The lookup tables may comprise an addressable random access memory, a content addressable memory (CAM), or any other suitable memory which may be used for storing information and searching the stored information. The lookup tables include a search field and a key field. The search field, for example, for MAC layer addresses, is comprised of three fields, namely, an address field, a DLDID field and an ordered list Selection (Type) field. MAC source and destination addresses are stored in the address field. The DLDID field is loaded with a Data Link Domain identifier which serves to associate as a virtual lan, various workstations which may reside on different physical ports of the frame processor. The ordered list Selection (Type) field is loaded with a value which selects one of the plurality of lookup tables each of which comprises an ordered list. For example, the MAC Source Addresses, the MAC Destination Addresses, the Network Layer Source Addresses, concatenated MAC Destination and Source compact keys (MD Key and MS Key respectively) and the concatenated Network Destination and Source compact keys (ND and NS Key respectively) are each contained within separate ordered lists uniquely identified by the value within the ordered list selection field. A classification key associated with each entry in the ordered lists is stored in the key field.

Following extraction of the MAC source and destination addresses by the parsing engine state machine and the storing of such addresses within the respective binary compare registers, binary searches of appropriate ordered lists are initiated under the control of the Binary Compare Engine state machine to locate values within the search field which match the values within the Binary Compare Registers. In the event exact matches are identified as a result of the binary search, first and second MAC compact classification keys, identified as the MAC Destination Key (MD Key) and the MAC Source Key (MS Key) associated with the respective MAC destination and source addresses are retrieved. The MAC Destination and Source Keys identified as a result of the search are extracted from the respective ordered lists and stored in first and second processor registers.

In a similar manner, Network Layer source and destination addresses are parsed from the received frame, stored in the respective first and second Binary Compare Registers and a binary search of the respective ordered lists is performed under the control of the Binary Compare Engine state machine to locate entries within the respective search fields which match the data in the respective binary compare registers. In the event matches are identified, first and second Network Layer compact classification keys, identified as the Network Destination Key (ND Key) and the Network Source Key (NS Key) associated with the respective Network Layer destination and source addresses are retrieved and are stored in specified processor registers.

The MD Key and MS Key are then concatenated to form a first search value which is stored within one of the Binary Compare Registers and the ND Key and NS Key may be concatenated with an FEP/XPT value which identifies the transport layer to form a second search value which is stored in the other of the Binary Compare Registers. The search field of the respective ordered list is next searched using the Binary Compare Engine state machine in an effort to locate entries within the lookup tables which present exact matches of the first and second search values contained in the first and second Binary Compare Registers, once again searching the appropriate ordered lists identified by the ordered list Selection (Type) field. In the event exact matches are identified, a MAC address classification key (referred to as a Midget) and a Network Layer address classification key, (referred to as a Nidget) are extracted from the associated key field of the respective lookup tables and are stored in specified processor registers.

The Network Layer compact classification key (the Nidget) and the MAC Layer compact classification key (the Midget) are encapsulated within the initial framelet for the respective frame. The framelet including the encapsulated compact classification keys is next forwarded to subsequent stages such as a Search Engine state machine, a Modification Engine state machine, and a Statistics Engine state machine for further classification, modification and/or processing. More specifically, the Search Engine state machine resolves the Nidget and the Midget into a Vector ID which is subsequently employed by the Modification Engine state machine to decide what processing routine to apply to the respective frame.

The processing time for each state machine to perform its prescribed function is no greater than the frame reception time. Thus, the frame processor can process incoming frames at frame reception rates without the performance penalties associated with a software implementation or the cost or performance penalties associated with frame processors which store the received frame in memory prior to parsing and processing of the frame.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will be more fully understood by reference to the detailed description of the invention and the accompanying drawing of which:

FIG. 1 is a block diagram of a frame processor in accordance with the present invention;

FIG. 2a is a diagram illustrating the Ethernet version 2 frame format;

FIG. 2b is a diagram illustrating the IEEE 802.3 frame format;

FIG. 2c is a diagram illustrating the IEEE 802.5 token ring frame format;

FIG. 2d is a diagram illustrating the fiber distributed data interface frame format;

FIG. 3a is a representation of the format of a frame as received at the input to the frame processor;

FIG. 3b is a representation illustrating the frame format as transmitted by the LAN Adaption Interface to the Mapper Engine state machine;

FIG. 4 illustrates the bit format within the Start of Transfer (SOT) field;

FIG. 5 illustrates the bit format within the Adaptation Layer Encapsulation (ALE) field;

FIG. 6 is a block diagram of the Mapper Engine state machine employed in the frame processor of FIG. 1;

FIG. 7 illustrates the organization of the lookup tables which are stored in the Pattern Memory of FIG. 1;

FIG. 8 illustrates the input and output data associated with frame classification by the Mapping Engine state machine for MAC and Network Layer address classification;

FIG. 9a illustrates the input format for frames entering the Mapper Engine state machine;

FIG. 9b illustrates the output format for frames exiting the Mapper Engine state machine;

FIG. 10a illustrates the input format for frames entering the Search Engine state machine;

FIG. 10b illustrates the output format for frames exiting the Search Engine state machine;

FIG. 11 is a timing diagram illustrating the pipelining of frames through the respective state machines of the frame processor of FIG. 1; and

FIG. 12 is a diagram illustrating a five level compaction process in accordance with the present classification method.

DETAILED DESCRIPTION OF THE INVENTION

In accordance with the present invention a method and apparatus for recognizing, categorizing and processing frames received over one or more local area networks at a rate approximating or equal to the frame reception rate is disclosed. The ability to recognize and categorize received frames at high frame rates is achieved through the use of a novel technique for frame recognition and categorization.

Referring to FIG. 1, a frame processor 10 which permits the implementation of various bridging and routing protocols includes a forwarding engine state machine 12 which includes a LAN Adaptation Interface (LAI) 14 having a plurality of Input/Output ports 16, a Mapper Engine state machine 18, a pattern memory 20, a Search Engine state machine 22, a Modification Engine state machine 24, a Statistics Engine state machine 26, a multiplexer 28 and a fragmentation engine state machine 30. The frame processor further includes a Buffer Manager state machine 32 and a Segmentation and Reassembly (SAR) state machine 34.

The output of the LAI 14 is coupled to the input of the Mapper Engine state machine 18 via a bus 36 such as a bus operative in accordance with the Interchip Bus protocol. Similarly, the Mapper Engine state machine 18 is coupled to the Search Engine state machine 22 via a bus 38, the Search Engine state machine 22 is coupled to the Modification Engine state machine (Mod. Engine) 24 via a bus 40, the Modification Engine state machine 24 is coupled to the Statistics Engine (Stat. Engine) state machine 24 via a bus 42 and the Statistics Engine state machine 26 is coupled to the input of a multiplexer 26 via bus 44. Additionally, the Modification Engine is coupled to the multiplexer 28 via bus 46. The buses 38, 40, 42, 44, and 46 are all operative in accordance with the Interchip Bus protocol although it should be appreciated that any suitable high speed bus may be employed. In a preferred embodiment the buses are 21 bits wide and include 16 data lines, 4 cycled definition lines and 1
halt line. The Interchip Bus permits a unidirectional flow of data from each engine within the frame processor to the connected engine or engines via a single and consistent interface mechanism.

The output of the multiplexer 28 is coupled to a Buffer Manager state machine 32 via a bus 48 and the output of the Buffer Manager state machine 32 is coupled to the input of the fragmentation engine (Frag. Engine) state machine 30 via bus 50. Additionally, the output of the fragmentation engine 30 is coupled to the input of the LAN Adaptation Interface 14 via a bus 52. Finally, the Buffer Manager may transmit or receive frames over a bus 54 to or from the Segmentation and Reassembly (SAR) state machine 34 for communication with an Asynchronous Transfer Mode (ATM) switch (not shown) over an ATM link 56. The Segmentation and Reassembly state machine segments frames into cells as required in accordance with the ATM protocol prior to transmittal or alternatively, reassembles cells received over the ATM link into frames.

The input/output ports 16 of the LAN Adaptation Interface 14 provide connections to various local area networks such as networks complying with the Ethernet version 2 protocol, the IEEE 802.3 protocol, the IEEE 802.5 Token Ring protocol, the fiber distributed data interface (FDDI) protocol or any other suitable network protocol. The LAN Adaptation Interface, performs the function of providing a physical interface to the various attached local area networks and additionally provides bandwidth normalization to adapt the frame reception rates of the various ports 16 to the processing rates within the forwarding engine 12 of the frame processor 10. More specifically, referring to FIGS. 2a-2d, each port 16 provides an interface to a specific local area network which in the preferred embodiment may be compliant with the frame format for Ethernet version 2 as depicted in FIG. 2a, the IEEE 802.3 protocol frame format as indicated in FIG. 2b, the IEEE 802.5 frame format is illustrated in FIG. 2c or the FDDI frame format as indicated in FIG. 2d. The network protocols associated with all of the ports 16 may be the same for simplicity, or alternatively, may be configured for compliance with different network protocols.

Received frames are conveyed from the LAI 14 to the Mapper Engine 18 employing an adaptation layer encapsulation (ALE) technique. More specifically, referring to FIGS. 3a and 3b, the LAI 14 receives a frame 58 having a header 60, a data portion
62 and a Frame Status Word (FSW) 64 marking the end of the frame. The LAI 14 segments the received frame into framelets 66, 68 and 70 which may be separately transmitted over the Interchip Bus 36 to the Mapper Engine 18.

As illustrated in FIG. 3b, the first framelet 66 comprises a Start of Transfer (SOT) field 72, an Adaptation Layer Encapsulation (ALE) field 74 and header data 76 which corresponds to the header data 60 in the received frame 58. An End of Transfer (EOT) delineator marks the end of the initial framelet 66. The received frame is further segmented into additional framelets as reflected by framelets 68 and 70. Each additional framelet commences with an SOT field and ends with a EOT delineator. In the presently disclosed embodiment, framelets may have a length up to 128 bytes although any suitable fixed predetermined framelet length may be employed.

Referring to FIG. 4, the SOT field comprises a 16 bit word which includes Segment # bits (bits 0-4), a Length bit (bit 5), a Start of Frame (SOF) bit (bit 7), Engine Control bits (bits 8-12) and Type bits (bits 13-15). The Segment # bits identify the port 16 of the LAI 14 over which the frame was received. The Length bit within the SOT field is used to identify whether the last word in the framelet is two bytes in length (bit=0) or one byte in length (bit=1). The SOF bit serves to indicate that the present framelet begins a new frame. The Engine Control bits permit the specific engines to be separately enabled or disabled. More specifically, the M bit (bit 12) is the Mapper Engine enable/disable bit, the S bit (bit 11) is the Search Engine enable/disable bit, the X bit (bit 10) is the Modification Engine enable/disable bit, the ST bit (bit 9) is the Statistics Engine enable/disable bit and the FR bit (bit 8) is the Fragmentation Engine enable/disable bit. The ability to separately enable and disable the respective engines within the frame processor 10 permits a frame to be presented to the Mapper Engine 18 through the Host Interface 19 (shown in FIG. 1) from a microprocessor (not shown), which, for example, is ready for processing by the Modification Engine 22 and to allow the Mapper and Search Engines state machines 18 and 20 to skip processing of the frame. The Type and Segment # fields within the SOF field describe the source of the framelet, thereby permitting multiple frames to utilize the Interchip Buses on a time shared basis.

As indicated in FIG. 3b, the second word in the initial framelet 66 for a frame that is output from the LAN Adaptation Interface comprises the ALE field 74. The format for the ALE field 74 is illustrated in FIG. 5. The ALE field 74 includes Type bits (bits 0-3) which are used to identify a parsing algorithm to be employed in the Mapper Engine 18. More specifically, the LAI identifies the received frame as arriving over a network having a particular network protocol and the Type bits are set to permit the Mapping Engine to select a parsing algorithm which is appropriate for the respective network protocol. Additionally, the ALE field contains Pad bits (bits 4-5) which identify the number of words of pad before the actual Media Access Control (MAC) Protocol Data Unit (PDU), and a Z bit (bit 6) which is used to indicate whether the received frame is in "standard" or "swizzled" format as hereinafter discussed.

The ALE field further contains Data Link Domain Identification bits (DLDID) (bits 8-12) which permit workstations associated with certain ports 16 to be identified as having a common DLDID. Workstations having a common DLDID may thus be associated as a virtual lan.

Structure of the Mapper Engine State Machine

The Mapper Engine state machine 18 of FIG. 1 is illustrated in greater detail in FIG. 6. The Mapper Engine 18 receives framelets, such as framelets 66, 68 and 70 over the bus 36 and parses the received framelet to ascertain the Media Access Control (MAC) and network layer source and destination addresses. Referring to FIG. 6, the Parsing Engine state machine 90, which operates under the control of a Microprogrammable Control Store 92, is operative to perform this function. Additionally, the Mapper Engine includes a Binary Compare Engine state machine 94 which commences operation upon receipt of a start signal 96 from the Parsing Engine 90. The Mapper Engine 18 further includes first and second fifty-six bit Binary Compare Registers 98
and 100 respectively which are used to hold specified search patterns. The Binary Compare Registers 98 and 100 are loaded by the Parsing Engine 90 prior to initiating the operation of the binary compare engine 94. The Binary Compare Engine 94 is used in conjunction with a Pattern Memory 102 which contains a plurality of lookup tables in the form of ordered lists which are hereinafter described in greater detail. The Mapper Engine state machine 18 also includes first and second comparators 104 and
106 respectively which are employed in conjunction the Binary Compare Engine 94 to ascertain whether data in the first and second Binary Compare Registers 98 and 100 matches entries within the respective lookup tables stored in the Pattern Memory 102. The operation of the Parsing Engine 90 commences when Input Control Circuitry 108 recognizes the arrival of an SOT field over the Interchip Bus 110 from the LAI 14. The incoming framelet data is stored in a frame FIFO 112 and the initial framelet of a frame is parsed as discussed above to extract from the framelet the MAC source and destination addresses and the network source destination addresses contained within the framelet. The Mapper Engine 18 also includes an Output Data Buffer 114 which is used to store data to be inserted into the initial framelet of a frame prior to the forwarding of the framelets from the output of the Mapper Engine state machine 18 over bus 116 to the Search Engine state machine. Merging of the received framelet data with the data stored in the Output Data Buffer 114 is accomplished via the use of a Multiplexer 118 which receives framelet data over Signal Lines 120 and output data over Signal Lines 122. The merging of the output data with the framelet data prior to transmission by the Mapper Engine over the Mapper Output Interchip Bus 16 is controlled by Output Control Circuitry 124 upon activation by the Parsing Engine state machine 90.

Lookup Table Organization

An example of the lookup table 130 organization within the pattern memory 102 is depicted in FIG. 7. The data included within the table is purely exemplary data for purposes of discussion. The lookup tables, in a preferred embodiment, comprise ordered lists which include a forty-eight bit (6 byte) Address Field 132, a five bit Data Link Domain Identification Field (DLDID) 134, a three bit Selection (Type) Field 136, a one bit Valid Field 138, two single bit Status Fields S1 140 and S2 142, a five bit Segment Field 144 and a sixteen bit (two byte) Key Field 146. For MAC addresses, the Address Field 132, DLDID Field 134 and Selection (Type) Field 136 in combination form a fifty-six bit Search Field 148. It should be appreciated that the search field may contain any desired information to be searched.

The lookup tables may be considered as a single table comprising a plurality of ordered lists which are selectable, for example by the Type field (see FIGS. 7 and 8), or alternatively, as a plurality of individual tables which are co-resident in the pattern memory, with the particular table being selected, for example, by the Type field.

As a software maintenance function, the lookup tables 130 are initially set up and are modified during the operation of the frame processor to reflect valid connection information. The Address Field of the respective ordered lists contain MAC source addresses, MAC destination addresses, network source addresses, network destination addresses, a first group of concatenated classification keys, a second group of concatenated classification keys and optionally additional groups of classification keys. As an example, a given MAC source address is stored in the Address field 132, the associated DLDID in 134 and the Selection (Type) Field 136 is loaded with a unique identifier which specifies the particular ordered list within one of the lookup tables 130. Thus, in the present embodiment, the three bit Selection (Type) Field allows up to eight ordered lists may be maintained. The Segment Field 144 is loaded with the segment number corresponding to the physical port over which the frame was received (0-15) or alternatively, a virtual port identification (16-31). The Status Bits S1 140 and S2 142 may be used as an indicator to the parsing engine. More specifically, the Status Bits contain information relevant to the particular frame format being processed. The Binary Compare Engine presents the state of the status bits to the parsing engine upon completion of a successful match operation and the Parsing Engine uses the state information in a manner appropriate to the frame format being processed. A two-byte classification key is stored in the Key Field 146, each being associated with one of the entries in the Search Field 148 of the respective ordered list.

Operation of the Mapper Engine State Machine

The Mapper Engine state machine 18 classifies pairs of addresses, such as MAC source and destination addresses and Network Layer source and destination addresses, to generate particular compact classification keys which may be used by subsequent engines within the frame processor for the processing of the frame in the manner hereinafter described.

Referring to FIGS. 7 and 8, after parsing the initial framelet of a frame to identify the MAC source and destination addresses, the parsing engine causes the respective 48 bit addresses to be stored in the first and second Binary Compare Registers 98 and 100. These addresses are concatenated with the associated DLDID for the frame and a value identifying a specific ordered list to be searched within the lookup tables 130. While in the present embodiment, the MAC source and destination addresses are forty-eight bits (6 bytes) wide, the DLDID information is five bits wide and the ordered list Selection (Type) Field is three bits wide, it is recognized that the width of the respective fields is a matter of design choice. The concatenation of a MAC address, the DLDID information and the ordered list selection information forms a fifty-six bit value which is stored in the respective Binary Compare Registers 98 and 100.

Upon activation by the Parsing Engine 90, the Binary Compare Engine 94 causes a binary search of the appropriate ordered lists within the lookup tables 130 to be conducted to ascertain whether exact matches for the data contained in the Binary Compare Registers 98 and 100 are to be found within the respective ordered lists of the lookup table 130. Each binary search is performed of the appropriate ordered list by initially accessing an entry approximately at the middle of the first field of the respective ordered list and testing the entry in the ordered list to see if such entry matches, is greater than or less than the value of the data in the respective Binary Compare Register. If a match is detected, the binary search is complete and an indication of the match is provided to the Binary Compare Engine. If the data value being searched is greater than the accessed entry in the ordered list, the relevant portion of the ordered list comprising the lower ordered list values is again divided, the middle entry of such portion accessed, and the respective value being searched is again compared to the value retrieved from the respective lookup table. The above process is reiterated until a match is detected or until it is determined that no corresponding entry is present within the respective lookup table. Alternatively, if the search value is less than the initial accessed entry from the ordered list, the relevant portion of the ordered list comprising the upper ordered list portion is again divided, the middle entry of such portion accessed and the respective search value is compared to the value retrieved from the lookup table. This process is likewise repeated until it is determined that a match exists or that there is no entry within the respective lookup table to the value being searched. To improve bandwidth utilization during the search process within the Binary Compare Engine, the comparison of a search value to a retrieved value is performed concurrently with a lookup of a second search value.

If a search of the Search Field 148 of the specified ordered list produces a match with the data in the Binary Compare Register 98 which contains a MAC source address, associated DLDID and ordered list selection information, a two byte classification key (MS Key) is retrieved from the Key field 146 and stored in a specified output register. Similarly, if the comparison between the information in the second Binary Compare Register 100, which contains a MAC destination address, associated DLDID and ordered list selection information, with the Search Field 148 yields an exact match, the associated two byte key (MD Key) is retrieved from the Key Field 146 and stored in an a specified output register. For example, if the search resulted in a match with the second entry in the MDA ordered list, the key MD Key.sub.2 would be retrieved from the respective Lookup table and if the search revealed a match with the third entry of the MSA ordered list, the key MS Key.sub.3 would likewise be retrieved from the respective Lookup table.

It should further be appreciated that in certain circumstances, such as in cases in which all of the bits within the search field are not employed in the search, a match may be indicated where the search value is related to or associated with the value within the search field of the respective ordered. More particularly, if some of the bits within the search value constitute "don't care" conditions for the search, a match may be indicated if the comparison of the remaining bits of interest constitute a match.

To expedite the search process, the binary searches for the keys associated with MAC source and destination addresses are performed on an overlapping basis such that the two classification keys (MD Key and MS Key) are identified at substantially the same time. The two classification keys derived from the searches of the respective ordered lists thus produce two two-byte classification keys which serve to classify the respective source and destination addresses and associated data link domain identification information. The Binary Compare Engine state machine 94 next causes the first and second classification keys (MD Key and MS Key) derived from the search of MAC related ordered lists to be concatenated and stored temporarily within a register. In the above referenced example, MD Key.sub.2 would be concatenated with MS Key.sub.3 and thereafter stored temporarily in a register.

Following identification of the MAC compact keys, in a manner similar to that described above in connection with the MAC source and destination addresses, the parsed Network Layer source and destination addresses are loaded into the Binary Compare Registers and first and second Network Layer classification keys (ND Key and NS Key) are identified in the event that matches are located within appropriate ordered lists of the lookup tables 130. These keys are also concatenated, stored in one of the Binary Compare Registers and employed as the basis for a further binary search of the appropriate ordered list within the lookup tables 130 on an overlapping basis to provide the search output results substantially simultaneously. In the above referenced example, if a search of the Search Field 148 yielded a match with the NDA.sub.1 entry and a search of the Search Field 148 yielded a match with the NSA.sub.2 entry, the ND Key.sub.1 and NS Key.sub.2 entries would be retrieved, concatenated and stored temporarily in a register.

If one or both of the binary comparisons of the respective registers 98 and 100 with the respective search fields of the lookup tables do not produce match, the frame is passed to a microprocessor through the host interface 19 for special handling.

Following the identification of the compact classification keys for the MAC addresses (MD Key and MS Key) and the Network Layer addresses (ND Key and NS Key), the concatenation of such compact classification keys and the loading of such keys into respective first and second Binary Compare Registers, the Binary Compare Engine cause further binary searches of the lookup tables 130 to ascertain whether there is an exact match for the two concatenated key pairs within the respective ordered lists of the lookup tables 130. If exact matches are identified via the comparators 104 or 106 between the concatenated keys stored in the respective Binary Compare Registers and entries within the respective ordered lists, the associated keys are retrieved by the Parsing Engine and caused to be stored in the Output Data Buffer 114. In the above referenced example, the search of the search field 148 for the concatenated entry MD Key.sub.2 -MS Key.sub.3 would yield a match with the second entry in the fifth ordered list. The associated classification key (Midget.sub.2) would then be retrieved and stored in the Output Data Buffer 114 for insertion into the initial framelet of the frame in a manner hereinafter discussed. Additionally, the search of the Search Field for the concatenated entry ND Key.sub.1 -NS Key.sub.2 would yield a match with the entry in the sixth ordered list. The associated classification key (Nidget.sub.1) which then be retrieved and stored in the Output Data Buffer 114 for insertion into the initial framelet of the frame.

Following the generation of the two two-byte compact keys (namely, the Midget and the Nidget) in the above described manner, the header framelet is further modified to insert into the initial framelet the classification key information obtained via the binary searches of the lookup tables. More specifically, referring to FIG. 9b, prior to transmission of the initial framelet of a frame out of the Mapper Engine state machine 18 to the Search Engine state machine 22 via the bus 36 (see FIG. 1), the Mapper Engine state machine 18 causes the Network Layer classification key referred to as the Nidget to be inserted into the initial framelet of the frame after the SOT field and causes the MAC layer address classification Key referred to as the Midget to be inserted into the framelet after the Nidget. Additionally, following the Midget, a MAC descriptor field is inserted into the framelet. The MAC descriptor field contains two information fields; namely, the MAC destination indicator and the MAC source address classification key. The MAC destination indicator identifies the frame as either a unicast or a multicast frame. Such information is employed within the frame processor for source aging. The initial framelet of the frame, as thus modified and as illustrated in FIG. 9b, is transmitted out of the Mapper Engine 18 for processing by the Search Engine 22.

The Mapper Engine state machine further provides for same port bridge filtering in the following manner. If the Parsing Engine state machine determines that both the source and destination MAC addresses are valid within the Pattern Memory 102
and both the source and destination addresses reside on the same segment, the frame qualifies for discard. In such event, the Mapping Engine ascribes specific Nidget and Midget classification keys which indicate to the Search Engine that the frame is to be discarded. The Nidget includes an identification of the involved input port to insure proper statistics updating. The MAC source key is transmitted to insure proper source aging in the Statistics Engine state machine and the MAC Protocol Data Unit (PDU) is discarded by the Mapper Engine state machine 18. The Start of Frame framelet is truncated following transmission of the MAC Source Key. Further, the Mapper Engine records the state of the segment as qualifying for Source Port Bridge Filtering and discards any further information from the respective segment until an End of Frame condition is detected. If the source or destination MAC address are not mapped in the Pattern Memory, the frame will be processed as an unrecognized frame.

Use of Mapper Engine Generated Keys

The Search Engine 22, within the frame processor 10 continues and completes the process of frame recognition and classification based upon the compact classification keys embedded within the initial framelet of the received frame. The framelet format for framelets received and transmitted by the Search Engine state machine are illustrated in FIGS. 10a and 10b. Based upon the Network Address classification key (Nidget) and the MAC address classification key (Midget), the Search Engine state machine performs byte serial testing of the frame header against string values known for the specified frame addresses. In this manner both discrimination of the frame's vector and validation of the protocol requirements are completed.

The Search Engine state machine validates the header data and completes the frame recognition process. More specifically, the Search Engine accesses a search table to associate a Vector ID with the received Nidget and Midget. Following identification of the Vector ID, the Search Engine inserts the Vector ID (VID) into the initial framelet in a field after the SOT field and before the Nidget, as depicted in FIG. 10b. The Vector ID is used as a pointer to particular microcode and/or software for subsequent processing of the frame in the respective frame processor engines. Basing the search table on compact (two byte) classification keys rather than full width addresses allows the keys to be sequentially assigned via software administration routines and permits the search table within the Search Engine state machine 22 to have a much higher density of known values than would be achieved if the addresses themselves were used for indexing or searching.

The Modification Engine state machine 24 uses the Vector ID embedded within the initial framelet of frames received from the Search Engine state machine to select a particular processing routine to be employed within the engine. One such application in the Modification Engine involves "swizzling", the process of reformatting addresses within frames received in one protocol to comply with the address format of another protocol. In particular, MAC layer addresses as defined in the IEEE
802.3 and 802.4 specifications are defined as six octets wherein the least significant bit stored in memory is transmitted first and the most significant bit is transmitted last. In the IEEE 802.5 and FDDI protocols, the most significant bit is transmitted first and the least significant bit is transmitted last. Thus, in translating frames received in the 802.3 frame format for transmission in compliance with the 802.5 frame format, the MAC address bits must be reversed prior to transmission. In the present example, the Vector ID conveyed from the Search Engine to the Modification Engine may be used to select a particular software control routine for swizzling and/or translation from one protocol as well as other frame modifications that must be made prior to transmission of the frame. The output of the Modification Engine state machine 24 is forwarded to the Buffer Manager state machine 32 through multiplexer 28 via bus paths 46 and 48.

As discussed above the Modification Engine state machine 22 is also coupled to the Statistics Engine 24 through bus 42. The Statistics Engine state machine 24 records frame and byte statistics for the respective protocols, records interface frame and byte statistics and performs MAC source address timestamp updating for source address aging.

The Buffer Manager transmits received frames to the Fragmentation Engine state machine 30 over bus 50. The Fragmentation Engine state machine applies protocol specific routines in order to segment oversized frames. More specifically, the frame's length, as supplied by the Buffer Manager state machine is tested against a maximum permitted length for the respective protocol for the respective input/output port 16. Frames which do not require segmentation are retransmitted by the Fragmentation Engine state machine 30 to the LAI 14 unchanged over bus 52 for transmittal over one of the ports 16 to the connected network.

Alternatively, with respect to a frame directed for transmission over the ATM link 56, the Buffer Manager state machine 32 transmits the frame to the Segmentation and Reassembly state machine 34 for proper segmentation in accordance with the ATM protocol requirements prior to transmission of the frame.

Frame processing, as depicted in FIG. 11 is handled on a pipelined basis. More specifically, each state machine sequentially processes received frames at the frame reception rate, which, in the present embodiment is approximately 4 microseconds. Thus, as apparent from FIG. 11 frames may be processed at a rate which avoids overrun conditions and the necessity to discard frames.

Finally, it should be appreciated that the compaction technique described above may be extended through multiple levels of compaction. More specifically, some network protocols, such as ISO CLNP, have addresses which exceed six b