One document matched: draft-shiomoto-ccamp-misconnection-analysis-00.txt



   CCAMP Working Group                                K. Shiomoto (NTT) 
   Internet Draft                            V. Sharma (Metanoia, Inc.) 
   Document: draft-shiomoto-ccamp-                     Y. Suemura (NEC) 
   misconnection-analysis-00.txt 
   Expires: December 2004                                     June 2004 
    
    
           Analysis of Misconnection Scenarios in GMPLS Networks 
            draft-shiomoto-ccamp-misconnection-analysis-00.txt 
    
    
  Status of this Memo 
    
   This document is an Internet-Draft and is in full conformance 
   with all provisions of Section 10 of RFC2026. 
    
   Internet-Drafts are working documents of the Internet Engineering 
   Task Force (IETF), its areas, and its working groups.  Note that      
   other groups may also distribute working documents as Internet-
   Drafts. 
    
   Internet-Drafts are draft documents valid for a maximum of six months 
   and may be updated, replaced, or obsoleted by other documents at any 
   time.  It is inappropriate to use Internet-Drafts as reference 
   material or to cite them other than as "work in progress." 
    
   The list of current Internet-Drafts can be accessed at 
        http://www.ietf.org/ietf/1id-abstracts.txt 
   The list of Internet-Draft Shadow Directories can be accessed at 
        http://www.ietf.org/shadow.html. 
    
    
   Abstract 
    
   The purpose of this memo is to describe and analyze the occurrence of 
   misconnections in GMPLS networks.  We present a problem statement, 
   and describe some scenarios under which misconnections may happen. We 
   then outline some common causes of misconnections and discuss some 
   solutions to address them. Our goal is to facilitate further 
   discussions of this issue and the associated solutions within the 
   CCAMP Working Group. 
 
 
   Table of Contents 
    
   1. Introduction...................................................2 
   2. Conventions used in this document..............................2 
   3. Problem Statement..............................................2 
   4. Analysis of Misconnection Scenarios............................3 
 4.1.  Activating Backup LSPs in Shared Mesh/1:1 Restoration........3 
 4.2.  Activating High-priority versus low-priority LSPs............5 
 4.3.  Failure at a switch or node..................................7 

     
   Shiomoto/Sharma/Suemura    Expires December 2004                  1 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
   5. Common Causes of Misconnections................................8 
   6. Possible Solutions to Handle Misconnections....................9 
   7. Conclusions....................................................9 
   8. Security Considerations.......................................10 
   9. References....................................................10 
 9.1.  Normative References..............Error! Bookmark not defined. 
 9.2.  Informative References............Error! Bookmark not defined. 
   10.  Author's Addresses..........................................10 
   11.  Intellectual Property Considerations........................11 
 11.1. IPR Disclosure Acknowledgement..............................11 
   12.  Full Copyright Statement....................................11 
    
 
1. Introduction 
    
   As providers gain experience with MPLS deployments, and begin 
   utilizing their MPLS/GMPLS networks as a common infrastructure upon 
   which to offer a variety of converged services (e.g., Layer 2 
   transport over MPLS, Layer 2 and Layer 3 MPLS VPNs, VoIP services, 
   and critical business data transport), a number of different aspects 
   related to MPLS/GMPLS network stability, operations, and management 
   become increasingly important. 
    
   In particular, with new restoration methods and sophisticated pre-
   emption schemes that are possible, providers are interested in 
   understanding the nature and types of misconnections that may occur 
   in GMPLS-based networks, and in developing ways to either prevent 
   misconnections (where possible) or to detect misconnections and 
   minimize their impact in a timely manner on the network operations 
   and the services offered. 
    
2. Conventions used in this document 
    
   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 
   "SHOULD", "SHOULD NOT", "RECOMMENDED",  "MAY", and "OPTIONAL" in this 
   document are to be interpreted as described in RFC-2119 [1]. 
    
3. Problem Statement 
    
   Based on the description in Section 1, our problem can be stated as: 
   "To analyze when and how misconnections may happen during the 
   activation of LSPs in an GMPLS network; to identify some common 
   causes of misconnections; and, to propose some initial solutions for 
   avoiding or minimizing misconnections." 
    
   The purpose of this document, therefore, is to focus on 
   misconnections, and discuss when they may occur and evaluate some of 
   their common causes.  The goal is also to propose some initial 
   solutions to address the problem of misconnections. 
    

     
   Shiomoto/Sharma/Suemura    Expires December 2004                  2 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
4. Analysis of Misconnection Scenarios 
    
   Misconnections in an MPLS/GMPLS network may happen in a variety of 
   circumstances. In the following sections, we focus on three cases of 
   misconnections that we have identified during discussions on the 
   CCAMP WG mailing list. These include: 
   i)   Misconnections during the activation of a backup LSP in a shared 
        mesh or 1:1 restoration scenario. 
   ii)  Misconnections during the activation of a high-priority LSP in 
        favor of a low-priority one. 
   iii) Misconnections due to a failure at a node or a switch. 
 
    
4.1. Activating Backup LSPs in Shared Mesh/1:1 Restoration  
    
   A misconnection may occur in several ways during the activation of a 
   backup LSP in shared mesh restoration [3], [4], [5]. 
    
   We discuss three cases below with an example setup. 
    
   Consider the following network scenario: 
    
        A-----------B 
         \         / 
          C-------D 
         /         \ 
        E           F 
    
   Where A-B:     Path of the Primary LSP 
         A-C-D-B: Path of the Secondary LSP 
         E-C-D-F: Path of an Extra (preemptible) LSP 
    
   Case 1: 
    
   Assume that an LSP is always activated upon the receipt of a Path 
   message, and assume that initially the primary LSP A-B, and the 
   extra-traffic LSP E-C-D-F are active. Upon a failure in the primary 
   LSP, it is necessary to activate the secondary LSP A-C-D-B. 
    
   We consider what happens at node C upon the arrival of a Path message 
   (that signals the secondary LSP A-C-D-B) carrying a Protection Object 
   [2]. If node C immediately reconfigures its cross-connect to connect 
   A-C to C-D, then, for a short period of time (until node D receives 
   the Path message and reconfigures its cross-connect to connect C-D to 
   D-B), it is possible for traffic from A to end-up at F via the path 
   A-C-D-F, thus leading to a misconnection. 
    
   One way to eliminate this would be to have a two-way activation 
   scheme that makes use of both the Path and the Resv messages. For 
   example, the extra LSP E-C-D-F is deleted upon the receipt of a Path 
   message (that corresponds to the secondary LSP) at a node, while the 
   secondary LSP A-C-D-B is activated upon the receipt of a Resv message 
   (that corresponds, as before, to the secondary LSP) at a node. Thus, 
     
   Shiomoto/Sharma/Suemura    Expires December 2004                  3 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
   in the example above, upon the detection of a failure on the primary 
   LSP A-B, a Path message with the Protection Object is sent along the 
   secondary LSP. Upon receipt of this message, node C deletes the 
   extra-LSP (removing state for it both from the control plane and the 
   data plane), and activates the secondary LSP when it receives a Resv 
   message for the secondary LSP. 
    
   However, even in this case a misconnection may happen as follows. 
    
    
    
        A-----------B 
         \         / 
          C-------D 
         /         \ 
        E           F 
    
   Where A-B:     Path of the Primary LSP 
         A-C-D-B: Path of the Secondary LSP 
    E-C-D-F: Path of an Extra (preemptible) LSP  
         
    
   Case 2: 
    
   Assume that upon the receipt of a Path message(containing the 
   Protection Object) at node C, corresponding to the secondary LSP, the 
   cross-connect for the extra LSP is deleted.(Recall that both Path and 
   Resv messages must be  periodically transmitted along the secondary 
   LSP to maintain the state of the provisioned secondary LSP.) 
    
   If node C now receives a Resv message from node D, it may set up the 
   cross-connect for the secondary LSP, thus connecting A-C to C-D. This 
   could happen, for example, when node C has no way of distinguishing a 
   refreshing Resv message (corresponding to the  provisionedsecondary 
   LSP) whose purpose is merely to maintain state for this LSP from an 
   activating Resv message (corresponding to the secondary LSP) whose 
   purpose is to actually activate the secondary LSP, by causing 
   intermediate nodes to reconfigure their cross-connects. Therefore, if 
   the cross-connect at  node C for the secondary LSP is (erroneously) 
   made before node D has had a chance to delete the cross-connect 
   corresponding to the extra LSP, there is a misconnection, with 
   traffic from node A  ending up at node F via the path A-C-D-F. 
 
   A possible solution to this would be to also carry the Protection 
   object in the Resv message that is sent to activate the secondary 
   LSP, thus allowing a node to distinguish between a Resv message for 
   activation from a refresh Resv message, say via the S bit in the 
   Protection object. 
    
   Even when the above changes to the Resv object and the processing 
   rules are made, there remains the possibility of misconnections 
   because the nodes may not consistently bring down the state of the 
   extra- LSP. We explain this next. 
     
   Shiomoto/Sharma/Suemura    Expires December 2004                  4 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
    
        A-----------B 
         \         / 
          C-------D 
         /         \ 
        E           F 
    
   Where A-B:     Primary LSP 
         A-C-D-B: Secondary LSP 
         E-C-D-F: Extra (preemptible) LSP 
    
   Case 3:  
    
   Consider again the earlier network example, where we assume that the 
   Resv message is enhanced to carry the Protection object, and that the 
   cross-connections for the secondary LSP at a node are made upon the 
   receipt of the Resv message for activation. 
    
   Note that because of the way vendors interpret and implement rules 
   for hard/soft pre-emption of LSPs, it is possible that the extra LSP 
   is not brought down upon the receipt of the Path message for the 
   secondary LSP. 
    
   Thus, it is possible that node C, upon receipt of a Path message for 
   the secondary LSP, does not immediately delete the extra LSP. 
    
   Suppose node D, upon the receipt of a Resv message with a Protection 
   Object for the secondary LSP, brings down the extra LSP, and 
   activates its cross-connect for the secondary LSP, that is, it 
   reconfigures its cross-connect to switch from C-D -> D-F to C-D -> D-
   B. However, since node C may not have deleted its cross-connect 
   corresponding to the extra LSP, traffic may now be misconnected from 
   node E to node B via the path E-C-D-B. 
    
   This may be prevented by enforcing a rule that an intermediate node 
   along the route of a secondary LSP must remove the state, in control 
   plane and data planes, associated with the extra LSP (that uses 
   resources required by the secondary LSP) *before* forwarding the Path 
   message for the secondary LSP. 
    
    
4.2. Activating High-priority versus low-priority LSPs 
    
   The situations that lead to misconnections during the activation of 
   LSPs with different priorities are almost analogous to those that 
   arise in the shared mesh restoration case for activation of the 
   secondary LSP in place of the extra LSP. In fact, we have three cases 
   here that parallel those discussed in Section 4.1. 
    
   Consider the following network scenario: 
    
        A-----------B 
         \         / 
     
   Shiomoto/Sharma/Suemura    Expires December 2004                  5 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
          C-------D 
         /         \ 
        E           F 
    
   Where  
         A-C-D-B: High-priority LSP 
         E-C-D-F: Low-priority LSP 
    
   Assume that initially LSP E-C-D-F is active, and that there is then a 
   need to set up the higher priority LSP A-C-D-B. 
    
   Case 1: 
    
   Assume that an LSP is always activated upon the receipt of a Path 
   message, and that initially the low-priority LSP E-C-D-F is active. 
   At some time, it becomes necessary to activate the high-priority LSP 
   A-C-D-B. 
    
   We consider what happens at C upon the arrival of Path message (that 
   signals the high-priority LSP A-C-D-B). If node C immediately 
   reconfigures its cross-connect to connect A-C to C-D, then, for a 
   short period of time (until node D receives the Path message and 
   reconfigures its cross-connect to connect C-D to D-B), it is possible 
   for traffic from A to end-up at F via the path A-C-D-F, thus leading 
   to a misconnection. 
    
   A possible solution to this is to, as before, use a two-way 
   activation scheme that uses both the Path and the Resv messages with 
   a way to distinguish between Resv messages that refresh the state of 
   an LSP from Resv messages that are supposed to activate the 
   configuration of a cross-connect for an LSP. 
    
    
    
        A-----------B 
         \         / 
          C-------D 
         /         \ 
        E           F 
    
   Where  
         A-C-D-B: High-priority LSP 
         E-C-D-F: Low-priority LSP 
   Case 2: 
    
   Consider again the earlier network example, where we assume that the 
   Resv message is enhanced to distinguish a refresh Resv message from a 
   Resv message for activation of an LSP. In other words, the cross-
   connections for an LSP at a node are made upon the receipt of the 
   Resv message for its activation. 
    
   Note that because of the way vendors interpret and implement rules 
   for hard/soft pre-emption of LSPs , it is possible that the low-
     
   Shiomoto/Sharma/Suemura    Expires December 2004                  6 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
   priority LSP is not brought down upon the receipt of the Path message 
   for the high-priority LSP. 
    
   Thus, it is possible that node C, upon receipt of a Path message for 
   the high-priority LSP A-C-D-B, does not immediately delete the low 
   priority LSP E-C-D-F. 
    
   If node D, upon the receipt of a Resv message for the high-priority 
   LSP, brings down the low-priority LSP, and activates its cross-
   connect for the high priority LSP, that is, it reconfigures its 
   cross-connect to switch from C-D -> D-F to C-D -> D-B, then we could 
   have a misconnection. This is because node C may not have deleted its 
   cross-connect corresponding to the low priority LSP, so traffic may 
   now be misconnected from E to B via the path E-C-D-B. 
    
   This may be prevented by enforcing a rule that an intermediate node 
   along the route of a high-priority LSP must remove the state, in the 
   control plane and data planes, associated with a low-priority LSP 
   (that uses resources required by the high-priority LSP) before 
   forwarding the Path message for the high-priority LSP. 
    
    
4.3. Failure at a switch or node 
    
   Another situation when misconnections may arise is when a node or a 
   switch malfunctions. The nature of malfunctions could span the 
   control plane or the data plane. 
    
   Let us consider again the network scenario from Section 4.1. 
    
        A-----------B 
         \         / 
          C-------D 
         /         \ 
        E           F 
    
   where A-B:     Primary LSP 
         A-C-D-B: Secondary LSP 
         E-C-D-F: Extra (preemptible) LSP 
    
   Suppose that we follow the rule of activating the secondary LSP upon 
   the receipt of a Resv message. 
    
   If there is a malfunction at node C, such that upon the receipt of a 
   Path message for the secondary LSP, it deletes the state of the extra 
   LSP from its control plane but fails to alter its cross-connect in 
   the data plane, thus allowing E-C to remain connected to C-D, we can 
   have a misconnection as follows. 
    
   Node D, upon receiving a Resv message (with a Protection object) from 
   node B, correctly switches its cross-connect from C-D  -> D-F to C-D 
   -> D-B. However, since the cross-connect at C has not been altered, 
   traffic may now flow incorrectly from E to B via the path E-C-D-B. 
     
   Shiomoto/Sharma/Suemura    Expires December 2004                  7 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
    
   If the control plane at C is somehow disconnected from its data 
   plane, with neither plane having stopped working, this situation 
   could continue indefinitely. It would then be necessary to have a way 
   to quickly detect and recover from such a misconnection scenario.  
    
5. Common Causes of Misconnections 
    
   When examining the common causes of misconnections, there are two 
   issues to keep in mind: 
    
        - One issue is when (in time) a cross-connection is made. 
        - Second issue is the signaling method and rules used to drop 
          an extra LSP and activate a secondary LSP (in the restoration 
          case), or to activate a high-priority LSP and deactivate a 
          low-priority LSP. 
        -  
   In order to analyze the common causes of misconnections, we 
   distinguish between the logical association of an incoming port with 
   an outgoing port from the physical association.  
    
   The logical association is an association between the incoming and 
   outgoing ports that is stored in the MPLS/GMPLS controller software. 
   On the other hand, the physical association is the actual association 
   between the incoming and the outgoing ports. When a physical 
   association between two ports is made, the data traffic is 
   immediately carried from the incoming port to the outgoing port over 
   the cross-connection made by the physical association. 
    
   The case where low-priority traffic is preempted by high-priority 
   traffic is explained by noting the difference between the logical 
   association and the physical one. Suppose the low-priority traffic 
   has a logical association between an incoming port (say, port-i1) and 
   an outgoing port (say, port-o1), while the high-priority traffic has 
   a logical association between an incoming port (say, port-i2) and the 
   outgoing port (say, port-o1). The outgoing port (port-o1) is shared. 
   The logical association for the traffic is maintained by the 
   signaling protocol. If both logical associations exist, the logical 
   association for the high priority traffic is selected to make the 
   cross-connection and realize a physical association between port i2 
   and port o1. If a logical association for only the low priority 
   traffic exists, it is selected to make the cross-connection. That is, 
   a physical association is made between port i1 and port o1. 
    
   A common cause of misconnection lies in the situation where the 
   priority traffic selected to make the cross-connection is 
   inconsistently selected among the nodes along with the route.  
    
   Here we define the possible states for the cross-connection as:  
   (1) high,  
   (2) low, and  
   (3) null.  
    
     
   Shiomoto/Sharma/Suemura    Expires December 2004                  8 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
   High and low states indicate that the cross-connection is made 
   according to the logical association of the high and low priority 
   traffic, respectively. Null state indicates that no cross-connection 
   is made (i.e., no input and output ports associated with low and high 
   priority traffic are connected). If we have both high and low state 
   for the cross-connections along with the route simultaneously, we 
   could have a misconnection. If we have only high and/or null states 
   for the cross-connections, we cannot have a misconnection. Similarly, 
   if we have only low and/or null state for the cross-connections, we 
   cannot have misconnections, either. In order to avoid misconnections, 
   we need a mechanism to have mutually exclusive high (including null) 
   or low (including null) state for the cross-connections along with 
   the route. 
    
   We observe, however, that the use of the null state may not always be 
   possible in optical switches, such as MEMS switches, which do not 
   allow for a null state. This is because a MEMS switch changes cross-
   connection by tilting a mirror. So, an input port is always 
   connected to one of the output ports. This results in the same 
   situation as described in Section 4.3. A possible solution to this 
   for the case of LSPs with multiple priorities is explained below. 
   The applicability of this solution to the restoration case (or the 
   development of other solutions to address this problem in the 
   restoration case) requires further study. 
    
    
6. Possible Solutions to Handle Misconnections 
    
   In order to have mutually exclusive high (including null) or low 
   (including null) state for the cross-connections along the route, one 
   may use a two-way activation approach. A Path message for the high 
   priority LSP activation is used to make the cross-connection state 
   æænullÆÆ while a Resv message for the high priority activation is used 
   to make the cross-connection state ææhighÆÆ.  
    
   In order to distinguish the refresh message from the activation one 
   for high priority traffic, we can use the Admin Status object. 
   Another option is to use the Protection object in the Resv message.  
   Finally, to solve the possible misconnection problem in optical 
   switches that do not allow for a null state, the source node must 
   wait for the high-priority LSP to be fully established before sending 
   out data on it. In other words, the initiator/ingress node sends out 
   data after receiving the RESV message for activation. For triggering 
   the terminator/egress node to send out data, a ResvConf message 
   should be sent from the initiator/ingress, and received by the 
   terminator/egress, before it begins transmitting data on the high-
   priority LSP. 
 
    
7. Conclusions 
    
   This document provided an initial discussion of misconnection 
   scenarios in GMPLS networks, analyzes some common scenarios, outlines 
     
   Shiomoto/Sharma/Suemura    Expires December 2004                  9 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
   some common causes of these, and then posits some alternative ways of 
   handling misconnections.  
    
8. Security Considerations 
    
  This document does not introduce any new security considerations in 
  the protocols considered herein. 
 
9. References 
    
9.1. Normative References 
    
    [1] RFC 2119 Bradner, S., "Key words for use in RFCs to Indicate 
        Requirement Levels", BCP 14, RFC 2119, March 1997.  
     
    [2] L. Berger (Ed.), "Generalize Multi-Protocol Label Switching 
        (GMPLS) Signaling Resource Reservation Protocol-Traffic 
        Engineering Extensions" RFC 3473, January 2003. 
 
9.2. Informative References 
 
    [3] D. Papadimitriou, et al, "Analysis of Generalized Multi-Protocol 
        Label Switching-based Recovery Mechanisms", Internet Draft, Work 
        in Progress, draft-ietf-ccamp-gmpls-recovery-analysis-03.txt, 
        April 2003. 
 
    [4] J. P. Lang, D. Papadimitriou, Y. Rekhter (Eds.),  "RSVP-TE 
        Extensions in Support of End-to-End GMPLS-Based Recovery", 
        Internet Draft, Work in Progress, draft-ietf-ccamp-gmpls-
        recovery-e2e-signaling-02.txt, May 2003. 
 
    [5] A. Farrel, "Multi-protocol Label Switching Pre-emption" Internet 
        Draft, Work in Progress, draft-farrel-mpls-preemption-01.txt, 
        April 2004 
 
 
10.     Author's Addresses 
    
   Kohei Shiomoto                          Vishal Sharma 
   NTT Corporation                         Metanoia, Inc. 
   9-11 Midori-Cho 3-Chome                 888 Villa St., Suite 200B 
   Musashino-Shi, Tokyo 180-8585, Japan    Mountain View, CA 94041, USA 
   Phone: +81 (422) 59 4402                 Phone: +1 408 530 8313 
   Email: Shiomoto.Kohei@lab.ntt.co.jp     Email: v.sharma@ieee.org 
                                            
   Yoshihiko Suemura                        
   NEC Corporation                          
   1753, Shimonumabe, Nakahara-ku           
   Kawasaki 211-8666, Japan                 
   Phone: +81 44 396 2804                   
   Email: y-suemura@bp.jp.nec.com           

     
   Shiomoto/Sharma/Suemura    Expires December 2004                 10 
                  Analysis of Misconnection Scenarios        June 2004 
                          in GMPLS Networks 
    
    
  11.     Intellectual Property Considerations 
   The IETF takes no position regarding the validity or scope of any 
   Intellectual Property Rights or other rights that might be claimed to 
   pertain to the implementation or use of the technology described in 
   this document or the extent to which any license under such rights 
   might or might not be available; nor does it represent that it has 
   made any independent effort to identify any such rights. Information 
   on the procedures with respect to rights in RFC documents can be 
   found in BCP 78 and BCP 79. 
    
   Copies of IPR disclosures made to the IETF Secretariat and any 
   assurances of licenses to be made available, or the result of an 
   attempt made to obtain a general license or permission for the use of 
   such proprietary rights by implementers or users of this 
   specification can be obtained from the IETF on-line IPR repository at 
   http://www.ietf.org/ipr. 
    
   The IETF invites any interested party to bring to its attention any 
   copyrights, patents or patent applications, or other proprietary 
   rights that may cover technology that may be required to implement 
   this standard. Please address the information to the IETF at ietf-
    ipr@ietf.org. 
 11.1. IPR Disclosure Acknowledgement 
   By submitting this Internet-Draft, I certify that any applicable 
   patent or other IPR claims of which I am aware have been disclosed, 
   and any of which I become aware will be disclosed, in accordance 
   with RFC 3668. 
    
12.     Full Copyright Statement 
    
   "Copyright (C) The Internet Society (2004).  This document is subject 
   to the rights, licenses and restrictions contained in BCP 78, and 
   except as set forth therein, the authors retain all their rights." 
    
   "This document and the information contained herein are provided on 
   an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE 
   REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE 
   INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR 
   IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE Of 
   THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 
   WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE." 
    
 








      
   Shiomoto/Sharma/Suemura    Expires December 2004                 11 
 

PAFTECH AB 2003-20262026-04-23 12:32:24