RSVP-TE Signaling Procedure for End-to-End GMPLS Restoration and Resource Sharing
draft-ietf-teas-gmpls-resource-sharing-proc-01

TEAS Working Group                                            Xian Zhang
Internet-Draft                                        Haomian Zheng, Ed.
Intended Status: Informational                                    Huawei
Expires: August 3, 2015                               Rakesh Gandhi, Ed.
                                                               Zafar Ali
                                               Gabriele Maria Galimberti
                                                     Cisco Systems, Inc.
                                                        Pawel Brzozowski
                                                            ADVA Optical
                                                        January 30, 2015

    RSVP-TE Signaling Procedure for End-to-End GMPLS Restoration and
                             Resource Sharing
             draft-ietf-teas-gmpls-resource-sharing-proc-01

Abstract

   In transport networks, there are requirements where Generalized
   Multi-Protocol Label Switching (GMPLS) end-to-end recovery scheme
   needs to employ restoration Label Switched Path (LSP) while keeping
   resources for the working and/or protecting LSPs reserved in the
   network after the failure occurs.

   This document reviews how the LSP association is to be provided using
   Resource Reservation Protocol - Traffic Engineering (RSVP-TE)
   signaling in the context of GMPLS end-to-end recovery scheme when
   using restoration LSP where failed LSP is not torn down.  In
   addition, this document clarifies the RSVP-TE signaling procedure to
   support resource sharing-based setup and teardown of LSPs as well as
   LSP reversion.  No new extensions are defined by this document, and
   it is strictly informative in nature.

Status of this Memo

   This Internet-Draft is submitted to IETF in full conformance with the
   provisions of BCP 78 and BCP 79.

   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."

 

Zhang, et al             Expires August 3, 2015                 [Page 1]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

   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.

Copyright Notice

   Copyright (c) 2015 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document. Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document. Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents

   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
   2.  Overview . . . . . . . . . . . . . . . . . . . . . . . . . . .  4
     2.1.  1+R Restoration  . . . . . . . . . . . . . . . . . . . . .  4
     2.2.  1+1+R Restoration  . . . . . . . . . . . . . . . . . . . .  5
     2.3.  Resource Sharing By Restoration LSP  . . . . . . . . . . .  6
   3.  RSVP-TE Signaling Procedure  . . . . . . . . . . . . . . . . .  6
     3.1.  Restoration LSP Association  . . . . . . . . . . . . . . .  6
     3.2.  Resource Sharing-based Restoration LSP Setup . . . . . . .  7
     3.3.  LSP Reversion  . . . . . . . . . . . . . . . . . . . . . .  8
       3.3.1.  Make-while-break Reversion . . . . . . . . . . . . . .  8
       3.3.2.  Make-before-break Reversion  . . . . . . . . . . . . .  9
   4.  Security Considerations  . . . . . . . . . . . . . . . . . . . 10
   5.  IANA Considerations  . . . . . . . . . . . . . . . . . . . . . 11
   6.  Acknowledgement  . . . . . . . . . . . . . . . . . . . . . . . 11
   7.  References . . . . . . . . . . . . . . . . . . . . . . . . . . 12
     7.1.  Normative References . . . . . . . . . . . . . . . . . . . 12
     7.2.  Informative References . . . . . . . . . . . . . . . . . . 12
   8.  Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 13

 

Zhang, et al             Expires August 3, 2015                 [Page 2]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

1.  Introduction

   Generalized Multi-Protocol Label Switching (GMPLS) [RFC3945] defines
   a set of protocols, including Open Shortest Path First - Traffic
   Engineering (OSPF-TE) [RFC4203] and Resource ReserVation Protocol -
   Traffic Engineering (RSVP-TE) [RFC3473].  These protocols can be used
   to setup Label Switched Paths (LSPs) in transport networks.  The
   GMPLS protocol extends MPLS to support interfaces capable of Time
   Division Multiplexing (TDM), Lambda Switching and Fiber Switching. 
   These switching technologies provide several protection schemes
   [RFC4426][RFC4427] (e.g., 1+1, 1:N and M:N).

   Resource Reservation Protocol - Traffic Engineering (RSVP-TE)
   signaling has been extended to support various GMPLS recovery
   schemes, such as end-to-end recovery [RFC4872] and segment recovery
   [RFC4873].  As described in [RFC6689], ASSOCIATION object can be used
   to identify the LSPs for restoration using Association Type set to
   "Recovery" [RFC4872].  [RFC6689] Section 2.2 reviews the procedure
   for providing LSP associations for GMPLS end-to-end recovery and
   covers the schemes where the failed working LSP and/or protecting LSP
   are torn down.

   In GMPLS end-to-end recovery schemes generally considered,
   restoration LSP is signaled after the failure has been detected and
   notified on the working LSP.  For revertive recovery mode, a
   restoration LSP is signaled while working LSP and/or protecting LSP
   are not torn down in control plane due to a failure.  In transport
   networks, as working LSPs are typically signaled over a nominal path,
   service providers would like to keep resources associated with the
   working LSPs reserved.  This is to make sure that the service
   (working LSP) can be reverted to the nominal path when the failure is
   repaired to provide deterministic behavior and guaranteed Service
   Level Agreement (SLA).

   Following behaviors are not fully documented in the existing
   standards for LSP associations, resource sharing based LSP setup,
   teardown and LSP reversion in transport networks:

   o  The procedure for providing LSP associations for the GMPLS
   recovery using restoration LSP where working and protecting LSPs are
   not torn down after the failure is not clearly documented. 

   o  In [RFC3209], the MBB method assumes the old and new LSPs share
   the SESSION object and signal Shared Explicit (SE) flag in
   SESSION_ATTRIBUTE object.  According to [RFC6689], ASSOCIATION object
   with Association Type "Resource sharing" enables the sharing of
   resources across LSPs with different SESSION objects.  However,
   existing documents do not mention the usage of SE flag for resource
 

Zhang, et al             Expires August 3, 2015                 [Page 3]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

   sharing with ASSOCIATION object. 

   o  As described in [RFC3209], Section 2.5, the purpose of make before
   break (MBB) is "not to disrupt traffic, or adversely impact network
   operations while TE tunnel rerouting is in progress".  In transport
   networks, the label has a mapping into the data plane resource used
   and the nodes along the LSP need to send triggering commands to data
   plane for setting up cross-connections accordingly during the RSVP-TE
   signaling procedure.  Due to the nature of transport networks, node
   may not be able to fulfill this purpose when sharing resources in
   some scenarios.   

   o  When using end-to-end recovery with revertive mode, methods for 
   LSP reversion and resource sharing have not been described. 

   This document reviews how the LSP association is to be provided for
   GMPLS end-to-end recovery when using restoration LSP where working
   and protecting LSP resources are kept reserved in the network after
   the failure.  In addition, this document clarifies the signaling
   procedure for sharing resources during setup and teardown of LSPs as
   well as LSP reversion.  This document is strictly informative in
   nature and does not define any RSVP-TE signaling extensions.

2.  Overview

   The GMPLS end-to-end recovery scheme, as defined in [RFC4872] and
   being considered in this document, "fully dynamic rerouting switches
   normal traffic to an alternate LSP that is not even partially
   established only after the working LSP failure occurs.  The new
   alternate route is selected at the LSP head-end node, it may reuse
   resources of the failed LSP at intermediate nodes and may include
   additional intermediate nodes and/or links".  Two examples, 1+R and
   1+1+R are described in the following sections.

2.1.  1+R Restoration

   One example of the recovery scheme considered in this document is 1+R
   recovery.  The 1+R recovery is exemplified in Figure 1.  In this
   example, working LSP on path A-B-C-Z is pre-established.  Typically
   after a failure detection and notification on the working LSP, a
   second LSP on path A-H-I-J-Z is established as a restoration LSP.
   Unlike protection LSP, restoration LSP is signaled per need basis.

 

Zhang, et al             Expires August 3, 2015                 [Page 4]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

          +-----+    +-----+     +-----+     +-----+
          |  A  +----+  B  +-----+  C  +-----+  Z  |
          +--+--+    +-----+     +-----+     +--+--+
              \                                /
               \                              /
             +--+--+       +-----+        +--+--+
             |  H  +-------+  I  +--------+  J  |
             +-----+       +-----+        +-----+

          Figure 1: An Example of 1+R Recovery Scheme

   During failure switchover with 1+R recovery scheme, in general,
   working LSP resources are not released so that working and
   restoration LSPs coexist in the network.  Nonetheless, working and
   restoration LSPs can share network resources.  Typically when failure
   is recovered on the working LSP, restoration LSP is no longer
   required and torn down, while the traffic is reverted to the original
   working LSP.

2.2.  1+1+R Restoration

   Another example of the recovery scheme considered in this document is
   1+1+R.  In 1+1+R, a restoration LSP is signaled for the working LSP
   and/or the protecting LSP after the failure has been detected, and
   this recovery is exemplified in Figure 2.  

             +-----+       +-----+        +-----+
             |  D  +-------+  E  +--------+  F  |
             +--+--+       +-----+        +--+--+
               /                              \
              /                                \
          +--+--+    +-----+     +-----+     +--+--+
          |  A  +----+  B  +-----+  C  +-----+  Z  |
          +--+--+    +-----+     +-----+     +--+--+
              \                                /
               \                              /
             +--+--+       +-----+        +--+--+
             |  H  +-------+  I  +--------+  J  |
             +-----+       +-----+        +-----+

          Figure 2: An Example of 1+1+R Recovery Scheme

   In this example, working LSP on path A-B-C-Z and protecting LSP on
   path A-D-E-F-Z are pre-established.  After a failure detection and
   notification on a working LSP or protecting LSP, a third LSP on path
   A-H-I-J-Z is established as a restoration LSP.  The restoration LSP
   in this case provides protection against a second order failure. 
 

Zhang, et al             Expires August 3, 2015                 [Page 5]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

   During failure switchover with 1+1+R recovery scheme, in general,
   failed LSP resources are not released so that working, protecting and
   restoration LSPs coexist in the network.  Nonetheless, restoration
   LSP with working LSP it is restoring as well as restoration LSP with
   protecting LSP it is restoring can share network resources. 
   Typically, restoration LSP is torn down when the failure on the
   working or protecting LSP is repaired and while the traffic is
   reverted to the original LSP.

2.3.  Resource Sharing By Restoration LSP

                               +-----+      +-----+
                               |  F  +------+  G  +--------+
                               +--+--+      +-----+        |
                                  |                        |
                                  |                        |
        +-----+    +-----+     +--+--+      +-----+     +--+--+
        |  A  +----+  B  +-----+  C  +--X---+  D  +-----+  E  |
        +-----+    +-----+     +-----+      +-----+     +-----+

          Figure 3: Resource Sharing in 1+R Recovery Scheme

   Using the network shown in Figure 3 as an example, LSP1 (A-B-C-D-E)
   is the working LSP and it allows for resource sharing when the LSP is
   dynamically rerouted due to link failure.  Upon detecting the failure
   of a link along the LSP1, e.g. Link C-D, node A needs to decide which
   alternative path it will use to signal restoration LSP and reroute
   traffic.  In this case, A-B-C-F-G-E is chosen as the restoration LSP
   path and the resources on the path segment A-B-C are re-used by this
   LSP when working LSP is not torn down as in 1+R recovery scheme.

3.  RSVP-TE Signaling Procedure 

3.1.  Restoration LSP Association

   Where GMPLS end-to-end recovery scheme needs to employ restoration
   LSP while keeping resources for the working and/or protecting LSPs
   reserved in the network after the failure, restoration LSP is
   signaled with ASSOCIATION object that has Association Type set to
   "Recovery" [RFC4872] with the association ID set to the LSP ID of the
   LSP it is restoring.  For example, when a restoration LSP is signaled
   for a working LSP, the ASSOCIATION object in the restoration LSP
   contains the association ID set to the LSP ID of the working LSP. 
   Similarly, when a restoration LSP is signaled for a protecting LSP,
   the ASSOCIATION object in the restoration LSP contains the
   association ID set to the LSP ID of the protecting LSP.

   The procedure for signaling the PROTECTION object is specified in
 

Zhang, et al             Expires August 3, 2015                 [Page 6]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

   [RFC4872].  Specifically, restoration LSP being used as a working LSP
   is signaled with P bit cleared and being used as a protecting LSP is
   signaled with P bit set.

3.2.  Resource Sharing-based Restoration LSP Setup

   GMPLS LSPs can share resources if they have Shared Explicit (SE) flag
   set in their SESSION_ATTRIBUTE objects and:

   o  As defined in [RFC3209], LSPs have identical SESSION objects
   and/or

   o  As defined in [RFC6689], LSPs have matching ASSOCIATION object
   with Association Type set to "Resource Sharing".  LSPs in this case
   can have different SESSION objects i.e. different tunnel ID, source
   and destination.

   For LSP restoration upon failure, as explained in Section 11 of
   [RFC4872], reroute procedure may re-use existing resources.  The
   behavior of the intermediate nodes during rerouting process to
   reconfigure cross-connections does not further impact the traffic
   since it has been interrupted due to the already failed LSP.

   The node behavior for setting up the restoration LSP can be
   categorized into the following three categories:

           Table 1: Node Behavior during Restoration LSP Setup
   ---------+---------------------------------------------------------
   Category |       Node Behavior during Restoration LSP Setup
   ---------+---------------------------------------------------------
      C1    + Reusing existing resource on both input and output
            + interfaces (node A & B in Figure 3).
            +
            + This type of nodes only needs to book the existing 
            + resources and no cross-connection setup 
            + command is needed.
   ---------+---------------------------------------------------------
      C2    + Reusing existing resource only on one of the interfaces,
            + either input or output interfaces and need to use new
            + resource on the other interface. (node C & E in Figure 3).
            +
            + This type of nodes needs to book the resources and send 
            + the re-configuration cross-connection command to its 
            + corresponding data plane node on the interfaces where new 
            + resources are needed and re-use the
            + existing resources on the other interfaces.
   ---------+---------------------------------------------------------
       C3   + Using new resources on both interfaces. 
 

Zhang, et al             Expires August 3, 2015                 [Page 7]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

            + (node F & G in Figure 3).
            +
            + This type of nodes needs to book the new resources 
            + and send the cross-connection setup 
            + command on both interfaces.
   ---------+---------------------------------------------------------

   Depending on whether the resource is re-used or not, the node
   behaviors differ.  This deviates from normal LSP setup since some
   nodes do not need to re-configure the cross-connection, and it should
   not be viewed as an error.  Also, the judgment whether the control
   plane node needs to send a cross-connection setup/modification
   command to its corresponding data plane node(s) relies on the check
   whether the LSPs are sharing resources. 

3.3.  LSP Reversion

   If the end-to-end LSP recovery is revertive, as described in Section
   2, traffic can be reverted from the restoration LSP to the working or
   protecting LSP after its failure is recovered.  The LSP reversion can
   be achieved using two methods:

     1. Make-while-break reversion, where resources associated with
     working or protecting LSP are reconfigured while removing
     reservations for the restoration LSP.

     2. Make-before-break reversion, where resources associated with
     working or protecting LSP are reconfigured before removing the
     restoration LSP.

   In transport networks, both of the above reversion methods will
   result in some traffic disruption when the restoration LSP and the
   LSP being restored are sharing resources and the cross-connections
   need to be reconfigured on intermediate nodes.

3.3.1.  Make-while-break Reversion

   In this reversion method, restoration LSP is simply requested to be
   deleted by the head-end.  Removing reservations for restoration LSP
   triggers reconfiguration of resources associated with working or
   protecting LSP on every node where resources are shared.  Whenever
   reservation for restoration LSP is removed from a node, data plane
   configuration changes to reflect reservations of working or
   protection LSP as signaling progresses.  Eventually, after the whole
   restoration LSP is deleted, data plane configuration will fully match
   working or protecting LSP reservations on the whole path.  Thus
   reversion is complete.

 

Zhang, et al             Expires August 3, 2015                 [Page 8]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

   Make-while-break, while being relatively simple in its logic, has few
   limitations as follows which may not be acceptable in some networks:

   o No rollback

   Deletion of restoration LSPs is not a revertive process.  If for some
   reason reconfiguration of data plane on one of the nodes to match
   working or protection LSP reservations fails, falling back to
   restoration LSP is no longer an option, as its state might have
   already been removed from other nodes.

   o No completion guarantee

   Deletion of an LSP provides no guarantees of completion.  In
   particular, if RSVP packets are lost due to nodal or DCN failures it
   is possible for an LSP to be only partially deleted.  To mitigate
   this, RSVP could maintain soft state reservations and hence
   eventually remove remaining reservations due to refresh timeouts. 
   This approach is not feasible in transport networks however, where
   control and data channels are often separated and hence soft state
   reservations are not useful.

   Finally, one could argue that graceful LSP deletion [RFC3473] would
   provide guarantee of completion.  While this is true for most cases,
   many implementations will time out graceful deletion if LSP is not
   removed within certain amount of time, e.g. due to a transit node
   fault.  After that, deletion procedures which provide no completion
   guarantees will be attempted.  Hence, in corner cases completion
   guarantee cannot be provided.

   o No explicit notification of completion to head-end node

   In some cases, it may be useful for a head-end node to know when the
   data plane has been reconfigured to match working or protection LSP
   reservations.  This knowledge could be used for initiating operations
   like enabling alarm monitoring, power equalization and others. 
   Unfortunately, for the reasons mentioned above, make-while-break
   reversion lacks such explicit notification.

3.3.2.  Make-before-break Reversion

   This reversion method can be used to overcome limitations of
   make-while-break reversion.  It is similar in spirit to MBB concept
   used for re-optimization.  Instead of relying on deletion of
   restoration LSP, head-end chooses to establish a new LSP to
   reconfigure resources on the working or protection LSP path, and uses
   identical ASSOCIATION and PROTECTION objects from the LSP it is
   replacing.  Only if setup of this LSP is successful will other
 

Zhang, et al             Expires August 3, 2015                 [Page 9]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

   (restoration and working/protecting) LSPs be deleted by the head-end.
    MBB reversion consists of two parts:

   A) Make part: 

   Creating a new reversion LSP following working or protection LSP's
   path.  Reversion LSP is sharing resources both with working and
   restoration LSPs.  As reversion LSP is created, resources are
   reconfigured to match its reservations.  Hence, after reversion LSP
   is created, data plane configuration essentially reflects working or
   protecting LSP reservations.

   B) Break part: 

   After "make" part is finished, working and restoration LSPs are torn
   down.  Removing reservations for working and restoration LSPs does
   not cause any resource reconfiguration on reversion LSP's path -
   nodes follow same procedures as for "break" part of any MBB
   operation.  Hence, after working and restoration LSPs are removed,
   data plane configuration is exactly the same as before starting
   restoration.  Thus reversion is complete.

   MBB reversion uses make-before-break characteristics to overcome
   challenges related to make-while-break reversion as follow:

   o Rollback

   If "make" part fails, (existing) restoration LSP will still be used
   to carry existing traffic.  Same logic applies here as for any MBB
   operation failure.

   o Completion guarantee

   LSP setup is resilient against RSVP message loss, as Path and Resv
   messages are refreshed periodically.  Hence, given that network
   recovers its DCN eventually, reversion LSP setup is guaranteed to
   finish with either success or failure.

   o Explicit notification of completion to head-end node

   Head-end knows that data plane has been reconfigured to match working
   or protection LSP reservations on intermediate nodes when it receives
   Resv for the reversion LSP.

4.  Security Considerations

   This document reviews procedures defined in [RFC3209] [RFC4872]
 

Zhang, et al             Expires August 3, 2015                [Page 10]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

   [RFC4873] and [RFC6689] and does not define any new procedure.  This
   document does not introduce any new security issues other than those
   already covered in [RFC3209] [RFC4872] [RFC4873] and [RFC6689].

5.  IANA Considerations

   This informational document does not make any request for IANA
   action.

6.  Acknowledgement

   The authors would like to thank George Swallow for the discussions on
   the GMPLS restoration.

 

Zhang, et al             Expires August 3, 2015                [Page 11]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

7.  References

7.1.  Normative References

   [RFC3209]   D. Awduche et al, "RSVP-TE: Extensions to RSVP for LSP
               Tunnels", RFC 3209, December 2001.

   [RFC4872]   J.P. Lang et al, "RSVP-TE Extensions in Support of
               End-to-End Generalized Multi-Protocol Label Switching
               (GMPLS) Recovery", RFC 4872, May 2007.

   [RFC4873]   L. Berger et al, "GMPLS Segment Recovery", RFC 4873, May
               2007.

   [RFC6689]   L. Berger, "Usage of the RSVP ASSOCIATION Object", RFC
               6689, July 2012.

7.2.  Informative References

   [PCEP-RSO]  X. Zhang, et al, "Extensions to Path Computation Element
               Protocol (PCEP) to Support Resource Sharing-based Path
               Computation", work in progress, February 2014.

   [RFC3473]   L. Berger, Ed., "Generalized Multi-Protocol Label
               Switching (GMPLS) Signaling Resource ReserVation
               Protocol-Traffic Engineering (RSVP-TE) Extensions", RFC
               3473, January 2003.

   [RFC3945]   Mannie, E., "Generalized Multi-Protocol Label Switching
               (GMPLS) Architecture", RFC 3945, October 2004.

   [RFC4203]   Kompella, K., and Rekhter, Y., "OSPF Extensions in
               Support of Generalized Multi-Protocol Label Switching
               (GMPLS)", RFC 4203, October 2005.

   [RFC4426]   Lang, J., Rajagopalan, B., and Papadimitriou, D.,
               "Generalized Multiprotocol Label Switching (GMPLS)
               Recovery Functional Specification", RFC 4426, March 2006.

   [RFC4427]   Mannie, E., and Papadimitriou, D., "Recovery (Protection
               and Restoration) Terminology for Generalized Multi-
               Protocol Label Switching", RFC 4427, March 2006.

 

Zhang, et al             Expires August 3, 2015                [Page 12]
Internet-Draft   GMPLS Restoration and Resource Sharing January 30, 2015

8.  Authors' Addresses

   Xian Zhang
   Huawei Technologies
   F3-1-B R&D Center, Huawei Base
   Bantian, Longgang District
   Shenzhen 518129 P.R.China

   Email: zhang.xian@huawei.com

   Haomian Zheng (editor)
   Huawei Technologies
   F3-1-B R&D Center, Huawei Base
   Bantian, Longgang District
   Shenzhen 518129 P.R.China

   Email: zhenghaomian@huawei.com

   Rakesh Gandhi (editor)
   Cisco Systems, Inc.

   Email: rgandhi@cisco.com

   Zafar Ali
   Cisco Systems, Inc.

   Email: zali@cisco.com

   Gabriele Maria Galimberti
   Cisco Systems, Inc.

   Email: ggalimbe@cisco.com

   Pawel Brzozowski
   ADVA Optical

   Email: PBrzozowski@advaoptical.com

Zhang, et al             Expires August 3, 2015                [Page 13]