One document matched: draft-karagiannis-ru-pdb-00.txt


TSVWG Working Group                                        
INTERNET-DRAFT                                    Georgios Karagiannis
                                       University of Twente / Ericsson
                                                           L. Westberg
                                                              A. Bader
                                                              Ericsson
                                                       
                                                      February 27, 2006

             Resource Unavailability (RU) Per Domain Behavior
                  draft-karagiannis-ru-pdb-00.txt


Status of this Memo
   By submitting this Internet-Draft, each author represents that any
   applicable patent or other IPR claims of which he or she is aware
   have been or will be disclosed, and any of which he or she becomes
   aware will be disclosed, in accordance with Section 6 of 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".

   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.

   This Internet-Draft will expire on September 1, 2006.

Copyright Notice

   Copyright (C) The Internet Society (2006). 


Karagiannis, et al.                                             [Page 1]

INTERNET-DRAFT                                                   RU-PDB


Abstract

   This draft specifies a Per Domain Behavior that provides the ability 
   to Difserv nodes located at the border or outside Diffserv domain(s) 
   to detect when the resources provided by the Diffserv domain(s) are 
   not available. This PDB is used when the negotiated SLS is 
   associated to throughput (or bandwidth) and when the SLS agreed 
   throughput bound is not static but loosely defined in order to allow 
   a more efficient utilization of the Diffserv domain(s) and a more 
   simple network management operation. This PDB can be applied in 
   association with either a single Diffserv domain or multiple 
   neighboring Diffserv domains. This specification is denoted as 
   Resource Unavailability (RU) PDB and it follows the guidelines given 
   in [RFC3086].


   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 [RFC2119].


Table of Contents

   1. Introduction. . . . .. . . . . . . . . . . . . . . . . . . . . . 3
   2.Applicability . . . . .. . . . . . . . . . . . . . . . . . . . . .3
   3.TCS and PHB configurations . . . . .. . . . . . . . . . . . . . . 3
   4.Attributes of this PDB . . . . .. . . . . . . . . . . . . . . . . 9
   5.Parameters . . . . .. . . . . . . . . . . . . . . . . . . . . . . 9
   6.Assumptions . . . . .. . . . . . . . . . . . . . . . . . . . . . .9
   7.Example uses . . . . .. . . . . . . . . . . . . . . . . . . . . .10
   8.Envinronmental concerns . . . . .. . . . . . . . . . . . . . . . 15
   9.Security considerations  . . . . .. . . . . . . . . . . . . . . .15
   10 IANA considerations. . .. . . . . . . . . . . . . . . . . . . . 15
   11.Acknowledgments . . . . .. . . . . . . . . . . . . . . . . . .  15
   12.Normative References . . . . .. . . . . . . . . . . . . . . . . 16
   13.Informative References . . . . .. . . . . . . . . . . . . . . . 17


Karagiannis, et al.                                             [Page 2]

INTERNET-DRAFT                                                   RU-PDB


1. Introduction
 
   to be done

2.  Applicability

   The RU PDB can be applied in the situation that Difserv nodes 
   located at the border or outside Diffserv domain(s) must detect when 
   the resources provided by the Diffserv domain(s) are not available.
   
   This PDB is used when the negotiated SLS is associated to throughput 
   (or bandwidth) and when the SLS agreed throughput bound is not 
   static but loosely defined. The main purpose of loosely defining the 
   SLS throughput bound is to allow a more efficient utilization of the 
   Diffserv domain(s) and a more simple network management operation. 
   This PDB can be applied in association with either a single Diffserv 
   domain or multiple neighboring Diffserv domains. 
   
   The resource unavailability on the Diffserv nodes within the 
   Diffserv domain(s) can be detected using a DSCP remarking approach 
   where the remarking is proportional to the amount of unavailable 
   resources. 
   
   The Diffserv nodes located at the border or outside the Diffserv 
   domain(s) can then use the remarked DSCP packets to calculate the 
   percentage of throughput (or bandwidth) that does exceed the loosely 
   agreed SLS throughput bound.
   
   These nodes can then, in combination with the sender of the traffic, 
   supported by the Diffserv domain(s), reduce the generated throughput 
   until the loosely agreed SLS throughput bound is satisfied. Possible 
   applicability areas on using the remarked DSCP packets/bytes are 
   related to the support of final handling decisions on the admission 
   control and/or severe congestion handling process.
   
   
3. TCS and PHB configurations
   
 
   Packets using any PHB can receive the RU PDB treatment. Furthermore, 
   the RU PDB can be used in combination with any other defined PDB. 
   

Karagiannis, et al.                                             [Page 3]

INTERNET-DRAFT                                                   RU-PDB

   
3.1. Specific Diffserv end system or Ingress edge router 
   configurations
   
   The Diffserv source end systems or the ingress edges, which support 
   the RU PDB, ensure that the packets are being marked with the right 
   PHB. Note that the process of marking can be specified by another 
   PDB. In this text, for simplicity reasons, the PDB that is defining 
   the PHB marking is denoted as another_PDB. For each of the chosen 
   PHB, the TCS and PHB configurations associated with the RU PDB are 
   following the rules defined by another_PDB, which MAY use the 
   specifications provided in [RFC2474], [RFC2475], [RFC3246], 
   [RFC2597] and [RFC3290]. 
   
   
3.2 Common Diffserv end system, edge and interior router 
   configurations
   
   The Diffserv nodes, which are supporting the RU PDB, ensure that for 
   each of the chosen PHB, the TCS and PHB configurations are 
   following the rules that are defined by the another_PDB, see above. 

   The classification of packets SHOULD be based on either the DSCP or 
   on a combination of IP header fields including the DSCP.
   
   In addition, the edge and interior routers MUST support the DSCP 
   remarking feature that is remarking the original DSCP into one or 
   more locally specified DSCPs, which SHOULD be associated with the 
   same PHB. In the following text, the original DSCP is denoted as 
   original_DSCP and the locally specified DSCPs that receive the same 
   PHB are denoted as locali_DSCP, where i= 1, 2, 3. 
   
   In order to provide this feature the Diffserv nodes SHOULD support 
   the following functions that can be implemented in a similar way as 
   specified in [RFC3290]:
   
   Classify: The Diffserv node SHOULD be configured to consider that 
   the packets marked either with the original_DSCP or with the 
   locali_DSCPs SHOULD receive the same per hop behavior treatment. 
   However, packets that are marked with the different DSCPs, i.e., 
   original_DSCP, local1_DSCP, local2_DSCP, local3_DSCP MAY be 
   classified to enter a different queue per DSCP. This classification 
   can be accomplished by the Classify function. 
   
   The classification of packets SHOULD be based on either the DSCP or 
   on a combination of IP header fields including the DSCP.

Karagiannis, et al.                                             [Page 4]

INTERNET-DRAFT                                                   RU-PDB

   
   Meter + Action: Each Diffserv node SHOULD maintain a metering 
   function (Meter) that measures/counts the bandwidth used by packets 
   belonging to a certain DSCP, e.g., original_DSCP, local1_DSCP, 
   local2_DSCP, local3_DSCP. In addition to that, either one 
   preconfigured loosely agreed SLS throughput (bandwidth) threshold or 
   two preconfigured loosely agreed SLS throughput (bandwidth) 
   thresholds SHOULD be maintained by the Diffserv node. The exact 
   definition of how these throughput (bandwidth) thresholds are 
   communicated from the location of where the SLS parameters are 
   maintained up to the Diffserv nodes it is outside the scope of this 
   draft. In the following text we denote the above mentioned 
   thresholds as Threshold1 and Threshold2. Note that if both 
   thresholds are maintained by a Diffserv node then the following 
   inequality SHOULD be valid: 
   
   Threshold2 > Threshold1. 
   
   If only one bandwidth threshold is maintained by the Diffserv node, 
   then this bandwidth threshold can be used either for resource 
   unavailability or for an overload situation detection. When this 
   threshold is used for resource unavailability detection then this 
   threshold is denoted in this text as Threshold1. Otherwise, this 
   bandwidth threshold is used for overload situation detection and is 
   denoted in this text as Threshold2.
   
   
3.2.1 Resource unavailability detection threshold
   
   If the Meter detects that the measured/count throughput (bandwidth) 
   exceeds the resource unavailability detection threshold (Threshold1) 
   then this exceeded throughput (bandwidth) (packets/bytes) SHOULD be 
   remarked accordingly (remarking whole exceeded bandwidth) by the 
   Action functionality block into a locally predefined DSCP, e.g., 
   local1_DSCP. 
   
   
3.2.2 Overload situation detection threshold
   
   If the Meter detects that the measured/count throughput (bandwidth) 
   exceeds the overload situation detection threshold (Threshold2) then 
   this exceeded bandwidth (packets/bytes) SHOULD be proportionally 
   remarked by the Action functionality block into two locally 
   predefined DSCP, e.g., local2_DSCP and local3_DSCP. The exact 
   definition of this is outside the scope of this draft. However, the 
   following text describes an example of the meter and action 
   functionality specification.

Karagiannis, et al.                                             [Page 5]

INTERNET-DRAFT                                                   RU-PDB

   
   The Diffserv node estimates the average amount of incoming 
   throughput (bandwidth) by counting the number of bytes N that arrive 
   during a measurement interval of T seconds and dividing N by T. If 
   this value is higher than the pre-configured threshold, i.e., 
   Threshold2, the Diffserv node knows that an overload situation has 
   occurred and estimates the overload O as O = N/T - Threshold2, which 
   is the amount of incoming traffic above the Threshold2. Using the 
   encoded DSCP (local2_DSCP), the Diffserv node encodes the value of O 
   into the outgoing packets as follows. 
   
   Firstly, at the end of each measurement period of T seconds the 
   estimated overload O of that period is converted to an equivalent 
   number of bytes B using the following formula: B = O x T. 
   
   Secondly, the Diffserv node re-marks the DSCP field of a certain 
   number of packets with the encoded DSCP (local2_DSCP) such that the 
   total sum of packet sizes of all the packets with an encoded DSCP 
   (local2_DSCP) is equal to B. The rest of the outgoing packets are 
   marked with the affected DSCP (local3_DSCP). In this way, when an 
   egress node receives a packet with a re-marked DSCP field 
   (local2_DSCP or local3_DSCP), it knows that there is an overloaded 
   node on the path the packet followed.
   
   Note that the packets that are used as input for the above described 
   remarking procedure have been marked with either the original_DSCP 
   or local1_DSCP. 
   
   Note that this procedure can be applied on more than one neighboring 
   Diffserv domains. Therefore, it is possible that a marked packet can 
   be received by the edge router of a new neighboring Diffserv domain 
   (and thus new domain operator). The new Diffserv domain may use 
   another type of Diffserv remarking scheme. Thus the original_DSCP, 
   local1_DSCP, local2_DSCP and local3_DSCP may be remarked to other 
   DSCP. However, the semantics and relations between these DSCPs 
   MUST remain. In the following text we assume that the same DSCP 
   names are kept. 
   
   If a Diffsev node receives marked packets, with one of the 
   locali_DSCPs, it means that another Diffserv node, located upstream, 
   on the same communication path remarked packets. Consider that the 
   remarking of the local2_DSCP and local3_DSCP proportion is 
   accomplished as above. Furthermore, consider that the total number 
   of bytes, B, that represent the percentage of overload above 
   Threshold2 and have to be remarked with the encoded DSCP 
   (local2_DSCP) is calculated as explained above. Moreover, consider 
   that the number of bytes that are received by the current Diffserv 
   node and marked as encoded DSCP (local2_DSCP) are denoted as IR. 
   Note that these IR remarked bytes SHOULD be transmitted over the 
   output link of the current Diffserv node.

Karagiannis, et al.                                             [Page 6]

INTERNET-DRAFT                                                   RU-PDB

   
   Furthermore, consider that the number of bytes that have to be 
   remarked by the current Diffserv node using the encoded DSCP 
   (local2_DSCP) are denoted as CR, which can be calculated as follows:
   
   IF (B > IR) 
   THEN  
     CR = B- IR
   ELSE
     CR = 0
   
   The rest of the outgoing packets are marked with the affected DSCP 
   (local3_DSCP).
   
   
3.3 Diffserv nodes supporting final handling decisions on resource 
   unavailability and/or overload situation handling process 
   
   When a Diffserv node located at the boundary or outside Diffserv 
   domain(s) is used to provide final decisions on the resource 
   unavailability and/or overload handling process then the following 
   steps SHOULD be followed by this Diffserv node. Note that the 
   Diffserv node that provides final decisions on resource 
   unavailability is denoted in this text as Resource Unavailability 
   Handling Diffserv node. Furthermore, the Diffserv node that provides 
   final decisions on overload situation handling process is denoted in 
   this text as Overload Situation Handling Diffserv node.
   
   
3.3.1 Resource unavailability handling
   
   When the Diffserv node that is located at the border or outside the 
   Diffserv domain(s) provides final handling decisions on the resource 
   unavailability functionality, then the amount of the original 
   (original_DSCP) and remarked (local1_DSCP) packets is counted by 
   this node to provide the basis of call acceptance for new flows. If 
   this Resource Unavailability Handling Diffserv node supports a QoS 
   agent, then this QoS agent can be informed when a new request has to 
   be accepted or rejected. 
   
   The downstream packets marked with the local1_DSCP are remarked back 
   to the original_DSCP by the Admission control Diffserv node.

Karagiannis, et al.                                             [Page 7]

INTERNET-DRAFT                                                   RU-PDB

   
   
3.3.2 Overload situation handling
   
   When the Diffserv node located at the boundary or outside the 
   Diffserv domain(s) provides final handling decisions on the overload 
   situation handling process, then the following steps must be 
   followed. 
   
   When the first packet with a re-marked DSCP field (encoded 
   (local2_DSCP) or affected (local3_DSCP) is received by this node, it 
   starts two counters that count the number of received packets with 
   encoded (local2_DSCP) and affected (local3_DSCP) DSCPs, 
   respectively. The counting interval of these counters is set to the 
   same length as the measurement periods used on the other Diffserv 
   nodes (T). At the end of the counting interval, the Overload 
   Situation Handling Diffserv node calculates an overload value by 
   summing the packet sizes of all the received encoded (local2_DSCP) 
   packets and dividing this sum by the length of the counting 
   interval. Based on this overload value, the Overload Situation 
   Handling Diffserv node can then decide whether flow termination is 
   necessary. If flow termination is needed, the Overload Situation 
   Handling Diffserv node using information provided by the QoS agent 
   or a local policy, it can select flows to terminate from the flows 
   that received packets with the encoded (local2_DSCP) or affected 
   (local3_DSCP) DSCP, since these are the flows that experienced 
   overload situation. 
   
   By giving preference to lower priority flows during the selection 
   procedure, higher priority flows such as emergency calls can be 
   protected. This is done by first trying to clear all overload by 
   terminating only affected flows from the lowest priority class. If 
   this is not sufficient, affected flows from the next priority class 
   will be terminated, etc. The flows to be selected from a certain 
   priority class are chosen in such a way, that the total number of 
   flows to be terminated is kept as low as possible. This is 
   accomplished by using the well-known "best-fit first" heuristic. For 
   example, if the overload to be cleared is 200 Kbps and the choice is 
   between three flows with reserved bandwidths of respectively 60, 180 
   and 210 Kbps, then the 180 Kbps flow will be selected first. Once 
   the flows that are to be terminated have been selected, then the QoS 
   agent is notified of which flows can be terminated.
   
   The downstream packets marked with the local2_DSCP and local3_DSCP 
   are remarked back to the original_DSCP by this egress edge.

Karagiannis, et al.                                             [Page 8]

INTERNET-DRAFT                                                   RU-PDB

   
4. Attributes of this PDB
   
   The new attributes that are related to this PDB are related to the 
   agreed SLS throughput (bandwidth) bound (threshold). 
   
   Different agreed SLS throughput thresholds, see Section 3, might be 
   used. Each of these throughput (bandwidth) thresholds are compared 
   to the measure throughput.
   
5. Parameters
   
   The used parameter is the SLS throughput (or bandwidth).
   
6. Assumptions
   
   The negotiated SLA may include either one preconfigured loosely 
   agreed SLS throughput (bandwidth) threshold or two preconfigured 
   loosely agreed SLS throughput (bandwidth) thresholds (bound). It is 
   assumed that the network operator communicates these throughput 
   (bandwidth) thresholds from the location of where the SLS parameters 
   are maintained up to the Diffserv nodes within the Diffserv 
   domain(s).
   
   The RU PDB can be applied on more than one neighboring Diffserv 
   domains. Therefore, it is possible that that a marked packet can be 
   received by the edge router of a new neighboring Diffserv domain 
   (and thus new domain operator). The new Diffserv domain may use 
   another type of Diffserv remarking scheme. Thus the original_DSCP, 
   local1_DSCP, local2_DSCP and local3_DSCP may be remarked to other 
   DSCP. However, the network operator MUST configure the Diffserv 
   remarking scheme such that the semantics and relations between the 
   original_DSCP, local1_DSCP, local2_DSCP and local3_DSCP remain even 
   when packets using the RU PDB are passing via multiple neighboring 
   Diffserv domains.
   
   Furthermore, a network operator may configure Diffserv nodes located 
   at the boundary or outside Diffserv domain(s) to provide final 
   handling decisions on the resource unavailability and/or overload 
   situation process, see section 3.3. 
   
   A network operator may configure the QoS Agent functionality (see 
   e.g., [RFC3290]) that may be used by the diffserv nodes located at 
   the boundary or outside Diffserv domain(s) to provide final handling 
   decisions on the resource unavailability and/or overload situation 
   process. The QoS Agent functionality can support a signaling 
   protocol, e.g., RSVP [RFC2205], or NSIS [RFC4080].
   

Karagiannis, et al.                                             [Page 9]

INTERNET-DRAFT                                                   RU-PDB

   
7. Example uses
   
   
7.1 Single Diffserv domain admission control based on probing: 
    controlled by egress
   
   In this scenario the RU PDB is applied only within one Diffserv 
   domain.
   
   The ingress router support the RU PDB functionality specified in 
   Section 3.1. The interior router supports the RU PDB functionality 
   as specified in section 3.2.1 and the egress edge supports the 
   functionality as specified in section 3.3.1. In addition to this is 
   is considered that each edge MAY maintain a QoS agent. 
   
   In this example, see Figure 1, a simple measurement-based admission 
   control within a Diffserv domain can be realised.  In the interior 
   nodes along the data path a threshold (Threshold1) is set in the 
   measurement based admission control function for the traffic 
   belonging to different PHBs. 
   
   When a probe packet (note that this packet MAY be sent by the QoS 
   Agent available at the Ingress as a signaling message), which is 
   marked with the original_DSCP arrives at the congested Interior node 
   (C), i.e., measured bandwidth is higher than Threshold2, then this 
   probe packet will be remarked to local1_DSCP. 
   
   When the egress receives this marked probe packet then it decides to 
   deny the request of the probe packet. If a QoS Agent is available at 
   the ingress and egress, then the QoS Agent at the egress SHOULD send 
   a notify like message, to the QoS Agent available at the ingress 
   edge to inform him about the denial of the probe packet request.
   
   The downstream packets marked with the local1_DSCP are remarked back 
   to the original_DSCP by this egress edge.

Karagiannis, et al.                                            [Page 10]

INTERNET-DRAFT                                                   RU-PDB

   
   
       Ingress            Interior          Interior            Egress
     data  |  user data       |                 |                  |
    ------>|----------------->|     user data   | user data        |
           |                  |---------------->C(# marked bytes)  |
           |                  |                 C----------------->|
           |                  |                 C(# unmarked bytes)|
           |                  |                 C----------------->|
   probe   |   probe          |                 C                  |
    ------>|----------------->|     probe       C                  |
           |                  |---------------->C probe            |
           |                  |                 C----------------->|
           |                  | notification QoS Agent              |
           |<------------------------------------------------------|
           |                  |                 C                  |
           |                  |                 C                  |
   
    Figure: 1  Admission control based on probing within one Diffserv 
                 Domain: controlled by egress
   
   7.2 Multiple Diffserv domains admission control based on probing: 
       controlled by egress
   
   In this scenario the RU PDB is applied within multiple neigboring 
   Diffserv domains. The operation of this scenario is similar to the 
   scenario described in Section 7.1. The main difference is related to 
   the fact that only the egress edge, which is located within the last 
   downstream Diffserv domain provides the final handling decision on 
   the admission control specified in Section 3.3.1. All the other end 
   systems, ingress and egress edges operate according to the 
   specification given in Section 3.2.1. Furthermore, the ingress edge 
   of the first downstream Diffserv domain supports the RU PDB 
   functionality specified in Section 3.1. 
   
   Note that also other ingress edges MAY support the RU PDB 
   functionality specified in Section 3.1. In addition to this it is 
   considered that each edge MAY maintain a QoS agent. Note that if the 
   egress edge located within the last downstream Diffserv domain 
   maintains a QoS Agent, then the QoS Agent at this egress SHOULD send 
   a notify like message to the QoS Agent available at an intermediate 
   ingress edge to inform him about the denial of the probe packet 
   request.
   
   The downstream packets marked with the local1_DSCP are remarked back 
   to the original_DSCP by the egress edge, which is located within the 
   last downstream Diffserv domain.
   

Karagiannis, et al.                                            [Page 11]

INTERNET-DRAFT                                                   RU-PDB

   
7.3 Single Diffserv domain admission control based on probing: 
    controlled by receiving end system
   
   This scenario is very much similar to the scenario described in 
   Section 7.1.
   
   The main difference is related to the fact that in this scenario the
   egress node does not support the functionality specified in Section 
   3.3.1, while the Diffserv receiving end system is supporting this 
   functionality, see Figure 2. In addition to this it is considered 
   that the end systems (Source and Receiver) are also supporting the 
   functionality specified in 3.2.1 and MAY maintain a QoS agent. The 
   downstream packets marked with the local1_DSCP are remarked back to 
   the original_DSCP by the receiver and not by the Diffserv egress.

   
   Source Ingress    Interior      Interior        Egress      Receiver
   
     data  |  user data  |             |                |            |
    ------>|------------>| user data   | user data      |            |
           |             |------- ---->C(# marked bytes)|            |
           |             |             C--------------->|            |
           |             |            (# unmarked bytes)|----------->|
           |             |             C--------------->|            |
   probe   |   probe     |             C                |----------->|
    ------>|------------>|     probe   C                |            |
           |             |------------>C probe          |            |
           |             |             C--------------->|            |
           |           notification QoS Agent           |----------->|
           |             |             C                |            |
           |             |             C                |            |
           |<-------------- ----------------------------|------------|
           |             |             C                |            |
   
   Figure: 2  Admission control based on probing within one Diffserv 
   domain:  controlled by receiver
   

Karagiannis, et al.                                            [Page 12]

INTERNET-DRAFT                                                   RU-PDB

   
7.4 Multiple Diffserv domains admission control based on probing: 
   controlled by receiving end system
   
   This scenario is very much similar to the scenario described in 
   Section 7.2.
   
   The main difference is related to the fact that in this scenario the
   egress node does not support the functionality specified in Section 
   3.3.1, while the Diffserv receiving end system (Receiver) is 
   supporting this functionality. In addition to this it is considered 
   that the end systems (Source and Receiver) are also supporting the 
   functionality specified in 3.2.1 and MAY maintain a QoS agent. The 
   downstream packets marked with the local1_DSCP are remarked back to 
   the original_DSCP by the receiver and not by the Diffserv egress.
   
   
7.5 Single Diffserv domain severe congestion handling: controlled by 
    egress
   
   In this scenario the RU PDB is applied only within one Diffserv 
   domain.
   
   The ingress router supports the RU PDB functionality specified in 
   Section 3.1. The interior router supports the RU PDB functionality 
   as specified in section 3.2.2 and the egress edge supports the 
   functionality as specified in section 3.3.2. In addition to this is 
   is considered that each edge MAY maintain a QoS agent. 
   
   In this example, see Figure 3, a severe congestion detection 
   procedure is shown. 
   
   When a failure in a communication path, e.g. router or link 
   failure, occurs, the routing algorithms will adapt to these failures 
   by changing the routing decisions to reflect changes in the topology 
   and traffic volume.  As a result, the re-routed traffic will follow 
   a new path, which may result in overloaded nodes as they need to 
   support more traffic.  This may cause severe congestion in the 
   communication path.  In this situation the available resources, are 
   not enough to meet the required QoS for all the flows along the new 
   path.  
   
   This severe congestion is detected in this example by using the RU 
   PDB. 

Karagiannis, et al.                                            [Page 13]

INTERNET-DRAFT                                                   RU-PDB

   
   When an Interior node becomes severe congested (S), i.e., measured 
   bandwidth is higher than Threshold2, then the remarking procedure 
   explained in Section 3.2.2 will be used. Note that the # marked 
   bytes, in Figure 2,represent the local2_DSCP and local3_DSCP bytes, 
   while the # unmarked bytes represent the original_DSCP bytes.
   
   If a QoS Agent is available at the ingress and egress, then the QoS 
   Agent at the egress SHOULD send for each selected flow (that should 
   be terminated) a notify like message to the QoS Agent available at 
   the ingress edge to inform him that the flow has to be terminated.
   
   The downstream packets marked with the local2_DSCP and local3_DSCP 
   are remarked back to the original_DSCP by this egress edge. 
   
   
     Ingress            Interior           Interior           Egress
  
    user  |                  |                 |                  |
    data  |  user data       |                 |                  |
   ------>|----------------->|     user data   | user data        |
          |                  |---------------->S(# marked bytes)  |
          |                  |                 S----------------->|
          |                  |                 S(# unmarked bytes)|
          |                  |                 S----------------->|Term.
          |            notification QoS Agent  S                  |flow?
          |<----------------|------------------S------------------|YES
   
   Figure: 3 Severe congestion detection within one Diffserv domain:  
             controlled by egress
   
   
7.6 Multiple Diffserv domains severe congestion handling: controlled 
    by egress
   
   In this scenario the RU PDB is applied within multiple neigboring 
   Diffserv domains. The operation of this scenario is similar to the 
   scenario described in Section 7.5. The main difference is related to 
   the fact that only the egress edge, which is located within the last 
   downstream Diffserv domain maintains the admission control 
   functionality specified in Section 3.3.2. All the other ingress and 
   egress edges operate according to the specification given in Section 
   3.2.2. Furthermore, the ingress edge of the first downstream 
   Diffserv domain supports the RU PDB functionality specified in 
   Section 3.1. 

Karagiannis, et al.                                            [Page 14]

INTERNET-DRAFT                                                   RU-PDB

   
   Note that also other ingress edges MAY support the RU PDB 
   functionality specified in Section 3.1. In addition to this it is 
   considered that each edge MAY maintain a QoS agent. Note that if the 
   egress edge located within the last downstream Diffserv domain 
   maintains a QoS Agent, then the QoS Agent at this egress SHOULD send 
   a notify like message to the QoS Agent available at an intermediate 
   ingress edge to inform him that the flow should be terminated. 
   
   The downstream packets marked with the local2_DSCP and local3_DSCP 
   are remarked back to the original_DSCP by the egress edge, which is 
   located within the last downstream Diffserv domain.
   
   
8. Environmental concerns
   
   There are no environmental concerns specific to this PDB. However, 
   depending on the the area wherein the RU PDB is applied (one 
   Diffserv domain or multiple neigboring domains), different 
   requirements have to be fulfilled by the network operators, see 
   Section 6.

   
9. Security considerations for RU PDB
   
   There are no specific security exposures for this PDB.  See the
   general security considerations in [RFC2474] and [RFC2475].

10. IANA Considerations
 
   
11. Acknowledgements
   
   We thank Kathie Nichols for reviewing this draft and providing 
   Useful comments.

Karagiannis, et al.                                            [Page 15]

INTERNET-DRAFT                                                   RU-PDB

 
12. Normative References
   
   [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
             Requirement Levels", BCP 14, RFC2119, March 1997.
   
   [RFC2474]  Nichols, K., Blake, S., Baker, F. and D. Black,
              "Definition of the Differentiated Services Field (DS
              Field) in the IPv4 and IPv6 Headers", RFC 2474, 
              December 1998.

   [RFC2475]  Blake, S., Black, D., Carlson, M., Davies, E., Wang, 
              and W. Weiss, "An Architecture for Differentiated
              Services", RFC 2475, December 1998.
   
      [RFC3086]  Nichols, K. and B. Carpenter, "Definition of
                 Differentiated Services Per Domain Behaviors and Rules 
                 For their Specification", RFC 3086, April 2001.
   
   
13 Informative References
   
   [RFC2205]  Braden, R., Zhang, L., Berson, S., Herzog, A., Jamin, 
              S., "Resource ReSerVation Protocol (RSVP)-- Version 1 
              Functional Specification", IETF RFC 2205, 1997.
   
   [RFC2597]  Heinanen, J., Baker, F., Weiss, W. and J. Wroclawski,
              "Assured Forwarding PHB Group", RFC 2597, June 1999.
   
   [RFC3290]  Bernet, Y., Blake, S., Grossman, D., Smith, A., "An 
              Informal Management Model for Diffserv Routers", RFC 3290, 
              May 2002.
   
   [RFC3246]  B. Davie, et al., "An Expedited Forwarding PHB (Per-
              Hop Behavior) ", RFC 3246, March 2002.
   
   [RFC4080] Hancock, R., "Next Steps in Signaling: Framework", RFC
             4080, December 2004.

Karagiannis, et al.                                            [Page 16]

INTERNET-DRAFT                                                   RU-PDB

   
   Authors' Addresses
   
   Georgios Karagiannis
   University of Twente
   P.O.  BOX 217
   7500 AE Enschede, The Netherlands
   EMail: g.karagiannis@ewi.utwente.nl

   Lars Westberg
   Ericsson Research
   Kistagangen 26
   SE-164 80 Stockholm
   Sweden
   EMail: Lars.Westberg@ericsson.com
   
   Attila Bader
   Ericsson Research
   Ericsson Hungary Ltd.
   Laborc 1, Budapest, H-1037 
   Hungary
   EMail: Attila.Bader@ericsson.com


  Intellectual Property Statement

   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.

Karagiannis, et al.                                            [Page 17]

INTERNET-DRAFT                                                   RU-PDB

Disclaimer of Validity

   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.

   Copyright (C) The Internet Society (2006).

   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.

PAFTECH AB 2003-20262026-04-24 05:54:03