4.  Configuration Parameters

   This section specifies the configuration and state parameters of the
   configuration data model separately for each class.

4.1.  ObservationPoint Class

         +-------------------------------+
         | ObservationPoint              |
         +-------------------------------+
         | name                          |
         | observationPointId {readOnly} |
         | observationDomainId           | 0..*
         | ifName[0..*]                  |-------------+
         | ifIndex[0..*]                 |             | 0..*
         | entPhysicalName[0..*]         |             V
         | entPhysicalIndex[0..*]        |    +------------------+
         | direction = "both"            |    | SelectionProcess |
         +-------------------------------+    +------------------+

                     Figure 7: ObservationPoint class

   Figure 7 shows the ObservationPoint class that specifies an
   Observation Point of the Monitoring Device.

   As defined in [RFC5101], an Observation Point can be any location
   where packets are observed.  A Monitoring Device potentially has more
   than one such location.  An instance of ObservationPoint class
   defines which location is associated with a specific Observation
   Point.  For this purpose, interfaces and physical entities are
   identified using their names.  Alternatively, index values of the
   corresponding entries in the ifTable (IF-MIB module [RFC2863]) or the
   entPhysicalTable (ENTITY-MIB module [RFC4133]) can be used as
   identifiers.  However, indices SHOULD only be used as identifiers if
   an SNMP agent on the same Monitoring Device enables access to the
   corresponding MIB tables.

   By its definition in [RFC5101], an Observation Point may be
   associated with a set of interfaces.  Therefore, the configuration
   data model allows configuring multiple interfaces and physical
   entities for a single Observation Point.

   The Observation Point ID (i.e., the value of the Information Element
   observationPointId [IANA-IPFIX]) is assigned by the Monitoring
   Device.  It appears as a state parameter in the ObservationPoint
   class.

   The configuration parameters of the Observation Point are:

   observationDomainId:  This parameter defines the identifier of the
      Observation Domain the Observation Point belongs to.  Observation
      Points that are configured with the same Observation Domain ID
      belong to the same Observation Domain.
      Note that this parameter corresponds to
      ipfixObservationPointObservationDomainId in the IPFIX MIB module
      [RFC6615].

   ifName/ifIndex/entPhysicalName/entPhysicalIndex:  These parameters
      identify interfaces and physical entities (e.g., linecards) that
      are on the Monitoring Device and are associated with the given
      Observation Point.
      An interface is either identified by its name (ifName) or the
      ifIndex value of the corresponding object in the IF-MIB module
      [RFC2863]. ifIndex SHOULD only be used if an SNMP agent enables
      access to the ifTable.
      Similarly, a physical entity is either identified by its name
      (entPhysicalName) or the entPhysicalIndex value of the
      corresponding object in the ENTITY-MIB module [RFC4133].
      entPhysicalIndex SHOULD only be used if an SNMP agent enables
      access to the entPhysicalTable.
      Note that the parameters ifIndex and entPhysicalIndex correspond
      to ipfixObservationPointPhysicalInterface and
      ipfixObservationPointPhysicalEntity in the IPFIX MIB module
      [RFC6615].

   direction:  This parameter specifies if ingress traffic, egress
      traffic, or both ingress and egress traffic is captured, using the
      values "ingress", "egress", and "both", respectively.  If not
      configured, ingress and egress traffic is captured (i.e., the
      default value is "both").  If not applicable (e.g., in the case of
      a sniffing interface in promiscuous mode), the value of this
      parameter is ignored.

   An ObservationPoint object MAY refer to one or more SelectionProcess
   objects configuring Selection Processes that process the observed
   packets in parallel.

4.2.  SelectionProcess Class

       +------------------+
       | SelectionProcess |
       +------------------+   1..* +----------+
       | name             |<>------| Selector |
       |                  |        +----------+
       |                  |
       |                  |   0..* +--------------------------------+
       |                  |<>------| SelectionSequence              |
       |                  |        +--------------------------------+
       |                  |        | observationDomainId {readOnly} |
       |                  |        | selectionSequenceId {readOnly} |
       |                  |        +--------------------------------+
       |                  |
       |                  | 0..*  0..1 +-------+
       |                  |----------->| Cache |
       +------------------+            +-------+

                     Figure 8: SelectionProcess class

   Figure 8 shows the SelectionProcess class.  The SelectionProcess
   class contains the configuration and state parameters of a Selection
   Process that selects packets from one or more Observed Packet Streams
   and generates a Selected Packet Stream as its output.  A non-empty
   ordered list defines a sequence of Selectors.  The actions defined by
   the Selectors are applied to the stream of incoming packets in the
   specified order.

   If the Selection Process receives packets from multiple Observation
   Points, the Observed Packet Streams need to be processed
   independently in separate Selection Sequences.  Each Selection
   Sequence is identified by a Selection Sequence ID that is unique
   within the Observation Domain the Observation Point belongs to (see
   [RFC5477]).  Selection Sequence IDs are assigned by the Monitoring
   Device.  As state parameters, the SelectionProcess class contains a
   list of (observationDomainId, selectionSequenceId) tuples specifying
   the assigned Selection Sequence IDs and corresponding Observation
   Domain IDs.  With this information, it is possible to associate
   Selection Sequence (Statistics) Report Interpretations exported
   according to the PSAMP protocol specification [RFC5476] with the
   corresponding object of the SelectionProcess class.

   A SelectionProcess object MAY include a reference to an object of the
   Cache class to generate Packet Reports or Flow Records from the
   Selected Packet Stream.

4.2.1.  Selector Class

    +--------------------------------------+
    | Selector                             |
    +--------------------------------------+      1 +-----------------+
    | name                                 |<>------+ SelectAll/      |
    | packetsObserved {readOnly}           |        | SampCountBased/ |
    | packetsDropped {readOnly}            |        | SampTimeBased/  |
    | selectorDiscontinuityTime {readOnly} |        | SampRandOutOfN/ |
    |                                      |        | SampUniProb/    |
    |                                      |        | FilterMatch/    |
    |                                      |        | FilterHash/     |
    +--------------------------------------+        +-----------------+

                         Figure 9: Selector class

   The Selector class in Figure 9 contains the configuration and state
   parameters of a Selector.  Standardized PSAMP Sampling and Filtering
   methods are described in [RFC5475]; their configuration parameters
   are specified in the classes SampCountBased, SampTimeBased,
   SampRandOutOfN, SampUniProb, FilterMatch, and FilterHash.  In
   addition, the SelectAll class, which has no parameters, is used for a
   Selector that selects all packets.  The Selector class includes
   exactly one of these sampler and filter classes, depending on the
   applied method.

   As state parameters, the Selector class contains the Selector
   statistics packetsObserved and packetsDropped as well as
   selectorDiscontinuityTime, which correspond to the IPFIX MIB module
   objects ipfixSelectionProcessStatsPacketsObserved,
   ipfixSelectionProcessStatsPacketsDropped, and
   ipfixSelectionProcessStatsDiscontinuityTime, respectively [RFC6615]:

   packetsObserved:  The total number of packets observed at the input
      of the Selector.  If this is the first Selector in the Selection
      Process, this counter corresponds to the total number of packets
      in all Observed Packet Streams at the input of the Selection
      Process.  Otherwise, the counter corresponds to the total number
      of packets at the output of the preceding Selector.
      Discontinuities in the value of this counter can occur at
      re-initialization of the management system, and at other times as
      indicated by the value of selectorDiscontinuityTime.

   packetsDropped:  The total number of packets discarded by the
      Selector.  Discontinuities in the value of this counter can occur
      at re-initialization of the management system, and at other times
      as indicated by the value of selectorDiscontinuityTime.

   selectorDiscontinuityTime:  Timestamp of the most recent occasion at
      which one or more of the Selector counters suffered a
      discontinuity.  In contrast to
      ipfixSelectionProcessStatsDiscontinuityTime, the time is absolute
      and not relative to sysUpTime.

   Note that packetsObserved and packetsDropped are aggregate statistics
   calculated over all Selection Sequences of the Selection Process.
   This is in contrast to the counter values in the Selection Sequence
   Statistics Report Interpretation [RFC5476], which are related to a
   single Selection Sequence only.

4.2.2.  Sampler Classes

        +----------------+   +----------------+   +----------------+
        | SampCountBased |   | SampTimeBased  |   | SampRandOutOfN |
        +----------------+   +----------------+   +----------------+
        | packetInterval |   | timeInterval   |   | population     |
        | packetSpace    |   | timeSpace      |   | size           |
        +----------------+   +----------------+   +----------------+

        +----------------+
        | SampUniProb    |
        +----------------+
        | probability    |
        +----------------+

                        Figure 10: Sampler classes

   The Sampler classes in Figure 10 contain the configuration parameters
   of specific Sampling algorithms:

   packetInterval, packetSpace:  For systematic count-based Sampling,
      packetInterval defines the number of packets that are
      consecutively sampled between gaps of length packetSpace.  These
      parameters correspond to the Information Elements
      samplingPacketInterval and samplingPacketSpace [RFC5477], as well
      as to the PSAMP MIB objects psampSampCountBasedInterval and
      psampSampCountBasedSpace [RFC6727].

   timeInterval, timeSpace:  For systematic time-based Sampling,
      timeInterval defines the time interval during which all arriving
      packets are sampled. timeSpace is the gap between two Sampling
      intervals.  These parameters correspond to the Information
      Elements samplingTimeInterval and samplingTimeSpace [RFC5477], as
      well as to the PSAMP MIB objects psampSampTimeBasedInterval and
      psampSampTimeBasedSpace [RFC6727].  The unit is microseconds.

   size, population:  For n-out-of-N random Sampling, size defines the
      number of elements taken from the parent population. population
      defines the number of elements in the parent population.  These
      parameters correspond to the Information Elements samplingSize and
      samplingPopulation [RFC5477], as well as to the PSAMP MIB objects
      psampSampRandOutOfNSize and psampSampRandOutOfNPopulation
      [RFC6727].

   probability:  For uniform probabilistic Sampling, probability defines
      the Sampling probability.  The probability is expressed as a value
      between 0 and 1.  This parameter corresponds to the Information
      Element samplingProbability [RFC5477], as well as to the PSAMP MIB
      object psampSampUniProbProbability [RFC6727].

4.2.3.  Filter Classes

          +---------------------------+
          | FilterMatch               |
          +---------------------------+
          | ieId/ieName               |
          | ieEnterpriseNumber = 0    |
          | value                     |
          +---------------------------+

          +---------------------------+
          | FilterHash                |
          +---------------------------+    1..* +---------------+
          | hashFunction = "BOB"      |<>-------| SelectedRange |
          | initializerValue[0..1]    |         +---------------+
          | ipPayloadOffset = 0       |         | name          |
          | ipPayloadSize = 8         |         | min           |
          | digestOutput = "false"    |         | max           |
          | outputRangeMin {readOnly} |         +---------------+
          | outputRangeMax {readOnly} |
          +---------------------------+

                         Figure 11: Filter classes

   The Filter classes in Figure 11 contain the configuration parameters
   of specific Filtering methods.  For property match Filtering, the
   configuration parameters are:

   ieId, ieName, ieEnterpriseNumber:  The property to be matched is
      specified by either ieId or ieName, specifying the identifier or
      name of the Information Element, respectively.  If
      ieEnterpriseNumber is zero (which is the default), this
      Information Element is registered in the IANA registry of IPFIX
      Information Elements [IANA-IPFIX].  A non-zero value of
      ieEnterpriseNumber specifies an enterprise-specific Information
      Element [IANA-ENTERPRISE-NUMBERS].

   value:  Matching value.

   For hash-based Filtering, the configuration and state parameters are:

   hashFunction:  Hash function to be used.  The following parameter
      values are defined by the configuration data model:
      *  BOB: BOB Hash Function as specified in [RFC5475], Appendix A.2
      *  IPSX: IP Shift-XOR (IPSX) Hash Function as specified in
         [RFC5475], Appendix A.1
      *  CRC: CRC-32 function as specified in [RFC1141]
      Default value is "BOB".  This parameter corresponds to the PSAMP
      MIB object psampFiltHashFunction [RFC6727].

   initializerValue:  Initializer value to the hash function.  This
      parameter corresponds to the Information Element
      hashInitialiserValue [RFC5477], as well as to the PSAMP MIB object
      psampFiltHashInitializerValue [RFC6727].  If not configured by the
      user, the Monitoring Device arbitrarily chooses an initializer
      value.

   ipPayloadOffset, ipPayloadSize:  ipPayloadOffset and ipPayloadSize
      configure the offset and the size of the payload section used as
      input to the hash function.  Default values are 0 and 8,
      respectively, corresponding to the minimum configurable values
      according to [RFC5476], Section 6.5.2.6.  These parameters
      correspond to the Information Elements hashIPPayloadOffset and
      hashIPPayloadSize [RFC5477], as well as to the PSAMP MIB objects
      psampFiltHashIpPayloadOffset and psampFiltHashIpPayloadSize
      [RFC6727].

   digestOutput:  digestOutput enables or disables the inclusion of the
      packet digest in the resulting PSAMP Packet Report.  This requires
      that the Cache Layout of the Cache generating the Packet Reports
      includes a digestHashValue field.  This parameter corresponds to
      the Information Element hashDigestOutput [RFC5477].

   outputRangeMin, outputRangeMax:  The values of these two state
      parameters are the beginning and end of the hash function's
      potential output range.  These parameters correspond to the
      Information Elements hashOutputRangeMin and hashOutputRangeMax
      [RFC5477], as well as to the PSAMP MIB objects
      psampFiltHashOutputRangeMin and psampFiltHashOutputRangeMax
      [RFC6727].

   One or more ranges of matching hash values are defined by the min and
   max parameters of the SelectedRange subclass.  These parameters
   correspond to the Information Elements hashSelectedRangeMin and
   hashSelectedRangeMax [RFC5477], as well as to the PSAMP MIB objects
   psampFiltHashSelectedRangeMin and psampFiltHashSelectedRangeMax
   [RFC6727].

4.3.  Cache Class

    +-----------------------------------+
    | Cache                             |
    +-----------------------------------+        1 +------------------+
    | name                              |<>--------| immediateCache/  |
    | meteringProcessId {readOnly}      |          | timeoutCache/    |
    | dataRecords {readOnly}            |          | naturalCache/    |
    | cacheDiscontinuityTime {readOnly} |          | permanentCache   |
    |                                   |          +------------------+
    |                                   |
    |                                   |     0..* +------------------+
    |                                   |--------->| ExportingProcess |
    +-----------------------------------+          +------------------+

                          Figure 12: Cache class

   Figure 12 shows the Cache class that contains the configuration and
   state parameters of a Cache.  Most of these parameters are specific
   to the type of the Cache and therefore contained in the subclasses
   immediateCache, timeoutCache, naturalCache, and permanentCache, which
   are presented below in Sections 4.3.1 and 4.3.2.  The following three
   state parameters are common to all Caches and therefore included in
   the Cache class itself:

   meteringProcessId:  The identifier of the Metering Process the Cache
      belongs to.
      This parameter corresponds to the Information Element
      meteringProcessId [IANA-IPFIX].  Its occurrence helps to associate
      Metering Process (Reliability) Statistics exported according to
      the IPFIX protocol specification [RFC5101] with the corresponding
      object of the MeteringProcess class.

   dataRecords:  The number of Data Records generated by this Cache.
      Discontinuities in the value of this counter can occur at
      re-initialization of the management system, and at other times as
      indicated by the value of cacheDiscontinuityTime.
      Note that this parameter corresponds to
      ipfixMeteringProcessDataRecords in the IPFIX MIB module [RFC6615].

   cacheDiscontinuityTime:  Timestamp of the most recent occasion at
      which dataRecords suffered a discontinuity.  In contrast to
      ipfixMeteringProcessDiscontinuityTime, the time is absolute and
      not relative to sysUpTime.
      Note that this parameter functionally corresponds to
      ipfixMeteringProcessDiscontinuityTime in the IPFIX MIB module
      [RFC6615].

   A Cache object MAY refer to one or more ExportingProcess objects
   configuring different Exporting Processes.

4.3.1.  ImmediateCache Class

         +-------------------------------+
         | ImmediateCache                |
         +-------------------------------+       1 +-------------+
         |                               |<>-------| CacheLayout |
         +-------------------------------+         +-------------+

                      Figure 13: ImmediateCache class

   The ImmediateCache class depicted in Figure 13 is used to configure a
   Cache that generates a PSAMP Packet Report for each packet at its
   input.  The fields contained in the generated Data Records are
   defined in an object of the CacheLayout class, which is defined below
   in Section 4.3.3.

4.3.2.  TimeoutCache, NaturalCache, and PermanentCache Class

         +-------------------------------+
         | TimeoutCache                  |
         +-------------------------------+       1 +-------------+
         | maxFlows {opt.}               |<>-------| CacheLayout |
         | activeTimeout {opt.}          |         +-------------+
         | idleTimeout {opt.}            |
         | activeFlows {readOnly}        |
         | unusedCacheEntries {readOnly} |
         +-------------------------------+

         +-------------------------------+
         | NaturalCache                  |
         +-------------------------------+       1 +-------------+
         | maxFlows {opt.}               |<>-------| CacheLayout |
         | activeTimeout {opt.}          |         +-------------+
         | idleTimeout {opt.}            |
         | activeFlows {readOnly}        |
         | unusedCacheEntries {readOnly} |
         +-------------------------------+

         +-------------------------------+
         | PermanentCache                |
         +-------------------------------+       1 +-------------+
         | maxFlows {opt.}               |<>-------| CacheLayout |
         | exportInterval {opt.}         |         +-------------+
         | activeFlows {readOnly}        |
         | unusedCacheEntries {readOnly} |
         +-------------------------------+

      Figure 14: TimeoutCache, NaturalCache, and PermanentCache class

   Figure 14 shows the TimeoutCache class, the NaturalCache class, and
   the PermanentCache class.  These classes are used to configure a
   Cache that aggregates the packets at its input and generates IPFIX
   Flow Records.  The three classes differ in when Flows expire:

   o  TimeoutCache: Flows expire after active or idle timeout.
   o  NaturalCache: Flows expire after active or idle timeout, or on
      natural termination (e.g., TCP FIN or TCP RST) of the Flow.
   o  PermanentCache: Flows never expire, but are periodically exported
      with the interval set by exportInterval.

   The following configuration and state parameters are common to the
   three classes:

   maxFlows:  This parameter configures the maximum number of entries in
      the Cache, which is the maximum number of Flows that can be
      measured simultaneously.
      If this parameter is configured, the Monitoring Device MUST ensure
      that sufficient resources are available to store the configured
      maximum number of Flows.  If the maximum number of Cache entries
      is in use, no additional Flows can be measured.  However, traffic
      that pertains to existing Flows can continue to be measured.

   activeFlows:  This state parameter indicates the number of Flows
      currently active in this Cache (i.e., the number of Cache entries
      currently in use).
      Note that this parameter corresponds to
      ipfixMeteringProcessCacheActiveFlows in the IPFIX MIB module
      [RFC6615].

   unusedCacheEntries:  The number of unused cache entries.  Note that
      the sum of activeFlows and unusedCacheEntries equals maxFlows if
      maxFlows is configured.
      Note that this parameter corresponds to
      ipfixMeteringProcessCacheUnusedCacheEntries in the IPFIX MIB
      module [RFC6615].

   The following timeout parameters are only available in the
   TimeoutCache class and the NaturalCache class:

   activeTimeout:  This parameter configures the time in seconds after
      which a Flow is expired even though packets matching this Flow are
      still received by the Cache.  The parameter value zero indicates
      infinity, meaning that there is no active timeout.
      If not configured by the user, the Monitoring Device sets this
      parameter.
      Note that this parameter corresponds to
      ipfixMeteringProcessCacheActiveTimeout in the IPFIX MIB module
      [RFC6615].

   idleTimeout:  This parameter configures the time in seconds after
      which a Flow is expired if no more packets matching this Flow are
      received by the Cache.  The parameter value zero indicates
      infinity, meaning that there is no idle timeout.
      If not configured by the user, the Monitoring Device sets this
      parameter.
      Note that this parameter corresponds to
      ipfixMeteringProcessCacheIdleTimeout in the IPFIX MIB module
      [RFC6615].

   The following interval parameter is only available in the
   PermanentCache class:

   exportInterval:  This parameter configures the interval (in seconds)
      for periodical export of Flow Records.
      If not configured by the user, the Monitoring Device sets this
      parameter.

   Every generated Flow Record MUST be associated with a single
   Observation Domain.  Hence, although a Cache MAY be configured to
   process packets observed at multiple Observation Domains, the Cache
   MUST NOT aggregate packets observed at different Observation Domains
   in the same Flow.

   An object of the Cache class contains an object of the CacheLayout
   class that defines which fields are included in the Flow Records.

4.3.3.  CacheLayout Class

         +--------------+
         | CacheLayout  |
         +--------------+   1..* +--------------------------------+
         |              |<>------| CacheField                     |
         |              |        +--------------------------------+
         |              |        | name                           |
         |              |        | ieId/ieName                    |
         |              |        | ieLength {opt.}                |
         |              |        | ieEnterpriseNumber = 0         |
         |              |        | isFlowKey[0..1] {not used with |
         |              |        |   ImmediateCache class}        |
         +--------------+        +--------------------------------+

                       Figure 15: CacheLayout class

   A Cache generates and maintains Packet Reports or Flow Records
   containing information that has been extracted from the incoming
   stream of packets.  Using the CacheField class, the CacheLayout class
   specifies the superset of fields that are included in the Packet
   Reports or Flow Records (see Figure 15).

   If Packet Reports are generated (i.e., if ImmediateCache class is
   used to configure the Cache), every field specified by the Cache
   Layout MUST be included in the resulting Packet Report unless the
   corresponding Information Element is not applicable or cannot be
   derived from the content or treatment of the incoming packet.  Any
   other field specified by the Cache Layout MAY only be included in the

   Packet Report if it is obvious from the field value itself or from
   the values of other fields in same Packet Report that the field value
   was not determined from the packet.

   For example, if a field is configured to contain the TCP source port
   (Information Element tcpSourcePort [IANA-IPFIX]), the field MUST be
   included in all Packet Reports that are related to TCP packets.
   Although the field value cannot be determined for non-TCP packets,
   the field MAY be included in the Packet Reports if another field
   contains the transport protocol identifier (Information Element
   protocolIdentifier [IANA-IPFIX]).

   If Flow Records are generated (i.e., if TimeoutCache, NaturalCache,
   or PermanentCache class is used to configure the Cache), the Cache
   Layout differentiates between Flow Key fields and non-key fields.
   Every Flow Key field specified by the Cache Layout MUST be included
   as Flow Key in the resulting Flow Record unless the corresponding
   Information Element is not applicable or cannot be derived from the
   content or treatment of the incoming packet.  Any other Flow Key
   field specified by the Cache Layout MAY only be included in the Flow
   Record if it is obvious from the field value itself or from the
   values of other Flow Key fields in the same Flow Record that the
   field value was not determined from the packet.  Two packets are
   accounted by the same Flow Record if none of their Flow Key fields
   differ.  If a Flow Key field can be determined for one packet but not
   for the other, the two packets are accounted in different Flow
   Records.

   Every non-key field specified by the Cache Layout MUST be included in
   the resulting Flow Record unless the corresponding Information
   Element is not applicable or cannot be derived for the given Flow.
   Any other non-key field specified by the Cache Layout MAY only be
   included in the Flow Record if it is obvious from the field value
   itself or from the values of other fields in same Flow Record that
   the field value was not determined from the packet.  Packets which
   are accounted by the same Flow Record may differ in their non-key
   fields, or one or more of the non-key fields can be undetermined for
   all or some of the packets.

   For example, if a non-key field specifies an Information Element
   whose value is determined by the first packet observed within a Flow
   (which is the default rule according to [RFC5102] unless specified
   differently in the description of the Information Element), this
   field MUST be included in the resulting Flow Record if it can be
   determined from the first packet of the Flow.

   The CacheLayout class does not have any parameters.  The
   configuration parameters of the CacheField class are as follows:

   ieId, ieName, ieEnterpriseNumber:  These parameters specify a field
      by the combination of the Information Element identifier or name,
      and the Information Element enterprise number.  Either ieId or
      ieName MUST be specified.  If ieEnterpriseNumber is zero (which is
      the default), this Information Element is registered in the IANA
      registry of IPFIX Information Elements [IANA-IPFIX].  A non-zero
      value of ieEnterpriseNumber specifies an enterprise-specific
      Information Element [IANA-ENTERPRISE-NUMBERS].
      If the enterprise number is set to 29305, this field contains a
      Reverse Information Element.  In this case, the Cache MUST
      generate Data Records in accordance to [RFC5103].

   ieLength:  This parameter specifies the length of the field in
      octets.  A value of 65535 means that the field is encoded as a
      variable-length Information Element.  For Information Elements of
      integer and float type, the field length MAY be set to a smaller
      value than the standard length of the abstract data type if the
      rules of reduced size encoding are fulfilled (see [RFC5101],
      Section 6.2).  If not configured by the user, the field length is
      set by the Monitoring Device.

   isFlowKey:  If present, this field is a Flow Key.  If the field
      contains a Reverse Information Element, it MUST NOT be configured
      as Flow Key.
      This parameter is not available if the Cache is configured using
      the ImmediateCache class since there is no distinction between
      Flow Key fields and non-key fields in Packet Reports.

   Note that the use of Information Elements can be restricted to
   certain Cache types as well as to Flow Key or non-key fields.  Such
   restrictions may result from Information Element definitions or from
   device-specific constraints.  According to Section 5, the Monitoring
   Device MUST notify the user if a Cache field cannot be configured
   with the given Information Element.

4.4.  ExportingProcess Class

     +-------------------------------+
     | ExportingProcess              |
     +-------------------------------+   1..* +-------------+
     | name                          |<>------| Destination |
     | exportingProcessId {readOnly} |        +-------------+
     | exportMode = "parallel"       |        | name        |<>-+
     |                               |        +-------------+   | 1
     |                               |                          |
     |                               |               +---------------+
     |                               |               | SctpExporter/ |
     |                               |               | UdpExporter/  |
     |                               |               | TcpExporter/  |
     |                               |               | FileWriter    |
     |                               |               +---------------+
     |                               |
     |                               |   0..* +------------------+
     |                               |<>------| Options          |
     +-------------------------------+        +------------------+

                     Figure 16: ExportingProcess class

   The ExportingProcess class in Figure 16 specifies destinations to
   which the incoming Packet Reports and Flow Records are exported using
   objects of the Destination class.  The Destination class includes one
   object of the SctpExporter, UdpExporter, TcpExporter, or FileWriter
   class which contains further configuration parameters.  These classes
   are described in Sections 4.4.1, 4.4.2, 4.4.3, and 4.4.4.

   As state parameter, the ExportingProcess class contains the
   identifier of the Exporting Process (exportingProcessId).  This
   parameter corresponds to the Information Element exportingProcessId
   [IANA-IPFIX].  Its occurrence helps to associate Exporting Process
   Reliability Statistics exported according to the IPFIX protocol
   specification [RFC5101] with the corresponding object of the
   ExportingProcess class.

   The order in which objects of the Destination class appear is defined
   by the user.  However, the order has a specific meaning only if the
   exportMode parameter is set to "fallback".  The exportMode parameter
   is defined as follows:

   exportMode:  This parameter determines to which configured
      destination(s) the incoming Data Records are exported.  The
      following parameter values are specified by the configuration data
      model:

      *  parallel: every Data Record is exported to all configured
         destinations in parallel
      *  loadBalancing: every Data Record is exported to exactly one
         configured destination according to a device-specific load-
         balancing policy
      *  fallback: every Data Record is exported to exactly one
         configured destination according to the fallback policy
         described below
      If exportMode is set to "fallback", the first object of the
      Destination class defines the primary destination, the second
      object of the Destination class defines the secondary destination,
      and so on.  If the Exporting Process fails to export Data Records
      to the primary destination, it tries to export them to the
      secondary one.  If the secondary destination fails as well, it
      continues with the tertiary, etc.
      "parallel" is the default value if exportMode is not configured.

   Note that the exportMode parameter is related to the
   ipfixExportMemberType object in [RFC6615].  If exportMode is
   "parallel", the ipfixExportMemberType values of the corresponding
   entries in ipfixExportTable are set to parallel(3).  If exportMode is
   "loadBalancing", the ipfixExportMemberType values of the
   corresponding entries in ipfixExportTable are set to
   loadBalancing(4).  If exportMode is "fallback", the
   ipfixExportMemberType value that refers to the primary destination is
   set to primary(1); the ipfixExportMemberType values that refer to the
   remaining destinations need to be set to secondary(2).  The IPFIX MIB
   module does not define any value for tertiary destination, etc.

   The reporting of information with Options Templates is defined with
   objects of the Options class.

   The Exporting Process may modify the Packet Reports and Flow Records
   to enable a more efficient transmission or storage under the
   condition that no information is changed or suppressed.  For example,
   the Exporting Process may shorten the length of a field according to
   the rules of reduced size encoding [RFC5101].  The Exporting Process
   may also export certain fields in a separate Data Record as described
   in [RFC5476].

4.4.1.  SctpExporter Class

    +------------------------------+
    | SctpExporter                 |
    +------------------------------+    0..1 +------------------------+
    | ipfixVersion = 10            |<>-------| TransportLayerSecurity |
    | sourceIPAddress[0..*]        |         +------------------------+
    | destinationIPAddress[1..*]   |
    | destinationPort = 4739|4740  |    0..1 +------------------------+
    | ifName/ifIndex[0..1]         |<>-------| TransportSession       |
    | sendBufferSize {opt.}        |         +------------------------+
    | rateLimit[0..1]              |
    | timedReliability = 0         |
    +------------------------------+

                       Figure 17: SctpExporter class

   The SctpExporter class shown in Figure 17 contains the configuration
   parameters of an SCTP export destination.  The configuration
   parameters are:

   ipfixVersion:  Version number of the IPFIX protocol used.  If
      omitted, the default value is 10 (=0x000a) as specified in
      [RFC5101].

   sourceIPAddress:  List of source IP addresses used by the Exporting
      Process.  If configured, the specified addresses are eligible
      local IP addresses of the multihomed SCTP endpoint.  If not
      configured, all locally assigned IP addresses are eligible local
      IP addresses.

   destinationIPAddress:  One or more IP addresses of the Collecting
      Process to which IPFIX Messages are sent.  The user must ensure
      that all configured IP addresses belong to the same Collecting
      Process.  The Exporting Process tries to establish an SCTP
      association to any of the configured destination IP addresses.

   destinationPort:  Destination port number to be used.  If not
      configured, standard port 4739 (IPFIX without TLS and DTLS) or
      4740 (IPFIX over TLS or DTLS) is used.

   ifIndex/ifName:  Either the index or the name of the interface used
      by the Exporting Process to export IPFIX Messages to the given
      destination MAY be specified according to corresponding objects in
      the IF-MIB [RFC2863].  If omitted, the Exporting Process selects
      the outgoing interface based on local routing decision and accepts
      return traffic, such as transport-layer acknowledgments, on all
      available interfaces.

   sendBufferSize:  Size of the socket send buffer in bytes.  If not
      configured by the user, the buffer size is set by the Monitoring
      Device.

   rateLimit:  Maximum number of bytes per second the Exporting Process
      may export to the given destination as required by [RFC5476].  The
      number of bytes is calculated from the lengths of the IPFIX
      Messages exported.  If this parameter is not configured, no rate
      limiting is performed for this destination.

   timedReliability:  Lifetime in milliseconds until an IPFIX Message
      containing Data Sets only is "abandoned" due to the timed
      reliability mechanism of the Partial Reliability extension of SCTP
      (PR-SCTP) [RFC3758].  If this parameter is set to zero, reliable
      SCTP transport MUST be used for all Data Records.  Regardless of
      the value of this parameter, the Exporting Process MAY use
      reliable SCTP transport for Data Sets associated with certain
      Options Templates, such as the Data Record Reliability Options
      Template specified in [RFC6526].

   Using the TransportLayerSecurity class described in Section 4.6,
   Datagram Transport Layer Security (DTLS) is enabled and configured
   for this export destination.

   If a Transport Session is established to the configured destination,
   the SctpExporter class includes an object of the TransportSession
   class containing state parameters of the Transport Session.  The
   TransportSession class is specified in Section 4.7.

4.4.2.  UdpExporter Class

    +-------------------------------------+
    | UdpExporter                         |
    +-------------------------------------+   0..1 +------------------+
    | ipfixVersion = 10                   |<>------| TransportLayer-  |
    | sourceIPAddress[0..1]               |        | Security         |
    | destinationIPAddress                |        +------------------+
    | destinationPort = 4739|4740         |
    | ifName/ifIndex[0..1]                |   0..1 +------------------+
    | sendBufferSize {opt.}               |<>------| TransportSession |
    | rateLimit[0..1]                     |        +------------------+
    | maxPacketSize {opt.}                |
    | templateRefreshTimeout = 600        |
    | optionsTemplateRefreshTimeout = 600 |
    | templateRefreshPacket[0..1]         |
    | optionsTemplateRefreshPacket[0..1]  |
    +-------------------------------------+

                       Figure 18: UdpExporter class

   The UdpExporter class shown in Figure 18 contains the configuration
   parameters of a UDP export destination.  The parameters ipfixVersion,
   destinationPort, ifName, ifIndex, sendBufferSize, and rateLimit have
   the same meaning as in the SctpExporter class (see Section 4.4.1).
   The remaining configuration parameters are:

   sourceIPAddress:  This parameter specifies the source IP address used
      by the Exporting Process.  If this parameter is omitted, the IP
      address assigned to the outgoing interface is used as the source
      IP address.

   destinationIPAddress:  Destination IP address to which IPFIX Messages
      are sent (i.e., the IP address of the Collecting Process).

   maxPacketSize:  This parameter specifies the maximum size of IP
      packets sent to the Collector.  If set to zero, the Exporting
      Device MUST derive the maximum packet size from path MTU discovery
      mechanisms.  If not configured by the user, this parameter is set
      by the Monitoring Device.

   templateRefreshTimeout, optionsTemplateRefreshTimeout,
      templateRefreshPacket, optionsTemplateRefreshPacket:  These
      parameters specify when (Options) Templates are refreshed by the
      Exporting Process.
      templateRefreshTimeout and optionsTemplateRefreshTimeout are
      specified in seconds between resendings of (Options) Templates.

      If omitted, the default value of 600 seconds (10 minutes) is used
      [RFC5101].
      templateRefreshPacket and optionsTemplateRefreshPacket specify the
      number of IPFIX Messages after which (Options) Templates are
      resent.  If omitted, the (Options) Templates are only resent after
      timeout.
      Note that the values configured for templateRefreshTimeout and
      optionsTemplateRefreshTimeout MUST be adapted to the
      templateLifeTime and optionsTemplateLifeTime parameter settings at
      the receiving Collecting Process (see Section 4.5.2).
      Note that these parameters correspond to
      ipfixTransportSessionTemplateRefreshTimeout,
      ipfixTransportSessionOptionsTemplateRefreshTimeout,
      ipfixTransportSessionTemplateRefreshPacket, and
      ipfixTransportSessionOptionsTemplateRefreshPacket in the IPFIX MIB
      module [RFC6615].

   Using the TransportLayerSecurity class described in Section 4.6, DTLS
   is enabled and configured for this export destination.

   If a Transport Session is established to the configured destination,
   the UdpExporter class includes an object of the TransportSession
   class containing state parameters of the Transport Session.  The
   TransportSession class is specified in Section 4.7.

4.4.3.  TcpExporter Class

    +------------------------------+
    | TcpExporter                  |
    +------------------------------+    0..1 +------------------------+
    | ipfixVersion = 10            |<>-------| TransportLayerSecurity |
    | sourceIPAddress[0..1]        |         +------------------------+
    | destinationIPAddress         |
    | destinationPort = 4739|4740  |    0..1 +------------------------+
    | ifName/ifIndex[0..1]         |<>-------| TransportSession       |
    | sendBufferSize {opt.}        |         +------------------------+
    | rateLimit[0..1]              |
    +------------------------------+

                       Figure 19: TcpExporter class

   The TcpExporter class shown in Figure 19 contains the configuration
   parameters of a TCP export destination.  The parameters have the same
   meaning as in the UdpExporter class (see Section 4.4.2).

   Using the TransportLayerSecurity class described in Section 4.6,
   Transport Layer Security (TLS) is enabled and configured for this
   export destination.

   If a Transport Session is established to the configured destination,
   the TcpExporter class includes an object of the TransportSession
   class containing state parameters of the Transport Session.  The
   TransportSession class is specified in Section 4.7.

4.4.4.  FileWriter Class

      +-----------------------------------------+
      | FileWriter                              |
      +-----------------------------------------+    0..* +----------+
      | ipfixVersion = 10                       |<>-------| Template |
      | file                                    |         +----------+
      | status {readOnly}                       |
      | bytes {readOnly}                        |
      | messages {readOnly}                     |
      | discardedMessages {readOnly}            |
      | records {readOnly}                      |
      | templates {readOnly}                    |
      | optionsTemplates {readOnly}             |
      | fileWriterDiscontinuityTime {readOnly}  |
      +-----------------------------------------+

                       Figure 20: FileWriter classes

   If an object of the FileWriter class is included in an object of the
   Destination class, IPFIX Messages are written into a file as
   specified in [RFC5655].  The FileWriter class contains the following
   configuration parameters:

   ipfixVersion:  Version number of the IPFIX protocol used.  If
      omitted, the default value is 10 (=0x000a) as specified in
      [RFC5101].

   file:  File name and location specified as URI.

   The state parameters of the FileWriter class are:

   bytes, messages, records, templates, optionsTemplates:  The number of
      bytes, IPFIX Messages, Data Records, Template Records, and Options
      Template Records written by the File Writer.  Discontinuities in
      the values of these counters can occur at re-initialization of the
      management system, and at other times as indicated by the value of
      fileWriterDiscontinuityTime.

   discardedMessages:  The number of IPFIX Messages that could not be
      written by the File Writer due to internal buffer overflows,
      limited storage capacity, etc.  Discontinuities in the value of
      this counter can occur at re-initialization of the management
      system, and at other times as indicated by the value of
      fileWriterDiscontinuityTime.

   fileWriterDiscontinuityTime:  Timestamp of the most recent occasion
      at which one or more File Writer counters suffered a
      discontinuity.  In contrast to discontinuity times in the IPFIX
      MIB module, the time is absolute and not relative to sysUpTime.

   Each object of the FileWriter class includes a list of objects of the
   Template class with information and statistics about the Templates
   written to the file.  The Template class is specified in Section 4.8.

4.4.5.  Options Class

                         +-----------------------+
                         | Options               |
                         +-----------------------+
                         | name                  |
                         | optionsType           |
                         | optionsTimeout {opt.} |
                         +-----------------------+

                         Figure 21: Options class

   The Options class in Figure 21 defines the type of specific
   information to be reported, such as statistics, flow keys, Sampling
   and Filtering parameters, etc.  [RFC5101] and [RFC5476] specify
   several types of reporting information that may be exported.  The
   following parameter values are specified by the configuration data
   model:

   meteringStatistics:  Export of Metering Process statistics using the
      Metering Process Statistics Options Template [RFC5101].

   meteringReliability:  Export of Metering Process reliability
      statistics using the Metering Process Reliability Statistics
      Options Template [RFC5101].

   exportingReliability:  Export of Exporting Process reliability
      statistics using the Exporting Process Reliability Statistics
      Options Template [RFC5101].

   flowKeys:  Export of the Flow Key specification using the Flow Keys
      Options Template [RFC5101].

   selectionSequence:  Export of Selection Sequence Report
      Interpretation and Selector Report Interpretation [RFC5476].

   selectionStatistics:  Export of Selection Sequence Statistics Report
      Interpretation [RFC5476].

   accuracy:  Export of Accuracy Report Interpretation [RFC5476].

   reducingRedundancy:  Enables the utilization of Options Templates to
      reduce redundancy in the exported Data Records according to
      [RFC5473].  The Exporting Process decides when to apply these
      Options Templates.

   extendedTypeInformation:  Export of extended type information for
      enterprise-specific Information Elements used in the exported
      Templates [RFC5610].

   The Exporting Process MUST choose a Template definition according to
   the options type and available options data.

   The optionsTimeout parameter specifies the reporting interval (in
   milliseconds) for periodic export of the option data.  A parameter
   value of zero means that the export of the option data is not
   triggered periodically, but whenever the available option data has
   changed.  This is the typical setting for options types flowKeys,
   selectionSequence, accuracy, and reducingRedundancy.  If
   optionsTimeout is not configured by the user, it is set by the
   Monitoring Device.