DHCP Options for the Port Control
Protocol (PCP)France TelecomRennes35000Francemohamed.boucadair@orange.comCiscoUSArepenno@cisco.comCisco Systems, Inc.170 West Tasman DriveSan JoseCalifornia95134USAdwing@cisco.comPCP Working GroupPCP Server discovery, Port Mapping, Shared AddressThis document specifies DHCP (IPv4 and IPv6) options to configure
hosts with Port Control Protocol (PCP) Server names. The use of DHCPv4
or DHCPv6 depends on the PCP deployment scenario.The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119.This document defines DHCPv4 and
DHCPv6 options which can be used to
provision PCP Server names.In order to make use of these options, this document assumes
appropriate name resolution means (e.g., Section 6.1.1 of ) are available on the host client.The use of DHCPv4 or DHCPv6 depends on the PCP deployment
scenarios.This document makes use of the following terms:PCP Server denotes a functional element which receives and
processes PCP requests from a PCP Client. A PCP Server can be
co-located with or be separated from the function (e.g., NAT,
Firewall) it controls. Refer to .PCP Client denotes a PCP software instance responsible for
issuing PCP requests to a PCP Server. Refer to .DHCPv4 refers to the Dynamic Host Configuration Protocol for IPv4.DHCP refers to both DHCPv4 and
DHCPv6 .DHCP client (or client) denotes a node that initiates requests to
obtain configuration parameters from one or more DHCP servers.DHCP server (or server) refers to a node that responds to
requests from DHCP clients.Name is a domain name that contains one or more labels. In
particular, a PCP name may be structured as DNS qualified name or be
composed of strings such as can be passed to getaddrinfo (Section
6.1 of ), including address literals,
etc.Both IP Address and Name DHCP options have been considered in early
stages of this specification. This flexibility aims to let service
providers to make their own engineering choices and use the convenient
option according to their deployment context. Nevertheless, DHC WG's
position is this flexibility has some drawbacks such as inducing errors
(See Section 7 of ). Therefore, only the
Name option is maintained within this document.This document defines an option to carry a name rather than an IP
address. This choice is motivated by operational considerations: In
particular, some Service Providers are considering two levels of
redirection:The first level is national-wise and undertaken by DHCP: a
regional-specific Name will be returned;The second level is done during the resolution of the
regional-specific Name to redirect the customer to a regional PCP
server among a pool deployed regionally.Distinct operational teams are responsible for each of the above
mentioned levels. A clear separation between the functional perimeter of
each team is a sensitive task for the maintenance of the offered
services. Regional teams will require to introduce new resources (e.g.,
new PCP-controlled devices such as Carrier Grade NATs (CGNs, )) to meet an increase
of customer base. Operations related to the introduction of these new
devices (e.g., addressing, redirection, etc.) are implemented locally.
Having this regional separation provides flexibility to manage portions
of network operated by dedicated teams. This two-level redirection can
not be met by the IP Address option.In addition to the operational considerations:The use of the Name for NAT64
might be suitable for load-balancing purposes;For the DS-Lite case , if the
encapsulation mode is used to send PCP messages, an IP address may
be used since the AFTR selection is already done via the AFTR_NAME
DHCPv6 option . Of course, this
assumes that the PCP Server is co-located with the AFTR function. If
these functions are not co-located, conveying the Name would be more
convenient.This DHCPv6 option conveys a domain name to be used to retrieve the
IP addresses of PCP Server(s). Appropriate name resolution queries
should be issued to resolve the conveyed name.The format of the DHCPv6 PCP Server option is shown in .The fields of the option shown in are as follows:Option-code: OPTION_PCP_SERVER (TBA, see )Option-length: Length of the 'PCP Server Domain Name' field in
octets.PCP Server Domain Name(s): The domain name s) of the PCP Server
to be used by the PCP Client. The OPTION_PCP_SERVER option can
include multiple PCP Server Domain Names; each Name is treated as
a separate PCP Server. The domain name(s) is encoded as string.
When several names are included, a space character is used as
separator.To discover a PCP Server ,
the DHCPv6 client MUST include an Option Request Option (ORO)
requesting the DHCPv6 PCP Server Name option as described in Section
22.7 of (i.e., include
OPTION_PCP_SERVER on its OPTION_ORO).If the DHCPv6 client receives an OPTION_PCP_SERVER option from the
DHCPv6 server, it extracts the Name(s) conveyed in the
OPTION_PCP_SERVER option and proceeds to validate it. Once each Name conveyed in the OPTION_PCP_SERVER option is
validated, the DHCPv6 client MUST follow the procedure specified in
.The PCP Server Name DHCPv4 option can be used to configure a name
to be used by the PCP Client to contact a PCP Server. The format of
this option is illustrated in .The description of the fields is as follows:Code: OPTION_PCP_SERVER (TBA, see );Length: Includes the length of the "PCP Server Domain Name"
field in octets; The maximum length is 255 octets.PCP Server Domain Name(s): The domain name(s) of the PCP Server
to be used by the PCP Client when issuing PCP messages. The
OPTION_PCP_SERVER option can include multiple PCP Server Domain
Names; each Name is treated as a separate PCP Server. The domain
name(s) is encoded as strings. When several names are included, a
space character is used as separator.The OPTION_PCP_SERVER DHCPv4 option is a
concatenation-requiring option. As such, the mechanism specified in
MUST be used if the PCP Server Name
option exceeds the maximum DHCPv4 option size of 255 octets.DHCPv4 client expresses the intent to get OPTION_PCP_SERVER by
specifying it in Parameter Request List Option .If the DHCPv4 client receives an OPTION_PCP_SERVER option from the
DHCPv4 server, it extracts the Name(s) conveyed in the option and
proceeds to validating it.Once each Name conveyed in the OPTION_PCP_SERVER option is
validated, the DHCPv4 client MUST follow the procedure specified in
.This section specifies the behavior to be followed by the PCP Client
to contact its PCP Server(s) when receiving one or several PCP Names.
This section is not specific to DHCP; it is applicable to any mechanism
that configures server names.Multiple Names may be configured to a PCP Client in some deployment
contexts such as multi-homing. It is out of scope of this document to
enumerate all deployment scenarios which require multiple Names to be
configured.Each configured Name is passed to the name resolution library
(e.g., Section 6.1.1 of or ) to retrieve the corresponding IP address(es)
(IPv4 or IPv6). Then, the PCP Client MUST follow the procedure
specified in to contact its PCP
Server(s).It is RECOMMENDED to associate a validity lifetime (e.g., TTL of
DNS record if the Name is resolved using DNS) with any address
resulting from resolving the PCP Server Name when stored in a local
name resolution cache. Considerations on how to flush out a local
cache are out of the scope of this document.A host may have multiple network interfaces (e.g, 3G, WiFi, etc.);
each configured differently. Each PCP Server learned MUST be
associated with the interface via which it was learned.This section specifies the behavior to be followed by the PCP
Client to contact its PCP Server(s) when receiving one or several PCP
Names:If only one PCP Name is configured: if a list of IP addresses
is returned as a result of resolving the PCP Server Name, the PCP
Client follows the procedure specified in .If several PCP Names are configured: each Name is treated as a
separate PCP Server. Moreover, each Name may be resolved into one
IP address or a list of IP addresses. The PCP Client contacts in
parallel the first IP address of each Name and follows the
procedure specified in for the list
of IP addresses returned for each Name. provides some examples to illustrate
this procedure.The discovery procedure may result in a PCP Client instantiating multiple
mappings maintained by distinct PCP Servers. The decision to use all
these mappings or delete some of them is deployment-specific. Only the
client can decide whether all the mappings are needed or only a subset
of them.The PCP Client initializes its retransmission timer, RETRY_TIMER,
to 2 seconds. The PCP Client sends its PCP message to the PCP Server
and waits 2 seconds for a response. If no response is received, it
doubles the value of RETRY_TIMER, sends another (identical) PCP
message and waits 2*RETRY_TIMER. This procedure is repeated three
(3) times, doubling the value of RETRY_TIMER each time. If no
response is received after four (4) attempts, the PCP Client tries
with the next IP address in its list of PCP Server addresses. If it
has exhausted its list, the procedure is repeated every fifteen
minutes until the PCP request is successfully answered. If, when
sending PCP requests the PCP Client receives an ICMP error (e.g.,
port unreachable, network unreachable) it SHOULD immediately try the
next IP address in the list. Once the PCP Client has successfully
received a response from a PCP Server address on that interface, it
sends subsequent PCP requests to that same server address until that
PCP Server becomes non-responsive, which causes the PCP client to
attempt to re-iterate the procedure starting with the first PCP
Server address on its list.The following sub-sections provide three examples to illustrate
the procedure.For all these examples, let's suppose pcpserver-x, pcpserver-y
and pcpserver-z are configured as PCP Names.Let's also suppose:The procedure to contact the PCP Servers is as follows:Now, if the following conditions are made:The procedure to contact the PCP Servers lead to the
following:Let's suppose now that:The procedure to contact the PCP Servers is as follows:In some deployment contexts, the PCP Server may be reachable with an
IPv4 address but DHCPv6 is used to provision the PCP Client. In such
scenarios, a plain IPv4 address or an IPv4-mapped IPv6 address can be
configured to reach the PCP Server.A Dual-Stack host may receive OPTION_PCP_SERVER via both DHCPv4 and
DHCPv6. The content of these OPTION_PCP_SERVER options may refer to the
same or distinct PCP Servers. This is deployment-specific and as such it
is out of scope of this document.The security considerations in , and
are to be considered.Authors of this document request the following DHCPv6 option
code:Option NameValueOPTION_PCP_SERVERTBAAuthors of this document request the following DHCPv4 option
code:Option NameValueOPTION_PCP_SERVERTBAMany thanks to B. Volz, C. Jacquenet, R. Maglione, D. Thaler, T.
Mrugalski, T. Lemon and M. Wasserman for their review and comments.