L3ND Upper-Layer Protocol Configuration Arrcus & IIJ
5147 Crystal Springs Bainbridge Island WA 98110 US randy@psg.com
Arrcus
2077 Gateway Place, Suite #400 San Jose CA 95119 United States of America keyur@arrcus.com
This document adds PDUs to the Layer-3 Neighbor Discovery protocol to communicate the parameters needed to exchange inter-device Upper Layer Protocol Configuration for upper-layer protocols such as the BGP family. The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.
Massive Data Centers (MDCs) which use upper-layer protocols such as BGP4 and other routing protocols may use the Layer-3 Neighbor Discovery Protocol, L3ND, to reveal the inter-device links of the topology. It is desirable for devices to facilitate the configuration parameters of those upper layer protocols to enable more hands-free configuration. This document defines a new L3ND PDU to communicate these Upper-Layer Protocol Configuration parameters.
The reader is assumed to have read Layer-3 Neighbor Discovery . The terminology and PDUs there are assumed here. Familiarity with the BGP4 Protocol is assumed.
To communicate parameters required to configure peering and operation of Upper-Layer Protocols at IP layer-3 and above, e.g., BGP sessions on a link, a neutral sub-TLV based Upper-Layer Protocol PDU is defined as follows:
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Version = 0 | Type = 8 | Payload Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | ULPC Type | AttrCount | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Serial Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Attribute List ... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The Version, Type, and Payload Length as defined in apply to this PDU. The BGP Authentication sub-TLV provides for provisioning MD5, which is a quite weak hash, horribly out of fashion, and kills puppies. But, like it or not, it has been sufficient against the kinds of attacks BGP TCP sessions have endured. So it is what BGP deployments use. As the ULPC PDU may contain keying material, e.g. , it SHOULD BE over TLS. ULPC Type: A one byte integer denoting the type of the upper-layer protocol Reserved BGP Reserved The one octet AttrCount is the number of attribute sub-TLVs in the Attribute List. The Attribute List is a, possibly null, set of sub-TLVs describing the configuration attributes of the specific upper-layer protocol. An Attribute consists of a one octet Attribute Type, a one octet Attribute Length of the number of octets in the Attribute, and a Payload of arbitrary length up to 253 octets.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Attr Type = 1 | Attr Len | Payload | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The parameters needed for BGP peering on a link are exchanged in sub-TLVs within an Upper-Layer Protocol PDU. The following describe the various sub-TLVs for BGP. The goal is to provide the minimal set of configuration parameters needed by BGP OPEN to successfully start a BGP peering. The goal is specifically not to replace or conflict with data exchanged during BGP OPEN. Multiple sources of truth are a recipe for complexity and hence pain. If there are multiple BGP sessions on a link, e.g., IPv4 and IPv6, then multiple sets of BGP sub-TLVs MAY BE exchanged within the BGP ULPC PDU or sepatate BGP ULPC PDUs may be sent, one for each address family. A peer receiving BGP ULPC PDUs has only one active BGP ULPC PDU for an particular address family on a specific link at any point in time; receipt of a new BGP ULPC PDU for a particular address family replaces any previous one. If there are one or more open BGP sessions, receipt of a new BGP ULPC PDU SHOULD not affect these sessions and the PDU SHOULD be discarded. If a peer wishes to replace an open BGP session, they MUST first close the running BGP session and then send a new BGP ULPC PDU. As a link may have multiple encapsulations and multiple addresses for an IP encapsulation, which address of which encapsulation is to be used for the BGP session MUST be specified. For each BGP peering on a link here MUST be one agreed encapsulation, and the addresses used MUST be in the corresponding L3ND IPv4/IPv6 Encapsulation PDUs. If the choice is ambiguous, an Attribute may be used to signal preferences. If a peering address has been announced as a loopback, i.e. MUST BE flagged as such in the L3ND Encapsulation PDU (see Sec. 10.2), a two or three hop BGP session MUST be established. Otherwise a direct one hop session is used. The BGP session to a loopback will forward to the peer's address which was marked as Primary in the L3ND Encapsulation Flags, iff it is in a subnet which is shared with both BGP speakers. If the primary is not in a common subnet, then the BGP speaker MAY pick a forwarding next hop that is in a subnet they share. If there are multiple choices, the BGP speaker SHOULD have signaled which subnet to choose in an Upper-Layer Protocol Configuration PDU Attribute. Attributes MUST be unique in the Attribute List. I.e. a particular Attr Type MUST NOT occur more than once in the Attribute List. If a ULPC PDU is received with multiple occurances of the same Attr Type, an Error ACK should be returned.
The four octet Autonomous System number MUST be specified. If the AS Number is less than 32 bits, it is padded with high order zeros.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Attr Type = 1 | Attr Len = 4 | My ASN ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ ~ | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The BGP IPv4 Address sub-TLV announces the sender's four octet IPv4 BGP peering source address and one octet Prefix Lenth to be used by the receiver. At least one of IPv4 or IPv6 BGP source addresses MUST be announced. As usual, the BGP OPEN capability negotiation will determine the AFI/SAFIs to be transported over the peering, see .
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Attr Type = 2 | Attr Len = 5 | My IPv4 Peering Address ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ ~ | Prefix Len | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The BGP IPv6 Address sub-TLV announces the sender's 16 octet IPv6 BGP peering source address and one octet Prefix Length to be used by the receiver. At least one of IPv4 or IPv6 BGP source addresses MUST be announced. As usual, the BGP OPEN capability negotiation will determine the AFI/SAFIs to be transported over the peering, see .
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Attr Type = 3 | Attr Len = 19 | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | | + + | My IPv6 Peering Address | + + | | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | Prefix Len | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The BGP Authentication sub-TLV provides any authentication data needed to OPEN the BGP session. Depending on operator configuration of the environment, it might be a simple MD5 key (see ), the name of a key chain in a KARP database (see ), or one of multiple Authentication sub-TLVs to support hop.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Attr Type = 4 | Attr Len | ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ ~ ~ BGP Authentication Data ... ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The BGP session OPEN has extensive, and a bit complex, capability negotiation facilities. In case one or more extra attributes might be needed, the two octet BGP Miscellaneous Flags sub-TLV may be used.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Attr Type = 5 | Attr Len = 2 | Misc Flags | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Misc Flags: GTSM Must be zero The GTSM flag, when 1, indicates that the sender wishes to enable the Generalized TTL Security Mechanism for the session.
All the Security considerations of apply to this PDU. As the ULPC PDU may contain keying material, see , it SHOULD BE over TLS, not clear TCP. Any keying material in the PDU SHOULD BE salted and hashed. The BGP Authentication sub-TLV provides for provisioning MD5, which is a quite weak hash, horribly out of fashion, and kills puppies. But, like it or not, it has been sufficient against the kinds of attacks BGP TCP sessions have endured. So it is what BGP deployments use.
This document requests the IANA create a new entry in the L3ND PDU Type registry as follows:
PDU Code PDU Name ---- ------------------- 9 ULPC
This document requests the IANA create a registry for L3ND ULPC Type, which may range from 0 to 255. The name of the registry should be L3ND-ULPC-Type. The policy for adding to the registry is RFC Required per , either standards track or experimental. The initial entries should be the following:
Value Name ----- ------------------- 0 Reserved 1 BGP 2-255 Reserved
The authors thank Rob Austein, Sue Hares, and Russ Housley.