VXLAN-GPE Encapsulation for In-situ OAM Data

In-situ OAM (IOAM) records OAM information within the packet while the packet traverses a particular network domain. The term "in-situ" refers to the fact that the IOAM data fields are added to the data packets rather than being sent within packets specifically dedicated to OAM. This document defines how IOAM data fields are transported as part of the VXLAN-GPE encapsulation. The IOAM data fields are defined in . An implementation of IOAM which leverages VXLAN-GPE to carry the IOAM data is available from the FD.io open source software project .

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 .

Abbreviations used in this document: In-situ Operations, Administration, and Maintenance Operations, Administration, and Maintenance Virtual eXtensible Local Area Network, Generic Protocol Extension

VXLAN-GPE is defined in . IOAM data fields are carried in VXLAN-GPE using a next protocol value of TBD_IOAM. An IOAM header is added containing the different IOAM data fields defined in . In an administrative domain where IOAM is used, insertion of the IOAM header in VXLAN-GPE is enabled at the VXLAN-GPE tunnel endpoints, which also serve as IOAM encapsulating/decapsulating nodes by means of configuration.

The VXLAN-GPE header and fields are defined in . The VXLAN Next Protocol value for IOAM is TBD_IOAM. The IOAM related fields in VXLAN-GPE are defined as follows: 8-bit field defining the IOAM Option type, as defined in Section 7.2 of . 8-bit unsigned integer. Length of the IOAM HDR in 4-octet units not including the first 4 octects. 8-bit reserved field MUST be set to zero upon transmission and ignored upon receipt. 8-bit unsigned integer that determines the type of header following IOAM protocol. The value is from the IANA registry setup for VXLAN GPE Next Protocol defined in . IOAM option header and data is present as specified by the IOAM-Type field, and is defined in Section 4 of . Multiple IOAM options MAY be included within the VXLAN-GPE encapsulation. For example, if a VXLAN-GPE encapsulation contains two IOAM options before a data payload, the Next Protocol field of the first IOAM option will contain the value of TBD_IOAM, while the Next Protocol field of the second IOAM option will contain the VXLAN "Next Protocol" number indicating the type of the data payload.

This section summarizes a set of considerations on the overall approach taken for IOAM data encapsulation in VXLAN-GPE, as well as deployment considerations.

This section is to support the working group discussion in selecting the most appropriate approach for encapsulating IOAM data fields in VXLAN-GPE. An encapsulation of IOAM data fields in VXLAN-GPE should be friendly to an implementation in both hardware as well as software forwarders. Hardware forwarders benefit from an encapsulation that minimizes iterative look-ups of fields within the packet: Any operation which looks up the value of a field within the packet, based on which another lookup is performed, consumes additional gates and time in an implementation - both of which are desired to be kept to a minimum. This means that flat TLV structures are to be preferred over nested TLV structures. IOAM data fields are grouped into three option categories: Trace, proof-of-transit, and edge-to-edge. Each of these three options defines a TLV structure. A hardware-friendly encapsulation approach avoids grouping these three option categories into yet another TLV structure, but would rather carry the options as a serial sequence. Two approaches for encapsulating IOAM data fields in VXLAN-GPE could be considered: Use a single GPE protocol type for all IOAM types: IOAM would receive a single GPE protocol type code point. A "sub-type" field would then specify what IOAM options type (trace, proof-of-transit, edge-to-edge) is carried. Use one GPE protocol type per IOAM options type: Each IOAM data field option (trace, proof-of-transit, and edge-to-edge) would be specified by its own "next protocol", i.e. each IOAM options type becomes its own GPE protocol type with a dedicated code point. This implies that in case additional IOAM option types would be added in the future, additional GPE protocol type code points would need to be allocated. The first option has been chosen here. Multiple back-to-back IOAM options can be encoded as a succession of IOAM headers, with the same single GPE protocol type appearing as the next protocol before each IOAM header, but different sub-types within each IOAM header.

defines an "O bit" for OAM packets. Per the O bit indicates that the packet contains an OAM message instead of data payload. Packets that carry IOAM data fields in addition to regular data payload / customer traffic must not set the O bit. Packets that carry only IOAM data fields without any payload must set the O bit.

If IOAM is deployed in domains where UDP port numbers are not controlled and do not have a domain-wide meaning, such as on the global Internet, transit devices MUST NOT attempt to modify the IOAM data contained in the IOAM header following the VXLAN-GPE header. In case UDP port numbers are not controlled there might be UDP packets specifying the same UDP port number that VXLAN-GPE utilizes, i.e. 4790, but with a payload that is not VXLAN-GPE. The scenario and associated reasoning is discussed in which states that "it is important to recognize that any interpretation of port numbers -- except at the endpoints -- may be incorrect, because port numbers are meaningful only at the endpoints."

IANA is requested to allocate a value in the VXLAN-GPE "Next Protocol" registry for IOAM, which is defined in .

IANA is requested to allocate a value in the LISP-GPE "Next Protocol" registry for IOAM, which is defined in .

The security considerations of VXLAN-GPE are discussed in , and the security considerations of IOAM in general are discussed in . IOAM is considered a "per domain" feature, where one or several operators decide on leveraging and configuring IOAM according to their needs. Still, operators need to properly secure the IOAM domain to avoid malicious configuration and use, which could include injecting malicious IOAM packets into a domain.

The authors would like to thank Eric Vyncke, Nalini Elkins, Srihari Raghavan, Ranganathan T S, Karthik Babu Harichandra Babu, Akshaya Nadahalli, Stefano Previdi, Hemant Singh, Erik Nordmark, LJ Wobker, and Andrew Yourtchenko for the comments and advice.