Stateless and Scalable Network Slice Identification for SRv6

Introduction SRv6 Network Programming enables the creation of overlays with underlay optimization to be deployed in an SR domain. As defined in , all inter-domain packets are encapsulated for the part of the packet journey that is within the SR domain. The outer IPv6 header is originated by a node of the SR domain and is destined to a node of the SR domain. This document describes a stateless encoding of slice identification in the outer IPv6 header of an SR domain. The slice identification is independent of topology and the QoS/DiffServ policy of the network, thus enabling scalable network slicing for SRv6 overlays. The definition of network slicing in the context of networks built from IETF technologies is specified in . It defines the term "IETF Network Slice" and establishes the general principles of network slicing in the IETF context. It also discusses the general framework for requesting and operating IETF Network Slices, the characteristics of an IETF Network Slice, the necessary system components and interfaces, and how abstract requests can be mapped to more specific technologies. The document also discusses related considerations with monitoring and security.

Slice Identifier The Slice Identifier (SLID) is a value encoded within the IPv6 packet that allows transit routers to process the packet according to network slice-based policy. An example of slice-based policy that can be enforced using the SLID is described in . The SLID may identify a unique IETF network slice or a group of slices that share the same policy. For example, a SLID may identify a slice aggregate . This document proposes to encode the SLID in a portion of the IPv6 Flow Label. The precise SLID location within the IPv6 Flow Label and the number of bits used to encode it are governed by local policy and uniform within the SR domain.

SLID Presence Indicator The SLID Presence Indicator (SPI) is set by a SLID-capable IPv6 source node to inform transit routers that a SLID is encoded in the packet. The SPI is encoded as a specific bit or range of values in the Traffic Class field of the IPv6 header. The encoding of the SPI in the IPv6 header is governed by local policy and uniform within the SR domain.

Ingress PE SLID Assignment When an ingress PE receives a packet that traverses the SR domain, it encapsulates the packet in an outer IPv6 header and optional SRH as defined in . The ingress PE MAY also classify the packet into a slice and set the slice identifier as follows: Set the SPI in the outer IPv6 header. Write the SLID in the outer IPv6 header. The slice classification method is outside the scope of this document.

Per-Slice Forwarding Any router within the SR domain that forwards a packet with SPI bit set uses the SLID to select a slice and apply per-slice policies. There are many different policies that could define a slice for a particular application or service. The most basic of these is bandwidth-allocation, an implementation complying with this specification SHOULD support the bandwidth-allocation slice as defined in the next section.

Bandwidth-Allocation Slice A per-slice policy is configured at each interface of each router in the SR domain, with one traffic shaper per SLID. The bitrate of each shaper is configured to reflect the bandwidth allocation of the per-slice policy. If shapers are not available, or desirable, an implementation MAY configure one scheduling queue per SLID with a guaranteed bandwidth equal to the bandwidth-allocation for the slice. This option allows a slice to consume more bandwidth than its allocation when available. Per-slice shapers or queues effectively provides a virtual port per slice. This solution MAY be complemented with a per-virtual-port hierarchical DiffServ policy. Within the context of one specific slice, packets are further classified into children DiffServ queues which hang from the virtual port. The DSCP value in the IPv6 header SHOULD be used for queue selection.

Backward Compatibility The Flow Label usage described in this document is consistent with and . PE routers that do not set the SPI do not enable the SLID semantic of the Flow Label bits. Hence, SLID-aware routers would not attempt to classify these packets into a slice. Any router that does not process the SPI nor the SLID forwards packets as usual.

Acknowledgements The authors would like to thank Darren Dukes, Ketan Talaulikar, Jisu Bhattacharya, John Bettink, Aman Manot, and David Melman for their insightful feedback on this document.