Internet-Draft | RIFT Auto-FR | July 2023 |
Head, et al. | Expires 11 January 2024 | [Page] |
This document specifies procedures where RIFT can automatically provision IS-IS Flood Reflection topologies by leveraging its native no-touch ZTP architecture.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 11 January 2024.¶
Copyright (c) 2023 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
[RIFT] is a protocol that focuses heavily on operational simplicity. It natively supports Zero Touch Provisioning (ZTP) functionality that allows each node to automatically derive its place in the topology and configure itself accordingly when properly cabled as a Clos, Fat-Tree, or other similarly structured variant.¶
IS-IS Flood Reflection [RFC9377] is a mechanism that enables flat single-area Level 2 IS-IS topologies to scale well beyond their typical properties when deployed in similar topological structures by:¶
RIFT Auto-Flood Reflection (Auto-FR) combines these technologies by using RIFT's ZTP functionality in order to automatically provision IS-IS Flood Reflection topologies in a completely distributed fashion.¶
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 [RFC2119].¶
IS-IS Flood Reflection operates using Flood Reflectors at the top of the fabric and Flood Reflector Clients at the bottom of the fabric. Any nodes in the middle are not required to support Flood Reflection functionality, nor do they need to support Auto-FR.¶
Nodes in a Flood Reflection topology require specific variables for deployment. For example, a Cluster ID that is unique to the particular fabric or loopback addresses that are unique to a particular node. RIFT has enough topological information to derive these variables with the appropriate scope in a distributed fashion automatically.¶
Once the Flood Reflection topology is built, RIFT Key-Value TIEs can be used to distribute operational state information to allow for cluster-wide validation without any additional tooling.¶
Auto-FR requires that each node understands its given role within the scope of the Flood Reflection deployment, so each node derives the necessary variables and resulting configuration.¶
Not all nodes have to participate in Auto-FR, however, if a node does assume an Auto-FR role, it MUST be capable of deriving the following variables:¶
This section defines an Auto-FR role whereby some ToF (Top-of-Fabric) nodes act as IS-IS Flood Reflectors. It is expected that Flood Reflectors will establish Level 2 IS-IS adjacencies with Flood Reflector Clients in the same area, in the same fabric. The typical Flood Reflector requirements do not change, however, determining which specific values to use requires further consideration.¶
ToF nodes performing Flood Reflector functionality MUST derive the following variables:¶
In the context of Auto-FR, a "spine" can simply be desribed as an Auto-FR node not running as a Flood Reflector (ToF) or Flood Reflector Client (leaf). While RIFT Auto-FR is not required to run on spine nodes and as such a spine role has no specific variables to derive at this time, it may be desirable to do so to facilitate the use of Auto-FR specific Key-Value TIEs.¶
This section also serves as a placeholder for future considerations and enhancements.¶
Although no specific variables for Flood Reflector Clients are described at this time, the generic role is specified as a placeholder for future enhancements.¶
As previously mentioned, not all nodes are required to derive all variables in a network. For example, spine nodes may only be required for transit traffic and not need to support Auto-FR at all. All variables are derived from RIFT's FSM or ZTP mechanism, so no additional flooding other than RIFT's typical flooding is required.¶
It is also important to mention that all variable derivation is in some way based on the RIFT System ID and/or Cluster ID and MUST comply precisely with calculation methods specified in the Auto-FR Variable Derivation section to allow interoperability between different implementations. All necessary foundational code elements are also mentioned there.¶
The 64-bit RIFT System ID that uniquely identifies a node as defined in RIFT [RIFT]. This not derived specifically for Auto-FR, but for all RIFT nodes and is used in the derivation procedures described in this section.¶
This section describes extensions to both the RIFT LIE and Node-TIE packet schemas in the form of a 16-bit value that identifies the Auto-FR Version. Auto-FR capable nodes MUST support this extension, but MAY choose not to advertise it in LIEs and Node-TIEs when Auto-FR is not being utilized.¶
The complete encoding.thrift schema documented in RIFT [RIFT] describes both major and minor protocol/schema versions. Auto-FR Version calculation is done by multiiplying a static value of 256 by the major version then adding the minor version, that is to say - 256 * MAJOR + MINOR.¶
This section also describes an extension to the NodeCapabilities schema indicating whether or not a node supports Auto-FR.¶
The appendix (Appendix B) details necessary changes to the LIEPacket, NodeTIEElement, and NodeCapabilities Thrift schemas.¶
This section describes extensions to both the RIFT LIE and Node-TIE packet schemas in the form of a 32-bit value that identifies the Auto-FR Cluster ID. Auto-FR capable nodes MUST support this extension, but MAY choose not to advertise it in LIEs and Node-TIEs when Auto-FR is not being utilized.¶
Deployments using more than one Auto-FR cluster MUST use different Cluster IDs. Failure to do so may cause sub-optimal routing as L1/L2 nodes from different clusters would belong to the same subnet.¶
A Cluster ID with a value of 0 is considered invalid and MUST NOT be used for any purpose.¶
The appendix (Appendix B) details necessary changes to the LIEPacket and NodeTIEElement Thrift schemas.¶
This section describes extensions to the Node-TIE packet schema in the form of a 32-bit value that indicates a Flood Reflection Preference value to be used during Flood Reflector election procedures with the higher value being preferred. Auto-FR capable nodes MUST support this extension.¶
The appendix (Appendix B) details necessary changes to the NodeTIEElement Thrift schemas.¶
Auto-FR nodes MUST derive a unique 8-byte IS-IS System ID for use in deriving the IS-IS NET Address. Calculation is done using the 8-byte RIFT System ID and 4-byte Cluster ID.¶
In order for nodes to derive an IS-IS System ID, the following algorithms are required - auto_fr_cidsid2isissid (Figure 11) and auto_fr_v6hash (Figure 16).¶
Auto-FR nodes MUST derive a unique 10-byte IS-IS NET (Network Entity Title) Address to uniquely identify itself within the Flood Reflection topology. The 1st byte (which indicates the AFI) MUST have a value of "49". The last byte (i.e. the NSEL) MUST have a value of 0. Remaining calculation is done using the 8-byte RIFT System ID and 4-byte Cluster ID.¶
In order for nodes to derive an IS-IS NET Address, the following algorithms are required - auto_fr_cidsid2isisnet (Figure 10) and auto_fr_cidsid2isissid (Figure 11).¶
Auto-FR nodes MUST derive a ULA-scoped IPv6 loopback address to be used in IS-IS. Calculation is done using the 6-bytes of reserved ULA space, the 4-byte Cluster ID, and the node's 8-byte RIFT System ID. Derivation of the IS-IS System ID varies slightly depending upon the node's location/role in the fabric and will be described in subsequent sections.¶
Leaf Nodes acting as Flood Reflector Clients MUST derive their loopback address according to the specific section describing the algorithm. Calculation is done using the 6-bytes of reserved ULA space, the 4-byte Cluster ID, and the 8-byte RIFT System ID.¶
In order for leaf nodes to derive IPv6 loopbacks, the following algorithms are required - auto_fr_cidsidv6loopback (Figure 13) and auto_fr_v6prefixcidsid2loopback (Figure 17).¶
IPv4 addresses MAY be supported, but it should be noted that they have a higher likelihood of collision. The appendix contains the required auto_fr_cidsid2v4loopback (Figure 12) algorithm to support IPv4 loopback derivation.¶
ToF nodes acting as Flood Reflectors MUST derive their loopback address according to the specific section describing the algorithm. Calculation is done using the 6-bytes of reserved ULA space, the 4-byte Cluster ID, and the 8-byte RIFT System ID.¶
In order for ToF nodes to derive IPv6 loopbacks, the following algorithms are required - auto_fr_cidsidv6loopback (Figure 13), auto_fr_v6prefixcidsid2loopback (Figure 17), and auto_fr_cidfrpref2frv6loopback (Figure 9).¶
IPv4 addresses MAY be supported, but it should be noted that they have a higher likelihood of collision. The appendix contains the required auto_fr_cidfrpref2frv4loopback (Figure 8) algorithm to support IPv4 loopback derivation.¶
Each ToF performs the election independently based on the RIFT System IDs and a Flood Reflection preference value of other ToF nodes in the fabric obtained via southbound reflection. The Flood Reflector election procedures are defined as follows:¶
This ordering is necessary to prevent a single node with either the highest or lowest System ID from triggering changes to flood reflector loopback addresses as it would result in all IS-IS adjacencies flapping.¶
For example, if ToF01 (System ID: 002c6af5a281c000 / FR Preference: 100) and ToF02 (System ID: 002c6bf5788fc000 / FR Preference: 1) went through the election process, ToF02 would be elected due to it having the highest System ID. If a ToF determines that it is elected as a Flood Reflector, it uses the knowledge of its position in the list to derive Flood Reflector IPv6 loopback address.¶
A topology MUST elect at least 1 ToF node as an IS-IS Flood Reflector, but SHOULD elect 3. The election process varies depending upon whether or not the topology is comprised of a single plane or multiple planes. The multiplane election procedure will be described in a future version of this document.¶
The algorithm shown in "auto_fr_sids2frs" (Figure 14) is required to perform the Flood Reflector election procedures.¶
RIFT FSM adjacency rules MUST consider Auto-FR Version (Section 4.2) and Auto-FR Cluster ID (Section 4.3) values so that nodes that do not support Auto-FR can interoperate with nodes that do. The LIE validation is extended with the following clause and if it is not met, miscabling should be declared:¶
(if auto_flood_reflection_version is not advertised by either node OR if auto_flood_reflection_version is identical on both nodes) AND (auto_flood_reflection_cluster_id is not advertised by either node OR auto_flood_reflection_cluster_id is identical on both nodes)¶
All nodes utilizing Auto-FR MUST advertise their Auto-FR Version (Section 4.2), Flood Reflection Cluster ID (Section 4.3), and Flood Reflection Preference (Section 4.4) values in at least one Node-TIE in each direction (i.e. North and South).¶
To fully realize the benefits of Auto-FR, it may help to describe the high-level method. Simply put, RIFT automatically provisions the underlay and Auto-FR provisions the Flood Reflection topology. The goal of this section is to draw simple lines between general fabric concepts, RIFT, and Auto-FR and how they fit into current network designs and practices.¶
This section also describes a set of optional Key-Value TIEs [RIFT-KV] that leverages the variables that have already been derived to provide further operational enhancement to the operator.¶
Figure 1 illustrates a typical 5-stage Clos IP fabric. Each node is named and labelled in such a way that conveys:¶
Table 1 should help further align these concepts.¶
RIFT Placement | IS-IS Level | IS-IS FR Role |
---|---|---|
ToF Nodes | L1/L2 | Flood Reflector |
Spine Nodes | L1 | N/A |
Leaf Nodes | L1/L2 | Flood Reflector Client |
Connections between various nodes can be understood in two different ways:¶
It is important to remember that Auto-FR is not altering the way in which IS-IS Flood Reflection operates in any way, it simply takes existing deployment scenarios and simplifies the provisioning process.¶
Leaf nodes MAY optionally advertise analytics information about the Auto-FR fabric to ToF nodes using RIFT Key-Value TIEs [RIFT-KV]. This may be helpful in that validation and troubleshooting activities can be performed on the ToF nodes rather than manually verifying the state separately on multiple leaf nodes.¶
This Key/Value pair describes node level information within the context of the Flood Reflection topology. The RIFT System ID of the advertising leaf node MUST be used to differentiate the node among other nodes in the fabric.¶
The Auto-FR Analytics (Global) Key/Value pair MUST be advertised with the 3rd and 4th bytes of the Key Identifier consisting of all 0s.¶
where:¶
A REQUIRED value indicating the node's Auto-FR role within the fabric.¶
Implementations leveraging Thrift for Key-Value functionality SHOULD refer to the auto_flood_reflection_kv.thrift (Appendix C) schema in the appendix.¶
This section requests the following suggested values from the RIFT Well-Known Key-Type Registry.¶
Value | Key-Identifier | Description | Status/Reference |
---|---|---|---|
5 | Auto-FR Analytics (Global) | Key/Value pair containing operational state of a Flood Reflector Client node. | This document. |
This document introduces no new security concerns to RIFT or other specifications referenced in this document as RIFT natively secures LIE and TIE packets as described in [RIFT].¶
This section will be used to acknowledge major contributors.¶
This section serves as a placeholder for a YANG model supporting RIFT Auto-FR.¶
This section contains the normative Thrift models required to support Auto-FR. Per the main RIFT [RIFT] specification, all signed values MUST be interpreted as unsigned values.¶
This section specifies extensions to RIFT common.thrift model.¶
These extensions are REQUIRED in order to support Auto-FR.¶
This section specifies extensions to RIFT encoding.thrift model.¶
These extensions are REQUIRED in order to support Auto-FR.¶
This section specifies common_flood_reflection.thrift as a normative THrift model to support core functionality for Auto-FR.¶
These extensions are REQUIRED in order to support Auto-FR.¶
This section defines auto_flood_reflection_kv.thrift as a method of supporting Auto-FR analytics functionality.¶
This section contains the normative variable derivation algorithms that are REQUIRED to support Auto-FR.¶