[doc/voq]: Add HLD for BGP in the VoQ chassis - #674
Merged
Conversation
Signed-off-by: jmmikkel@arista.com
|
|
eswaranb
reviewed
Sep 16, 2020
| # 1 Requirements | ||
|
|
||
| Traffic ingressing on any FSI should be routed the same, regardless of which | ||
| FSIs learned the route from their eBGP peers. |
Contributor
There was a problem hiding this comment.
Replace FSI with Asic Instance
eswaranb
reviewed
Sep 16, 2020
| - Each FSI must advertise eBGP-learned routes to the other FSIs in the chassis. | ||
| - Each FSI in a chassis should choose the same BGP route for each prefix. | ||
| - When ECMP is enabled and equal-cost routes are learned by any FSI, all FSIs should compute the same ECMP nexthop. | ||
|
|
Contributor
There was a problem hiding this comment.
Add a note on preserving the nexthop and the AS_PATH
eswaranb
reviewed
Sep 16, 2020
|
|
||
| # 1 Requirements | ||
|
|
||
| Traffic ingressing on any FSI should be routed the same, regardless of which |
Contributor
There was a problem hiding this comment.
Clarify that forwarding behavior can be different, but the nexthop set for the ECMP will be same
Contributor
There was a problem hiding this comment.
Clarify that this is substantially different from an internal-ethernet-clos based chassis because the forwarding decision is made once at the ingress in voq chassis and several times in an ethernet based chassis.
eswaranb
reviewed
Sep 16, 2020
| - Each FSI must advertise eBGP-learned routes to the other FSIs in the chassis. | ||
| - Each FSI in a chassis should choose the same BGP route for each prefix. | ||
| - When ECMP is enabled and equal-cost routes are learned by any FSI, all FSIs should compute the same ECMP nexthop. | ||
|
|
Contributor
There was a problem hiding this comment.
Restriction: The external routers are connected via eBGP and not iBGP
eswaranb
reviewed
Sep 16, 2020
| To ensure that each FSI computes the same ECMP group: | ||
| - Enable additional-path send all for each chassis iBGP peer. If FSI1 learns a route from both R1 and R4, both must be advertised to other FSIs. | ||
| - Allow BGP to form ECMP groups with paths learned from both eBGP and iBGP peers. The best path algorithm normally prefers eBGP paths (RFC 4271 section 9.1.2.2 step d). Without this change, if a route is learned from R1, R2, and R4, FSI1 would create ECMP group {R1,R4}, FSI2 would use {R2}, and FSI3 would use {R1,R2,R4}. | ||
| - The maximum ECMP group size must be set the same for eBGP and iBGP. |
Contributor
There was a problem hiding this comment.
Expand this point a bit more.
eswaranb
reviewed
Sep 16, 2020
| - `neighbor <neighbor> addpath-tx-all-paths` | ||
| - `bgp bestpath peer-type multipath-relax` | ||
| - `maximum-paths ibgp <n>` (where <n> is the same as used in `maximum-paths <n>`) | ||
| - `bgp disable-ebgp-connected-route-check` |
Contributor
There was a problem hiding this comment.
This can change existing eBGP behavior. Need to note down what FRR does today when the nexthop is not over a connected route.
eswaranb
reviewed
Sep 16, 2020
| <a:ASN>64542</a:ASN> | ||
| <a:Hostname>OCPSCH0104001MS</a:Hostname> | ||
| <a:RouteMaps/> | ||
| <a:VoQChassisPeer>1</a:VoQChassisPeer> |
Contributor
There was a problem hiding this comment.
Maybe add this to BGP_SESSION definition
eswaranb
reviewed
Sep 16, 2020
| <a:VoQChassisPeer>1</a:VoQChassisPeer> | ||
| </a:BGPRouterDeclaration> | ||
| ``` | ||
|
|
Contributor
There was a problem hiding this comment.
Add a Cli section
- "show ip bgp" shows both internal and external today, should we update the command to show only the external sessions and add a new command to include the internal.
- A flag that brings up the instance with bgp sessions shut down must not do it for the internal BGP sessions.
- 'shutdown' should bring down only the external sessions, not the internal BGP sessions. We need an additional token to also shutdown the internal sessions if needed.
The functional changes are: - Propose to modify bgpd's publish of nexthop groups to RIB instead of using "bgp disable-ebgp-connected-route-check". - Change the new element in the minigraph to BGPSession>VoQChassisInternal. - Add discussion of BGP commands in the CLI.
This was referenced Oct 13, 2020
Collaborator
lguohan
approved these changes
Oct 22, 2020
minionatwork
approved these changes
Oct 28, 2020
Contributor
eswaranb
approved these changes
Mar 10, 2021
arlakshm
approved these changes
Apr 5, 2021
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This document describes how BGP will be configured between each FSI in the chassis.