BGP Configuration Guide for Cisco 8000 Series Routers, Cisco IOS XR Releases

PDF

BGP slow peer automatic isolation from update group

Updated: February 28, 2026

Want to summarize with AI?

Log in

Overview

Explains automatic detection and isolation of slow peers into dedicated slow update groups to prevent group-wide stalls and allow non-slow peers to continue processing updates.

BGP slow peer automatic isolation from update group is a BGP feature that

  • detects neighbors in an update group that cannot keep up with sustained update generation over time

  • automatically moves detected slow peers into a dedicated slow update group and returns them to the original group upon recovery, and

  • prevents group-wide stalls by isolating slow peers, allowing non-slow members to continue processing new updates.

Slow update groups: structure and defaults

Table 1. Feature History Table

Feature Name

Release Information

Feature Description

BGP Slow Peer Automatic Isolation from Update Group

 Release 25.4.1

Introduced in this release on: Fixed Systems (8010 [ASIC: A100])(select variants only*)

*This feature is now supported on:

  • 8011-32Y8L2H2FH

  • 8011-12G12X4Y-A

  • 8011-12G12X4Y-D

BGP Slow Peer Automatic Isolation from Update Group

 Release 25.1.1

Introduced in this release on: Fixed Systems (8010 [ASIC: A100])(select variants only*)

*This feature is supported on Cisco 8011-4G24Y4H-I routers.

BGP Slow Peer Automatic Isolation from Update Group

 Release 24.4.1

Introduced in this release on: Fixed Systems (8200 [ASIC: P100], 8700 [ASIC: P100, K100])(select variants only); Modular Systems (8800 [LC ASIC: P100])(select variants only*)

*This feature is supported on:

  • 8212-48FH-M

  • 8711-32FH-M

  • 8712-MOD-M

  • 88-LC1-36EH

  • 88-LC1-12TH24FH-E

  • 88-LC1-52Y8H-EM

BGP Slow Peer Automatic Isolation from Update Group

 Release 7.3.1

A slow peer cannot keep up with the rate at which the router generates BGP update messages over a period of time, in an update group. This feature automatically detects a slow peer in an update group and moves it to a new update group. The feature is enabled on the router, by default.

New commands:

  • slow-peer detection enable

  • clear bgp slow-peers

Updated commands:

  • Short-lived churn: A peer that briefly falls behind during events such as connection resets but quickly recovers is not treated as a slow peer.

  • Impact of slow peers: The presence of a slow peer increases the number of formatted updates pending transmission in the original update group.

  • Group structure:

    • One slow update group exists for each original update group that contains slow peers.

    • Multiple slow peers in the same original group are isolated in separate sub-groups within that slow update group.

    • Slow peers from different original update groups cannot be combined because outbound policy configurations differ.

  • Defaults: The feature is enabled by default, and automatic splitting of update groups is enabled by default.

How automatic isolation works

Summary

The key components involved in the process are:

  • Original update group: Formats messages once and transmits to all members; deletes messages only after all members acknowledge receipt.

  • Slow update group: Dedicated group created when one or more peers in the original group are detected as slow.

  • Slow peer sub-groups: Per-peer sub-groups within the slow update group used to process each slow peer independently.

  • Recovery action: Moves peers back to the original update group when they catch up.

Automatic isolation mitigates backlog and contention in update groups by detecting slow peers, processing their updates separately in a dedicated slow update group, and restoring them when they recover.

Workflow

These stages describe how automatic isolation works:

  1. Detection: The system detects that a peer in the original update group cannot keep pace, causing formatted messages to accumulate.
  2. Isolation: The detected slow peer is moved automatically to a new slow update group; if multiple peers are slow, each is placed in its own slow sub-group.
  3. Parallel processing: The parent (original) update group continues sending newly modified nets to non-slow peers, while the slow update group processes backlogged updates for slow peers.
  4. Recovery: When a slow peer completes processing and catches up, it is moved back to the original update group.

Result

Non-slow peers advance at their regular pace, while slow peers process updates in isolation. This prevents stalls and reduces the risk of backlog growth across the entire group.

Configure slow peer automatic isolation

Detect, isolate, and manage slow BGP peers by configuring global and neighbor-level controls, and verify slow-peer status and queues.

Automatic isolation moves detected slow peers to a dedicated slow update group and returns them to the original group after recovery. Dynamic detection can be enabled globally or per neighbor address family; static marking isolates a specific peer immediately.

Before you begin

  • Identify the BGP autonomous system number.

  • Determine the neighbor address and address family (AFI/SAFI).

  • Decide whether to disable global detection, mark a peer as static, or enable dynamic detection with the permanent option.

Do one or more of the following to configure and manage slow peers.

Procedure

1.

Disable slow peer detection globally.

Example:

Router# configure
Router(config)# slow-peer-detection disable

Any slow peers that are detected are marked as normal peers and moved back to their original update groups. No more slow peers are detected.
2.

Mark a neighbor as a static slow peer at the neighbor address-family level.

Example:


Router(config)# router bgp 5
Router(config-bgp)# address-family ipv4
Router(config-bgp-af)# neighbor 172.60.2.3
Router(config-bgp-nbr-af)# slow-peer detection disable split-updategroup static

The peer becomes part of the slow update group.
3.

Enable dynamic detection with permanent option at the neighbor address-family level.

Example:

Router(config)# router bgp 5
Router(config-bgp)# address-family ipv4
Router(config-bgp-af)# neighbor 172.60.2.3
Router(config-bgp-nbr-af)# slow-peer detection enable split-update-group
dynamic permanent

The peer is moved to a slow update group when detected slow.

  • If only the split-update-group dynamic command is configured, a dynamically detected slow peer is moved to an existing slow update group or a new one is created. This behavior is enabled by default.

  • If the permanent keyword is not configured, the peer returns to the original update group after recovery. If the permanent keyword is configured, the peer does not return automatically; use the clear command to move it back. Use this option if a peer keeps becoming a slow peer and recovering.
4.

Clear dynamically detected slow peers.

  • Clear all slow peers for a specific AFI/SAFI.

    Router# clear bgp slow-peers <afi> <safi>

  • Clear all slow peers for a neighbor across AFI/SAFI.

    Router# clear bgp slow-peers <neighbor-address>

  • Clear a specific AFI, SAFI, and neighbor combination.

    Router# clear bgp slow-peers <afi> <safi> <neighbor-address>
5.

View the running configuration.

Example:


slow-peer-detection disable
router bgp 5
address-family ipv4
 neighbor 172.60.2.3
 slow-peer detection disable split-update-group static
router bgp 5
address-family ipv4
 neighbor 172.60.2.3
 slow-peer detection enable split-update-group dynamic permanent
6.

View slow-peers summary for neighbors.

Example:

show bgp update out neighbor slow-peers brief
Fri Feb 5 00:12:50.830 UTC

VRF "default", Address-family "IPv4 Unicast"
Main routing table version: 9819220
RIB version: 9819220

Neighbor FG SG SG-R UG Status OutQ OutQ-R Version

19.1.3.1 0.4 0.4 --- 0.2 Normal 4864200 0 7073474
19.1.4.1 0.4 0.4 --- 0.2 Normal 5206200 0 7073474
7.

Check slow-peers across all address families and neighbors, and compare behavior after time passes.

Example:

Router# show bgp all all update out neighbor slow-peers
Fri Sep 13 13:57:48.503 PDT
Address Family: IPv4 Unicast
----------------------------

After 5 minutes:

Router# show bgp all all update out neighbor slow-peers
Fri Sep 13 14:02:23.097 PDT
Address Family: IPv4 Unicast
----------------------------
VRF "default", Address-family "IPv4 Unicast"
Main routing table version: 3329832
RIB version: 3329832
Neighbor 11.11.11.21
Filter-group 0.3, Refresh filter-group ---
Sub-group 0.2, Refresh sub-group ---
Update-group 0.3
Update OutQ: 20447800 bytes (7680 messages) Refresh
update OutQ: 0 bytes (0 messages) Filter-group pending:
7680 messages

Slow peer detection is managed globally or per neighbor address family as configured. Detected slow peers are isolated, and recovery behavior is controlled. Use the verification commands to confirm slow-peers status, queues, and group membership.