Control Plane Optimization for Local and Remote Peers
Feature Summary and Revision History
Applicable Product(s) or Functional Area |
vDRA |
Applicable Platform(s) |
Not Applicable |
Default Setting |
Enabled – Always-on |
Related Changes in This Release |
Not Applicable |
Related Documentation |
|
Revision Details |
Release |
---|---|
In CPS vDRA:
|
24.2.0 |
Feature Description
In vDRA, all the control plane messages from local and remote were handled by a single thread pool cpSubscriberWorker. Whenever there is a network outage or peer fluctuation on the remote site, the local site was flooded with control plane messages from the remote site thus, leading to peer expiration in local site followed by traffic loss. Hence, following enhancements are made in CPS vDRA 24.2.0 and later releases:
-
A separate thread pool is available to handle the remote sites control plane messages. The addition of a new thread pool helps to overcome the exhaustion of thread pool in local sites during any fluctuation on the remote sites. The control plane worker threads and queue size are increased through:
-
Policy Builder - cpSubscriberWorker threading configuration to handle the peer status control messages from local sites.
-
Policy Builder - globalCPSubscriberWorker threading configuration to handle the control messages from remote sites.
-
-
The control-plane remote-peer-policy global filter-fqdn-with-string CLI command configures the RxB peer FQDN as a list of substring to skip the local publish of RxB peer status control messages from remote sites.
-
The subscribe_messages_drop KPI includes the following labels added along with the existing labels:
-
control_topic
-
system
-
For more information see, the Advanced Tuning section in the CPS vDRA Advanced Tuning Guide, CLI Commands section in the CPS vDRA Operations Guide, and Statistics/KPI Additions or Changes section in the CPS Release Change Reference Guide.