How to configure QoS to prioritize CallManager Intra-Cluster traffic

[dtran]

I have a single CallManager cluster of 4 servers, one Publisher and 3 Subscribers running version 4.0.2a with SR2a. I have the Publisher and one Subscriber located in Santa Ana Ca and the other 2 Subscribers are located in Dallas Tx. I am experienceing replication issues between the Publisher in Santa Ana Ca and the 2 Subscribers in Dallas Tx, I keep getting the red sad face when I run the DBLHelper.exe on the Publisher. I am pretty sure the issue is network related, I do have QoS applied for voice traffic but not for Intra-Cluster traffic. And I have a 10MB DS3 connection to an MPLS VPN network at both sites.

I am looking for some assistance on how to configure QoS to prioritize Intra-Cluster traffic. Any advices will be highly appriciated.

[anup.anand]

Check this section "Clustering over WAN" in the SRND guide.

Here is the link

http://www.cisco.com/en/US/products/sw/voicesw/ps556/products_implementation_design_guide_chapter09186a0080447510.html#wp1045846

Regards,

Anup

[dtran]

Hi Anup ! I have gone through the link you sent but still not sure how to configure it on the router.

Do you know actual commands or the step-by-step on how to configure it on the router ?

Thanks for your time ! I really appreciate it

[anup.anand]

There are three types of traffic in intra-cluster communication

1. Database replication from publisher to subscriber uses best effort (DSCP=0) by default.

2. Directory traffic from the LDAP directory also uses best effort packet marking.

3. For real-time traffic via Intercluster Communications (ICCS), which includes signaling, call admission control, and so forth, as well as CTI Manager realtime traffic, a DSCP value of 26 (AF31 or IP precedence 3) is used.

You would need to setup Low latency queuing in your routers to prioritize these traffic. Traffic patterns #1 and #2, use the DSCP value of 0. You would need to re-classify them to a different value to keep it separate from your regular data traffic.

You can re-classify the DSCP at the switch port that is connected to the Callmanager servers.

Check this QoS SRND guide for more details. Page 2-35 has a sample configuration for a 3550 switch.

http://www.cisco.com/univercd/cc/td/doc/solution/esm/qossrnd.pdf

You would need to use the TCP/UDP ports used by Callmanager to define the access list to match these traffic. Check this link for the ports used by Callmanager

http://www.cisco.com/en/US/products/sw/voicesw/ps556/products_tech_note09186a00801a62b9.shtml

After you mark the DSCP values, you can use Low Latency Queuing on the routers. Check QoS SRND, page 3-7 for a sample configuration.

As per the Callmanager SRND

Every 10,000 busy hour call attempts (BHCA) between sites that are clustered over the WAN

requires 900 kbps of bandwidth for Intra-Cluster Communication Signaling (ICCS).

You can set aside 900 kbps for the signaling class in LLQ.

An alternate and simple method would be to use LLQ by matching all traffic between the Callmanager servers by using an extended access list with source and destination IP addresses of the Callmanager servers at both sites. The drawback with this method being that it will prioritize all traffic between Callmanagers, not just the Intracluster traffic. Since most of the traffic between Callmanager servers are Intra cluster traffic, this should not be much of a concern.

Regards,

Anup

[dtran]

Thanks Anup ! I created an extended access list you suggested, will see if that helps.

I appreciate for your time !!