cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2523
Views
8
Helpful
6
Replies

How does the CUCM 2:1 redundancy work

Hatem_Hamdi
Level 1
Level 1

Hi all,

Suppose that I have a CUCM cluster consisting of 4 server : 1 publisher + 3 subscribers which I set up in the 2:1 redundancy model. My question is how my ip phones are going to register.

Are they going to register with the first sub till it reaches its full capacity then they register with the second sub?

How many redunandcy group do I need to achieve load sharing between my three servers? and is it recommended to load share between all of them or just between the two primary subs.

Best Regards,

1 Accepted Solution

Accepted Solutions

Rob Huffman
Hall of Fame
Hall of Fame

Hi Hatem,

This 2:1 model is explained nicely in the CUCM SRND Guide. Here's a clip;

A call processing subscriber is a server that has the Cisco CallManager Service enabled. A single server license is required to enable this service on a subscriber. The Cisco CallManager Service cannot be enabled on a server if the publisher is not available because the publisher acts as a licensing server and distributes the licenses needed to activate the Cisco CallManager Service. Once this service is enabled, the server is able to perform call processing functions. Devices such as phones, gateways, and media resources can register and make calls only to servers with this service enabled. Unified CM supports up to eight servers in a cluster with the Cisco CallManager Service enabled.

Depending on the redundancy scheme chosen (see Call Processing Redundancy), the call processing subscriber will be either a primary (active) subscriber or a backup (standby) subscriber. In the load-balancing option, the subscriber can be both a primary and backup subscriber. When planning the design of a cluster, you should generally dedicate the call processing subscribers to this function. In larger-scale or higher-performance clusters, the call processing service should not be enabled on the publisher and TFTP server. Call processing subscribers normally operate in either dedicated pairs or shared pairs, depending on the redundancy scheme adopted. One-to-one redundancy uses dedicated pairs, while two-to-one redundancy uses two pairs of servers that share one server from each pair (the backup server).

The choice of hardware platform depends on the scale, performance, redundancy, and cost of the servers. Scale and performance are covered in the section on Unified CM Platform Capacity Planning, and redundancy is covered in the section on Call Processing Redundancy.

Call Processing Redundancy

You can choose from the following Unified CM redundancy configurations:

Two to one (2:1) — For every two primary subscribers, there is one shared secondary or backup subscriber.

One to one (1:1) — For every primary subscriber, there is a secondary or backup subscriber.

http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/srnd/7x/callpros.html#wp1043707

So in a "true" 2:1 model like you want you would have this;

CCM Group 1

Sub A - Primary

Sub C - Backup (secondary)

Sub B - Tertiary

CCM Group 2

Sub B - Primary

Sub C - Backup (secondary)

Sub A - Tertiary

You would have Device Pool A with half the phones using CCM Group 1

and

You would have Device Pool B with half the phones using CCM Group 2

The only time Sub C would be used is if Sub A or Sub B failed.

Cheers!

Rob

View solution in original post

6 Replies 6

Hello,

The call manager group can contain to maximum 3 call managers .

For the sizing , it's all dependant of the call manager hardware specs , each hardware has it's limit.

let's say you have 7816 which has the capability to reach 500 IP Phones , if you have 1000 ip phones in your corporate then you need to creat two device pools and two call manager groups that each one can register with a different server.

Hope this helps.

Amer

Hi Amer,

My question is about the 2:1 reduandancy model in particular. The call manager group in this very case contains 3 servers. let 's say I have assigned a DP with a 3 servrers call manager group to an IP Phone. How this IP Phone is going to find out about the server he should be registering to?

Best Regrads,

Mar7aba Hatem,

Listen, when you assign three call managers in the call manager group , you can re-order them , you can move a specific call manager up or down , which means that the one in the top is gonna be the first server the ip phone is gonna try to register to , then the second then the third .

Amer

Rob Huffman
Hall of Fame
Hall of Fame

Hi Hatem,

This 2:1 model is explained nicely in the CUCM SRND Guide. Here's a clip;

A call processing subscriber is a server that has the Cisco CallManager Service enabled. A single server license is required to enable this service on a subscriber. The Cisco CallManager Service cannot be enabled on a server if the publisher is not available because the publisher acts as a licensing server and distributes the licenses needed to activate the Cisco CallManager Service. Once this service is enabled, the server is able to perform call processing functions. Devices such as phones, gateways, and media resources can register and make calls only to servers with this service enabled. Unified CM supports up to eight servers in a cluster with the Cisco CallManager Service enabled.

Depending on the redundancy scheme chosen (see Call Processing Redundancy), the call processing subscriber will be either a primary (active) subscriber or a backup (standby) subscriber. In the load-balancing option, the subscriber can be both a primary and backup subscriber. When planning the design of a cluster, you should generally dedicate the call processing subscribers to this function. In larger-scale or higher-performance clusters, the call processing service should not be enabled on the publisher and TFTP server. Call processing subscribers normally operate in either dedicated pairs or shared pairs, depending on the redundancy scheme adopted. One-to-one redundancy uses dedicated pairs, while two-to-one redundancy uses two pairs of servers that share one server from each pair (the backup server).

The choice of hardware platform depends on the scale, performance, redundancy, and cost of the servers. Scale and performance are covered in the section on Unified CM Platform Capacity Planning, and redundancy is covered in the section on Call Processing Redundancy.

Call Processing Redundancy

You can choose from the following Unified CM redundancy configurations:

Two to one (2:1) — For every two primary subscribers, there is one shared secondary or backup subscriber.

One to one (1:1) — For every primary subscriber, there is a secondary or backup subscriber.

http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/srnd/7x/callpros.html#wp1043707

So in a "true" 2:1 model like you want you would have this;

CCM Group 1

Sub A - Primary

Sub C - Backup (secondary)

Sub B - Tertiary

CCM Group 2

Sub B - Primary

Sub C - Backup (secondary)

Sub A - Tertiary

You would have Device Pool A with half the phones using CCM Group 1

and

You would have Device Pool B with half the phones using CCM Group 2

The only time Sub C would be used is if Sub A or Sub B failed.

Cheers!

Rob

Thank you rob for this wonderful answer.

Rob Huffman
Hall of Fame
Hall of Fame

Hi Hatem,

You are most welcome my friend

Cheers!

Rob

PS: Thanks for reminding me to give Amer +5 for his great contributions here!