BE6K v11.5 (CUCM) backup

[mapicallo]

is possible with a BE6K v11.5 (CUCM) to configure a cluster so that a server / node can be a backup of up to 10 active servers / nodes?

[Ritesh Desai]

@TONY SMITH 

CUCM 7500 and 10000 profiles are valid on BE7K UCS server models. BE6K UCS models are not supported for CUCM 7500 and above profiles according to compatibility matrix.

[mapicallo]

I have seen the following:

https://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/srnd/collab11/collab11/callpros.html

There is some way to get a redundancy model greater than 2: 1, for example up to 10: 1

[Maren Mahoney]

There are two different things at play here.

First, the 1:1 and 2:1 redundancy refers to how much of the total capacity you load onto a server. In 1:1 redundancy you are using 50% of any given server's capacity, which means that in the event of a server failure the entire load of the failed server can be failed over to a single other server.

In 2:1 redundancy you are using 66% of any given server's capacity, which means that in the event of a server failure the load on that server would need to be failed over to two other servers.

The other thing you may be seeing is that there can be 8 call processing nodes in a standard cluster, and 16 call processing nodes in a supercluster/megacluster.

With the BE6K, you can create a cluster but not a supercluster. And regardless of the number of nodes in your cluster you are limited to a total of 1200 devices.

If this does not address the question you are asking, can you provide the exact quote from the document where it references 10:1 redundancy?

Maren

[mapicallo]

I will try to explain the scenario

A company R, has some needs for fixed and mobile voice and data communications services, and among other specifications there is one in particular that is the following: 

"The Voice Communications solution, which supports Call Control and Call Management features, should be able to support a virtualization platform, and should be configured in Fault Tolerant and HA (High Availability) configuration.

The telephony server must have the following redundancy mechanisms that can coexist simultaneously in the same solution, the final implementation scenario could be: VMWare High Availability or Fault Tolerance.

Server redundancy: Ability to configure a cluster so that a redundant server (in stand-by mode) can be back-up of up to 10 active servers. The connectivity between these servers will be L2 level.

A plan and protocol of joint design of DR - Disaster Recovery for the virtual environment and storage of the voice solution should be presented. "

Another company A offers as a possible solution to this requirement:

2 servers with Cisco BE6K, "Cisco Bussines Edition 6000 v11.5 "

And another company B, says that solution does not meet that requirement, claiming:

"This switchboard or platform does not allow a configuration of a redundant server that acts in back-up mode of up to 10 active servers, given that, at the factory, the platform offered by company A allows, at most, only one redundant server to act in the back-up mode of another active server The capacity required by the specification, to configure a cluster of up to eleven servers (10 active + 1 back-up) is impossible to comply with this model of platform "

 To which company A responds with the following:

"The BE6K is a packaged CUCM for 1,000 users and 2 servers for quick sale without having to think about physical machines, virtual machines, licensing, ... It usually does not sell more than two servers because it is unnecessary in most cases. It does not mean that it does not support it, in fact it supports it because the CUCM is the same software as the one present in the BE6K edition and we could have N nodes in clusters for redundancy with the only requirement of not being able to exceed 1,000 users. "

or

https://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/srnd/collab11/collab11/callpros.html

* The important point is if:

Server redundancy: Ability to configure a cluster so that a redundant server (in stand-by mode) can be back-up of up to 10 active servers. The connectivity between these servers will be L2 level.

* Is company B really totally right?

* That level of backup requested, is really influenced by the proposed choice (2 servers with Cisco's BE6K, " Bussines Edition 6000 v11.5 from Cisco"),

Do you have to see one thing with the other?

Is it possible to achieve this level of backup using other standard tools / technologies complementary to the proposed solution?

[Maren Mahoney]

Company B is correct that with the BE6K it is impossible to have a 10:1 redundancy.

Company A is correct that BE6K is usually sold as either one or two servers, which provides sufficient redundancy for 1000 users/1200 devices. However, Company A is incorrect that "we could have N nodes in clusters for redundancy".

A CUCM cluster (whether it is on the BE6K platform or not) is limited to 20 total servers, and is limited to 8 call processing nodes - and redundancy generally refers to call processing redundancy. But even with that, you do not get 8:1 redundancy as a single device is given a list of up to 3 nodes to which it can register/failover.

There is no telephony system I know of that provides 10:1 redundancy.

I am also concerned about "The connectivity between these servers will be L2 level." While it is possible to have all of your CUCM servers in the same IP subnet (which would provide L2 connectivity) that is not a recommended practice as it means you have a single point of failure (default gateway). But if you can speak to that requirement in terms of datacenter spine/leaf architecture you may be able to address that requirement.

I hope this is helpful. Please let us know if you have more questions.

Maren

[TONY SMITH]

I think you need more basic information, I don't think you have stated how many users and/or devices are to be supported.  Or what is meant by "stand-by mode" which has no meaning in a CUCM cluster.  Or indeed how one server is expected to back up 10 servers - are each of those 10 only expected to be at 10% of their capacity, or is the backup only intended to protect against one of the 10 servers being offline?

Also I think we've been assuming the requirements refer to the CUCM application environment, however maybe even that's not the case, maybe they're referring to the VMware infrastructure.

As I noted earlier it would be better to start with the functional requirements and then see which cluster design and which platforms can best meet it.