BE6KS as remote-site redundancy?

[BenjaminAmick]

I know this question is outside of the reccomended-path for standard integration of the BE6K environment, but I was wondering if this would be a feasible path.

Obviously, for a primary installation, we prefer to have a redundant pair of BE6KH/M at the customer's headquarters, but when we deal with a new installation in a geographically divergent area we prefer to have local call processing for failover purposes over just SRST. I think a BE6KM is a bit overkill for this need, as we have no purpose for an additional unity server, with the 2 per cluster limit, so we are left with just CM and IM&P (if desired) on the BE6KM leaving it highly underutilized unless we have other application (Like UCCX) that we would want to move cross-site.

Therefore, I was wondering if it would be feasible or even reasonable to spec the BE6KS for these remote facilities, as it would be able to handle the core application without much overhead. The only issue I see with this is that Cisco advertises the BE6KS as no-redundancy and a limit of 150 users, and I do not know if this would be a limitation if we were just adding it as a member server in the cluster, as I have not worked with the BE6KS hardware before.

Could anyone shed some light for me? Thank you.

[skilambi]

See this post

Re: Design Question using BE6S as a subscriber

[BenjaminAmick]

The access to that page appears to be restricted, even though I am a partner.

[skilambi]

Yes it is on the private side of the community and needs to be applied for access

[BenjaminAmick]

I do have access to the partner side, but your link is providing me with an unauthorized link. Is this not on the partner section? Where do I have to go to find this?

[skilambi]

It is also partner plus customers but is a separate community that you need to be approved for access

Private - Customer Connection Program

You should be able to ask for access here or contact Denise Britten

[jarias]

The short answer is if you want to mix UCM VM configurations and/or BE6000 / 7000 appliances within the same UCM cluster, this is allowed and TAC supported, but the cluster capacity is usually going to be limited to that of the smallest member, due to UCM clustering rules based on how the application works and what it’s expecting from all cluster nodes.

BE6000S won’t work for SRST replacement. Any cluster built out of BE6000S nodes (running 150 user or 1K user VMs depending on UCM version) is limited to max 150 users / 300 devices regardless of node count or what the other nodes are.

Similar happens when folks put BE6M / BE6H at ROBO but BE7M/H at HQ.

There’s an edit to the UC SRND chapter on sizing coming that talks to this in more detail than what’s in there today.

-james

[BenjaminAmick]

Thank you for the answer.

Other than Virtual Machine sizing constraints to the OVAs, is there any limitations that would stop you from running the BE6KM load of UCM in order to increase the user counts? Is it a matter of pure compute power on the blade server?

Would you be able to implement your own UC on UCS style deployment with a router-based blade server in order to run without the 150 user limitation?

[jarias]

@Ben…

Isn’t really the VM specs it’s the hardware underneath and the plumbing of UCM clustering.

E.g. for a BE6000S box, the hardware is slow particularly the storage (mobo raid controller + disk count/choices = low IOPS compared to UCS C), which limits how much we can run on one UCS E blade server regardless of VM configuration.

Also as you know, UCM clustering is not linearly “scale-out” the way that say, total port count on stackable Catalysts are. You can’t stack N nodes of capacity C to get NxC total capacity. The SRND rules are a little more elaborate than that and come with inflection points w.r.t. both VM footprint and VM count to get to a particular capacity.

BE6000S and UConUCS E160D M2 TRC#1 are under the same restrictions.

1K user VM on UConUCS Small / Small Plus TRCs can be used to create >1000 user clusters following SRND rules.

Again, look for summer/fall SRND update (can’t remember which one it’s in) for updated verbage on all these use cases.

-james

[BenjaminAmick]

Thanks James.

I think I understand quite a bit more now. I'll be looking forward to that SRND update.