System Hardware Provisioning (design planning) question

[MitelInMyBlood]

My newly-inherited CM (rls 8) w/Unity Cxn was purchased with the intent of it being able to support 1600 local users in a medium-sized office setting. However, I'm starting to have concerns that the buffalo maybe got squeezed a little tight to contain costs. Rather than a conventional Pub/Sub configuration as I've seen in other Cisco deployments, the Publisher is the only thing here local to me and the Subscriber is 3 hours away by car. With all due respect for disaster planning, is this design really 'best practices' to 1) provision all your instruments off the Pub rather than the Sub and 2)to have your Sub sitting 150 miles away? If anything I'd have liked to see the Sub/Pub together in the same rack and an SRST router sitting at the D/R site. Is this configuration possibly a ticking bomb? What could we have done better? Any potential issues having your sets homed off of the Pub vs the Sub? In this configuration, what's going to happen when (not if) we have a failover event? (if you were near retirement would you put your papers in before it happens?)


Original MUG/NAMU Charter Member

 

[whykap]

The pub under no circumstances should have any devices registered to it, except in case of failover.
That said putting a sub in the DR site is fine but you should have another sub at the site where the devices are.
Or put the pub at the DR and the sub at the main site.

The nunber of devices that are supported on a single server depends on the server platform.

 

[MitelInMyBlood]

Thanks

That squares with what I always believed. Presently there are almost 1400 phones registered to the Pub and only 1 sub 3 hours away. (this was configured for us by a large, well-known VAR, whose name I shall not mention)

I have a spare Contact Center server (bought for CC redundancy but as yet not deployed) which someone today suggested reformatting the drive and installing CM 8 it, buying a node license and configuring it as a sub then moving all the phones over to it. We have KUWL licensing so I think we should be good there.



Original MUG/NAMU Charter Member

 

[whykap]

I would love to know who your VAR is although I have a vague idea.
You could deploy the ccx server as another sub and that would solve your problem. it would take you around 4hrs for the whole thing so no big deal.
I believe and don't quote me on that but with version 8 you do not need a node license anymore. Just the software and DLU licenses which you already have.
I would make sure however before deploying.

1700 devices on the pub is not best design and if you open a tac case and the engineer gets a wiff that you are doing that he won't help you further until you meet best design specs.

 

[MitelInMyBlood]

Well we're not at 1700 phones,but already 1380 and at the end of the day it will easily grow to something beyond 1600.

By the way, the Pub is also a 7835, not a 7845

As for the VAR, I like the guys and they were a big help on a lot of things getting us up and going and I know there were thumbscrews applied to meet our budgetary constraints and still get it done.

The phones are mostly all deployed and working now with about 150 or so left to do. What's on my plate presently is several hundred users still waiting (some less patiently than others) for CupC, which so far seems like we're in beta trials with it.... way too many issues to give it to end users. We have it in controlled introduction to about 50 "friendlies" abd it's generating 4 or 5 trouble tickets a day for problems. It also seems to be a big resource hog. If a user has multiple chats open it will eventually bring the desktop (Win7/Intel Core-I5/750) to its knees.


Original MUG/NAMU Charter Member