VOIP Rollout Suggestions Needed

[MitelInMyBlood]

Okay, sr mgmt has awakened from winter hibernation, seen the announcement of EOS for the 2K and is now in full hysteria mode wanting us to (finally) move forward with VOIP.

The question before us is what's the best plan? do we:

(A) Migrate all 18 (!) of the 2K expanded per nodes over to the 4 or so 3300s it'd take then begin gradual VOIP instrument replacements from that point?

or

(B) Draw a line in the sand and say no more TDM phones in newly remodeled areas of the bldg and get to VOIP that way
(they're constantly remodeling)

or

(C) Leave the TDM system completely alone, install 2 MXe resilient servers + a gateway node, add the new 3300s to the existing CLUSTER and begin somewhat orderly moves (a whole department at a time) onto the 3300.

I favor plan C. My boss likes plan A and his boss, who holds the purse strings likes plan B. Either plan has drawbacks. With plan A we sacrifice control redundancy & wind up fighting companion key appearances scattered across the cluster, which experience has shown don't work plus having to mirror the database and do a flash cut. With plan B it will take at least 5 years to get there and with plan C we're going to end up spending a lot of money up front where with plans A and B it could be dribbled out. Instrument cost being the major item here, as we're talking about roughly 2800 stations. I'm also thinking 4 3300s + gateway cuz they think they want 100% resiliency. Yee-haw!!!

Anyway others' thoughts and insights would be appreciated

 

[M1tel]

Install EM with Opsman first get it working. Install MXE Servers these are pure IP (5600 Sets) no trunks or analogue. Install mxe servers 1 main one resilient, 2 MXE systems with failover PRI's for trunking and analogue use a cabinet off your 2k's for ONS sets and backup ls trunks. Move any tdm with ops to IP.

Make sure to Flex dim the MXE servers to get the correct numbers

I just did this transition this is the easiest way and least intrusive.

 

[kwbMitel]

Option A would be a nightmare. If you could use 1 3300 to migrate the DB, maybe, but you're talking reprogramming from scratch.

B and C don't have much separation in scope and can probably be combined into a C+ or B- depending on your perspective.

The new Mitel Call Director can handle up to 5600 IP devices. Add one of these and Cluster it with your existing system. The MCD does not have any TDM capability period. All trunking and analog would have to route thru the SX2000. Add Other 3300 MXe's for resiliency as necessary and eventual displacement of the 2K devices.

All new adds and renovations would be VOIP and a migration plan could be implemented for the rest over a longer period of time to lessen the sticker shock and also reduce the likelyhood of error.

Changing all at once is a massive undertaking and requires unnecessary risk in my opinion. Take small bites. The 2K will be supported for some time yet. No need to panic. That being said, don't wait until panic is necessary either.

*******************************************************
Occam's Razor - All things being equal, the simplest solution is the right one.

 

[Lundah]

With that many stations, I would follow kwb's advice and go with the MXe Server (2 for resiliency). The expanded limits the MXe Server gives you will be nice (more conf. channels, increased multicall/BLF limits, etc). Leave the 2K in place for trunking, and use a combo of options B and C for the user rollout.

If the network will support VOIP, that would be the way to go. Migrating the per cabinets to the 3300 will be a major headache, and you'll have to migrate to IP sets eventually anyway, might as well get it done while you can get the funding for it.

 

[MitelInMyBlood]

Thanks!!

Well I agree option A would be a nightmare of collossal proportion. However we've done many NEW INSTALLATIONS in some small existing offices with old key systems where to shave costs they've had us deploy a new 3300 then sent them a spare 2K per bay, 8 DNI cards, 2 ONS/Class and 50 or so SS4150s and some blue-backers all from our basement inventory of several decommissioned systems acquired through reorganizations, business unit spin-offs and downsizing. The TDM hybrid deployment is done to leverage the "nearly new" but otherwise nearly obsolete stuff, plus all the copper is already in place. The secondary market offers only pennies on the dollar for it. Never mind the huge pile of loose phones, there's still 8 unopened cases of 4150s and a new/unopened 2K-Red MC and 4 unopened DSU nodes left over from a period of explosive growth in 2001 then never deployed. Anyway, Mgmt knows that the 3300 will drive a 2K per node, so that's the direction they were looking. (swap out the 2K main control for 3300 main control) Seriously flawed thinking, but hey I'm just the piano player in this here bordello.

I already have EMS w/OPS and the big 2K is already clustered with a dozen other 2K's and 40 or so small 3300s scattered around the US & all tied together w/VOIP trunking over a digital DS3, so that piece is in place (thankfully).

I had completely overlooked the fact that the 3300 MCD had no TDM capability. That's GREAT actually as it nullifies option A from further consideration, period. Keeping a 2K alive for legacy trunking and analog was also on the drawing board, short term at least as for right now the VM is on an Octel with 48 ports of COV integration. The Octel is going away too, but later.

Another major $ hit is the need for 166 48-port POE etherswitches (Cisco 3760-G) which equipped with the required options & IOS will probably come in around $7200 a copy. That's way more ports than it will take for VOIP, but that allows for a full one-for-one port replacement of what's in the 40 total wiring closets today. (40 closets because the building is 775 feet long x 10 stories high.) We're replacing the Etherswitches right now. We could do it for a lot less $ but we're a Cisco shop top to bottom and I know to be careful in picking my battles. Be satisfied we won with Mitel VOIP. That was a long, hard-fought battle.

 

[Mitelmatt]

Just don't forget if you install some MXE Servers to handle the migration, you will need a normal MXe controller to act as a gateway between the servers and the SX2000 cluster.

As as been said the MXE server can only talk IP

Be a good network when you're finished

 

[MitelInMyBlood]

I already have a 250-user controller in place today handling trunking between the office 2K and the balance of the corporate sites (mostly all 3300s themselves) so the "gateway" piece is there. One of the 4 SX2Ks in my office also has 40 links and 5 dual-fiber DSU cans hanging off of it. The one remaining fiber goes to a pair of expanded pers (384 ports) for those analog nuisance circuits you just can't seem to ever get away from.

I see this 2K as needing to remain as we have almost 200 (analog) Polycom conference phones which users will surrender only when you pry their cold, dead fingers from around them.

The network design has also been thoroughly reviewed (by Cisco when they thought they were about to get the phone deal) plus we have a couple really sharp network guys that I'm confident can handle anything that comes up. The only real concern is power in the computer room. Though all on large commercial (Liebert) systems we've actually had two total power interruptions in the past 5 years that were awfully high profile and took the core routers completely down. The saving grace at the time was the phones stayed up (all DC-powered TDM). Once we go VOIP the phones too will be at the mercy of the Liebert system. Hopefully I'll be retired by the next time it crashes as with both network PCs and phones down together there's going to be some major explaining to do.