CS1000E 5.5 Can I separate my CP MGate 32 cards into separate cabinets for redundancy?

[Funk49]

We have a CS1000E release 5.5 with NTDU14/15 MG1000 chassis and a 1005r Call Pilot server.
There was a service issue where all of our DSPs had been disabled in that same chassis.

The MGC w/DSP(s) is in slot 0 and the CP MGates 32 cards are in slots 3 & 4.

With the service issue we had, no one could access Call Pilot, well they could only there was no audio.
It took a little while before discovering the disabled DSPs on the MGC card so there was no voice mail user access and no Call Pilot Applications being answered.
None. We have quite a few Call Pilot Applications.

Once service was restored I was curious why both of our CP MGate 32 cards were in the same chassis, i.e., single point of failure.

My vendor engineer is looking into my request to separate these Call Pilot cards into separate cabinets.

But here are the various concerns he stated;

- This was not an oversight in the original install design, as matter of fact, in release 5.0 you had no choice but to keep the CP Mgate cards in the same chassis for clocking reasons
- They have separated the CP Mgate cards into different cabinets for some 5.5 customers but some are experiencing "popping and crackling" due to clocking issues with the Call Pilot (some are and some are not and they don't know why and that some customers don't even start hearing the issue for up to year later?)
- He stated I would require a special cable(?) for clocking issues when separating the CP MGates between cabinets
- Then he said maybe I wouldn't require this if the cable was long enough and within the same data rack housing the MG1000 chassis(s)
- Finally he stated the Call Pilot 1005r server may not allow this request?

I'm just curious if others have faced this same scenario.

Wouldn't separating the CP Mgate cards into separate cabinets allow some sort of Call Pilot channel redundancy in case there is an issue with the chassis/MGC/DSP?

Is there an issue with the DS30x clocking causing popping and crackling if separating these cards into different cabinets?

What could be an issue with the 1005r server that would not allow the separation of the MGate cards?

Thanks in advance

 

[KCFLHRC]

Those Mgate's can reside in separate chassis. But you would probably still have the same problem. Because the Call Pilot didn't detect an issue it would still try and send calls to that Mgate because the Agent TN's were still showing idle even though there was no TLAN connection.

 

[Funk49]

I am finding some Avaya documentation on the cable needed to go between cabinets (AUI connections on the back of the chassis) if separating the CP Mgate cards.
Document P-2008-0178

And it appears you did used to have to keep those in the same chassis, now you can seperate them.

But to KCFLHRC's point, you're right. Could I disable the MGate card associated with the chassis at fault and will that then force the calls to Call Pilot to go through the other MGate card? Worst case (until the vendor or I noticed and disabled the card) would it load share and at least give access to half of the channels?

 

[KCFLHRC]

Yes, you could have disabled that card and it should send traffic to the other card because then the agents would be disabled. It would never actually load share, just longest idle port typically.

 

[Funk49]

Thanks for the responses KCFLHRC, looks like we are going forward with relocating the Call Pilot Media Gateway card. I does take a special cable that connects both (separated) chassis via the AUI connection and provides clocking synchronization.
Apparently this was an issue that Nortel resolved - Nortel document P-2008-0178 "CS1000 Media Gateway Clock Synchronization and Applications Enhanced Resiliency". Big thanks to my awesome service tech (Z!) who occasionally visits this site. Next project will be looking into splitting up our MIRAN cards, all (3) in the same chassis?

 

[KCFLHRC]

It would appear that who ever engineered the switch and physical card layout didn't think too far ahead. Putting all 3 miran cards in the same chassis and both mgate cards in the same chassis either didn't know what they were doing or was lazy.

 

[bignose21]

The problem with mgates in different cabinets is clocking. The call pilot grabs one for its clock the other will slip. This could be over come by putting a pstn pri in each gateway, but then a pri fault could cause a similar issue. A way round this I think mentioned earlier is get a dect cable kit to sync the two cabinets.