3300 IP Flex Failover to branch office, Resiliency, DR disaster recovery, Business continuity 1

[TrentGreenawalt]

We have our main site and a branch office connected via MPLS. Each location has its own internet Main (2-T1) and Branch (1-T1). Each site has a 3300 CX-II MCD 4.2 Software 10.2.2.10 at the site, main office is around 35 phones and branch office 15 phones. What we are trying to accomplish should have been dealt with when we initially put the system in, but almost 2 years later we haven't had an outage of note that lasted longer than a few days...until now. Our branch office was beat up pretty bad by a storm this weekend and is going to be without power for a few days.

So at this point I have opened a ticket with AT&T to forward my branch office main number to my main line here at headquarters and this does work. However, none of my DIDs are working, and the calls aren't identified as calls for our branch vs main here on our 3300, so it is causing issues for the receptionist here (this is always the case). It does work for now, but not as ideal as we would like.

So in the past few weeks, AT&T has been pitching an automated failover system to us IAR (Free of charge - imagine that), that would automatically reroute our calls from either branch to the other instantly in the case of a power outage, failure, 3300 down, etc. Sounds great right? I agree... However, we need to ensure that the 3300s at either end can handle this change. The current forwarding simply translates one number to the other in AT&T's cloud, I believe this new service will actually deliver the branch office numbers to my main controller, which most likely doesn't have a clue what to do with those calls right, or no? Where do I check this? How would calls out work in this scenario?

Additionally, since power is out, obviously everything is out. However, what if it was just my internet IP Flex internet line that was out at the branch office and our MPLS was up, is there a way to route these new inbound calls back down to the correct phones, call groups, etc? What if they called out from the branch, would we be able to route that out our site?

To make it even more interesting, what if the MPLS was out so the 3300s couldn't talk to each other? I could setup a IPSEC VPN with my branch office and route internal calls across the internet (Granted their internet was up)? Is this possible?

Last scenario, controller failure: I know that my 3300 has built in resiliency MCD 4.2 and I have tried to setup as many Secondary elements that I can for the users, but not all users can have this setting. Especially my users that are able to log into multiple sites. I always get an error. I have seen demos where they unplug the 3300 while on the call, and the other controller just picks up where the call leaves off. Is this resiliency or something that I have to purchase?

So sorry to put this all in one post, but I feel the value of this thread will benefit the masses and get a good DR/Business continuity chat going on, all the while helping me out of the situation that I am unfortunately in right now.

Thanks for your time,
Trent

 

[TrentGreenawalt]

One additional note: I do have a Mitel border gateway for my Teleworkers offsite - 3 of them. Also we are using IP Flex SIP trunks from AT&T: 12 at main and 6 in our remote site.

Best Regards,

Trent Greenawalt
IT Manager
Milwaukee, WI

 

[LoopyLou]

Not sure about your AT&T products so I can't really comment effectively on those.

Basically by default the systems support resiliency. This is the ability of a phone to fail from its primary controller to its secondary controller ( at this time only one failover controller can be defined ).

Resiliency costs nothing ( sort of ). What you do need however for it to work is that both of your controllers need to be in a cluster with IP trunks ( $ ) between them. You also need a free device license ( $ ) on your secondary controller for each phone that fails over.

So to program a set you need a user and device license on the primary controller ( MCD 4.x software ) and a free ( and I mean unused when I say "free" ) device license on the secondary controller for fail-over. In MCD 5 software Mitel has new licensing standard vs Enterprise. When you buy an Enterprise user license you can automatically program it with a secondary controller for fail over. Thats why the Enterprise user license is more then a standard user license.

To make a set resilient you simply assign it a secondary controller in "user and device" programming. Thats it. There should be a pull down listing all the clustered controllers and you select the secondary element you want to use ( or fail to ). Hotdesk profiles can also be resilient and no additional licensing is required to make a hot desk profile resilient since the hot desk user only uses a "user" licenses and doesn't need a device license.

Just a note, hunt groups, consoles and ACD agents and groups can also be made resilient.

Teleworkers can also be setup to failover through the MBG although a network would need to be up for this to happen.

Ok back to your questions. What ever AT&T does you need both sites clustered to take advantage of the situation. In clustering the primary site knows all the extensions at the secondary site and visa versa. So on a normal DID call you might get the last 4 digits of the DID telephone number delivered via the SIP trunks at your main site. Those 4 digits usually correspond to the extension number of the phone. When the 3300 sees the 4 digits is says hmmmm is that an extension number here. No then it looks in the remote directory number form and see what cluster element it should send it to. If it finds the number it then forwards the call across the IP trunks to the element ( or the 3300 controller ) that contains the extension number. With AT&T's IAR it sounds like they can send the same digits down either of the SIP trunks and that is good because as long as either 3300 gets the correct digits then it can deal with them and send them to the correct element that will ring the right phone ( as long as they are clustered ).

For calls going out they typically go out their primary controllers trunks but with ARS route lists you could send calls out by sending them across the IP trunks to the other controller and using its trunks if your own are down. You just need to redo your ARS and keep in mind 911 calls go out the wrong location in the failed trunk senario ( if the 3300 is still up and has copper trunks use those for 911 ). Note: If the MBG is used as a SIP trunk gateway it can point its SIP trunks to another controller on failure of the primary. This might be helpful

As far as the using a VPN if the MPLS is down. The 3300 doesn't really care what you do for a network as long as it can route to the IP address of the other controller some how. The issue will be bandwidth and QoS for call quality across the VPN.

Clear as mud right. I will probably get bitch slapped by everyone for all my mistakes but in short you can probably give yourself a lot of survivability with what you have with not a lot of investment.


I'd tell you a UDP joke but I'm afraid you won't get it. TCP jokes are the best because you always get them.

 

[TrentGreenawalt]

Hi LoopyLou,

I appologize for not getting back to you sooner, I just got approval to move forward with this project. Your information is invaluable to keep our consultants in line when they are onsite working on this. I hope to be able to be fully redundant for a few hours of consulting time. Follow up question, when you mention clustered, you mean are they talking to each other correct? I get the other controller in the drop downs for backup controller, and have "sync" arrows everywhere, is that what you are talking about?

I really appreciate your help, and have given you a star for your feedback. Thanks again!

Trent Greenawalt

Best Regards,

Trent Greenawalt
IT Manager
Milwaukee, WI

Phone Hardware:
Release Level: 4.2 SP2
Active Software Load: 10.2.2.10
Platform CX-II, 512mb of RAM
Mitel 3300 (2 Controllers, 2 locations)
5340, 5330, 5320 Phones (SIP Service via AT&T IP FLEX)

 

[LoopyLou]

What I mean by clustered is the feature of Mitel products where numerous physical ( or virtual ) systems are setup to appear as one big PBX. This can be considered "talking" to one another I guess. Basically it means every 3300 in the cluster is aware of every phone in the cluster whether it is programmed on itself or it is programmed on another 3300 in the cluster. Not sure if that answers the question.




I'd tell you a UDP joke but I'm afraid you won't get it. TCP jokes are the best because you always get them.