SE Resilience

[AJ22]

Hi all

As above I have primary /secondary server 300 users load balanced between each server. Servers are on different sites and connect via MPLS. I have around 24 remote sites with 9608 IP handsets. Failover works good but I have a question on network failure. Say the link to main site goes down the 100 ip phones on the local primary site are still registered and working, my understanding is remote handsets will register to secondary when primary no longer visable due to system failure or network failure at Server locations. So vm pro back-up will kick in when the the secondary can no longer see the prinsry for all Auto attendants etc. Would the VMPRO still be operational on the main site too for the handsets that are still registered to that?


I am trying to think of all eventualities, I think certain sites would be better with IPO expansion then SCN backup to primary, then I would have 3 points of failure local system /primary/secondary.

Thanks again as always

Allan

 

[kyle555]

I have the same question. Read read read EVERYTHING about it 20 times and try to find answers to the conflicting information.

IPO 10.0 Resilience Overview

https://downloads.avaya.com/css/P8/documents/101028351

My issue is what you have to read in between the lines. Since they added active/active VMPro in IP Office Select, I reckon it's answering a question I have for a customer that can't be bothered to pay for Select for that one feature. It's not that the feature is particularly critical, but it does speak to there being different brains under the hood for how either voicemail can be active and have bi-directional sync.

Any resilience infromation pre-Select offering active/active refers to all kinds of failure but network.
That's to say - VMPro service failure at either site or IPO service failure at either site or either physical server going boom.

That means if either site A or B's WAN go down and people at site A and B both call remote user Warehouse Willy in West Virginia, and Willy's out to lunch, each site A will leave a message for Willy in his mailbox on server A, and B's will on B. And then in resyncing, bad things might happen. Think shared hunt groups across sites and people getting messages before sync is complete, deleting voicemail folders on both and starting over and re-recording greetings to make them happy again and work normally next time both VMPros go active.

 

[AJ22]

I have done the same, system failure when network is stable no problem I understand. But from experience it is 50/50 between system/network to fail.

Pulling the connection from the primary is really the same as the WAN to primary going down so everything registers to secondary. But what if only certain sites can see the primary and the rest cannot, do the sites that cannot see primary simply register to secondary? I think my question is does the primary have to be down completely dead in water for the phones to re register to the secondary or does it work Wherby if the phone simply can’t see primary it will jump out to secondary regardless?

Voicemail I would like to test this to see what happens..

 

[sizbut]

For the phones it is certainly a two step dance:
[ul]
[li]They will only attempt to register with their backup if they lose sight of their current system.
[/li]
[li]and their backup will only accept their registration if it too can't see their current system.[/li]
[/ul]

So if its a problem with the phone's current system, then yes, both criteria are probably meet at the same time. But if its a data routing issue then you can create scenarios where the phone's want to failover but the backup isn't accepting them.

I don't know what the break point would be in terms of number of phones at a remote site, but having a local IP Office at the site will always be more reliable.

Stuck in a never ending cycle of file copying.

 

[kyle555]

I disagree that pulling from the primary = WAN outage. You still have people at the site of the primary. With a WAN outage, they use the primary. Pull the cable and they don't. That's how you get 2 Voicemail Pro's running at the same time and run into something non-Select VMPros didn't think about.

 

[AJ22]

Kyle555- yes of course you are right I was incorrect to say that. There is many ways of looking at this. Seems they built this solely around system failure,

If remote phones don’t see primary they look to secondary but if secondary still sees the primary.. the phones won’t register anywhere. This might not always happen but I’d say it’s something I’d want to mitigate.

Solution- install IPO on each site..