Server edition split users between primary and secondary question. 2

[qtelcom]

Have a single site with primary and secondary server with around 400 users. I trying to best decide how to split users between the servers, having 200 on one and 200 on their other. I'm wondering what the best practice is for this for ease of use for deployment and ongoing administration.

My understanding you create 200 users on primary and 200 users on secondary, and register the appropriate phone to the corresponding server. What's the easiest way to have this be plug in play for the ip phones? I've always used dhcp option 242 utilizing a separate voice vlan.

Should there be 2 voice vlans? And subnet the primary and secondary servers? Then tell each phone which vlan to belong to, which then it gets it corresponding 242 option. Or is there a simpler way? Is their a mechanism in server edition that, for example, if the primary server receives a registration request for a user/Extn located on the secondary server, it tells the phone the register with the secondary server instead and vice versa?

Customer has windows dhcp servers on pdc and bdc that split the load for the entire single site location.

Thanks ahead for any advice.

 

[zoomatomic]

You should have all handsets register on the primary server. The secondary server is for failover should the Primary go down.


ACSS - SME
ASPS - SME

http://www.vodafone.ie/readybusiness

 

[tlpeter]

With 9.1 and the select version you can do this.
You can even do this for voicemail pro.


BAZINGA!

I'm not insane, my mother had me tested!

 

[qtelcom]

Tlpeter, just to clarify, must they have select? Or will regular 9.1 server edition do this? and where in the docs is this? I read the whole deployment guide and can't see how it's done.

Zoomatomic, I'm well aware of th failover, the point of splitting the load is so if primary server fails only 200 user are affected during the failover instead of all 400, and vice versa if the secondary failed.

 

[morenot]

Server Edition dual servers reference configuration

This reference configuration is for a customer who has one or two locations, IP telephony, and may require resilience.

For specific capacity information, see Deploying Avaya IP Office™ Platform Server Edition .

In this configuration two Server Edition servers are configured:

Server Edition Primary — You can configure trunks, users, and extensions on a primary server. You can distribute and administer all hunt groups on a primary server. The primary server offers resilience for the IP Phone and users of the secondary server.

Server Edition Secondary — You can configure trunks, users, and extensions on a secondary server. The secondary server offers resilience for IP Phone, Voicemail Pro, and hunt group of the primary server.

taken from IP Office Knowledgebase

/JLMT

 

[zoomatomic]

Zoomatomic, I'm well aware of th failover, the point of splitting the load is so if primary server fails only 200 user are affected during the failover instead of all 400, and vice versa if the secondary failed."

Sorry but you should have stated what version of IP Office, you asked a question I only answered

ACSS - SME
ASPS - SME

http://www.vodafone.ie/readybusiness

 

[derfloh]

Split the load is the only argument for me but not with 400 users because the small SE server supports 750 users

If primary goes down all calls between phones and between phones and IP500 wil stay connected if you use direct media and routing is well configured. Standby phones will register to the other server after three minutes. So usually no ongoing call should be lost.

If you want to split the load I see two different ways.

1st: The local switch puts the phone into a VLAN based on the location. It will work as long as the phone will stay in the same location. If someone takes the phone to another switch with another voice VLAN configured you will come into trouble because the phone will try to register with the server where the extension is not configured at.

2nd: you can configure group settings in 46xxsettings.txt with a group specific value for the call server address. It will work regardless which switch port the phone is connected to but you have to set the group value for every single phone manually.

 

[qtelcom]

Thanks derfloh, you are thinking exactly along the lines I was thinking. Some more backstory here; The direct Avaya guy we brought in during the presales mentioned the 200 by 200 split to the customer as an active/active, and if you have a failure only 200 users are affected at the time, so that is what is driving this whole conversation. I'm seeing that this is possible, but not completely "plug-n-play" as it appears for a single site, and I really like plug-n-play KISS deployments.

I figured the only ways are;

A)Have 2 vlan ids, (have default data scope option 242 VLANID point the phone in the primary server vlan subnet) and then manually change the vlan id for any phone required in the secondary vlan server subnet. That way they only need to manually change the vlan id for 200 phones/users that are going to register with the secondary server. Done deal once they are all deployed with the appropriate vlanid set, users/departments can move at will without problems.

B)Use lldp on the 2 switch stacks that tell the phones which vlan to belong to, but if users/departments move around this creates the problem you described, sounds very messy.

c)use the 46xxsettings.txt with group specific values like you mentioned, and manually set the group value for every single phone. (kinda like my option A, but required for every phone, however this option would work with a single voice vlanid.)

D)scrap the 200 on primary and 200 on secondary, and just put all 400 on primary and avoid the complexity for little gain. Unfortunately the customer has the 200 by 200 split load in their mind.

So does getting the Select version do anything extra fancier or easier for my scenario of the 200 by 200 split?

Thanks in advance!

 

[tlpeter]

If continuity is that important then ditch the IPO and go for Aura.
The IPO has no direct failover,Aura does do much better and really fast.

BAZINGA!

I'm not insane, my mother had me tested!

 

[Albus2]

@tlpeter: Just correct!
@qtelcom: This is another example of Avaya sales rep doing a bad job. They usually have no clue about IPO. They only know the 200 slide PPT version...
They just tell as story of 'go mid-marked' which is IPO SE. They forget to look at the customers needs. In fact there is only a limited number of possible customer served best with IPO SE. Most are better served with a 500v2 or Aura.

One important thing is to recognise the difference between IPO SCN, Server Edition, Server Edition Select and Aura.
- IPO SCN are independent systems, which are networked. Some feature make it look like 'one' system, but it's not.
In older times networking between units was only possible as star, not mesh, however the number of conversations between units was open. This changed later on, mesh was possible, but calls limited to 4 outgoing per license.
- Server Edition is not much different than IPO SCN. Yes, we have central licensing, and the number of conversations between units is open. But it's still star networked, or in case of a secondary server it's a double star. It's not just a funny coincidence that one can have 1 primary, 1 secondary and 30 expansion units. This is 32 units, just like it was with IPO SCN.
- Server Edition Select, if you have two of them, brings mash network. And the option for fall-back to be like 500v2 to 500v2, or even Server to 500v2.
- Aura is completely different. It is one system with one central server (with option for redundancy) and a number of gateways. But everything is controlled an managed centrally. Gateways are here to have local physical ports and local survivability in case the WAN fails.

A little thing to make this difference clear:
If a number of phones in one open space office are in a pick-up group, but some phones are registered to the backup unit, then pick-up fails. (I did not verify this in 9.1, but it was true for 9.0)


So, if you go for IPO, keep in mind to have just one unit per site. A second unit, like the Server Edition Secondary server is only for backup. This is Active-Ready redundancy.
If you have another large office somewhere else, then put the secondary there. In this case you have Active-Active redundancy. But still just one 1XPortal.
If it must be Server Edition, as in you case with 400 users, install SE in VMware. Even if you bring a dedicated server just for PBX needs. If it is just one site, you may skip the secondary. VMware brings another layer of possibilities for backing up.


Regarding the number of VLAN's and the consequences when moving phones around:
Some other posts above are incorrect. You can have just one VLAN or one per building or floor. As you like, or more likely, as the customers is doing the rest of his network.
In DHCP option 242 (Data-VLAN), you should add L2QVLAN=20,VLANTEST=60. This will send the phone to Voice-VLAN 20. The phone memorizes this for the next reboot, i.e. stays in this VLAN. But if it fails to get a DHCP offer in this VLAN it will ditch any VLAN info and start over without VLAN. In this (new) Data-VLAN it will get a new 242 option like L2QVLAN=30,VLANTEST=60 for Voice-VLAN 30.
Another option, especially if using AVAYA switches: Use LLDP. In this case the correct settings are configured in each switch. Any phone will immediately get the correct infos.

 

[derfloh]

Regarding the number of VLAN's and the consequences when moving phones around:
Some other posts above are incorrect.

What exactly is wrong?

 

[Albus2]

@derfloh

'wrong' is

1st: If someone takes the phone to another switch with another voice VLAN configured you will come into trouble because the phone will try to register with the server where the extension is not configured at.​

This will happen, if only L2QVLAN is given. However just because of this reason you can add VLANTEST.

The meccano is:
* If the phone receives L2QVLAN in the untagged DHCP request, it will do another DHCP request with this VLAN-ID
* If this DHCP-request, using VLAN-ID, is successful, this VLAN-ID is memorized.
* Next boot is directly using the last known VLAN-ID. Endless...
* Except, if VLANTEST=60 is given together wit L2QVLAN. In this case, the phone only waits for 60 seconds. Then it starts over without VLAN-ID.
-> Now, if this phone is attached in another environment where another Voice-VLAN is valid, it will learn this. So: There is no trouble with multiple Voice-VLAN's.

'causing even more troubles' is:

2nd: you can configure group settings in 46xxsettings.txt with a group specific value for the call server address. It will work regardless which switch port the phone is connected to but you have to set the group value for every single phone manually​

The problem here is, later on you will not know which phone is setup for server A or B. You would need both Voice-VLAN's everywhere. So, Yes, you can sent some phones to server A and the others to server B. But it will limit your flexibility, as you can only register half of the phones to server A. You need to have a spreadsheet at hand when registering phones, or stick a label on them. If you have only one Voice-VLAN per building/floor/switch, then nothing is won against the method above. Just more troubles.
Grouping is used for VPN Phones, or to give different settings i.e. to lobby phones.


Anyway, we both agree:
There is no need for load sharing. One server will handle this load. A second server is for backing up - and should be located at a different site. And it must not be considered a mirror of the primary server. It's another networked unit.

Another part of our discussion is, there is one registration on the extension level and another on the user level.
I.e. if we direct part of the phones to server A, but the User is created in system B, we have hot-desking users where we did not expect them. For normal office operation at a single site, all normal Users shall be from the same system.

 

[derfloh]

The first point was regarding provisioning the voice VLAN Very LLDP. DHCP is not an option if you have only one underlying data VLAN but different voice VLANs "on top".

The group based option would work with only one data and one voice VLAN. Same DHCP, same voice VLAN, same file server, same 46xxsettings.txt for all phones. Provisioning of call server address would be a group specific value for MCIPADD (don't know the correct name right now).

You are not restricted to use group settings for VPN or lobby phones. You can adjust every available value that can be used in 46xxsettings.txt

 

[qtelcom]

great discussion and explains exactly what i needed to hear. Thanks to both derfloh and Albus2.

 

[Albus2]

@derfloh

Valid points.

Maybe not relevant to this example, but just to give another option:
If there is one data lan and multiple voice vlans (problem: how to send phones to the correct voice vlan?) or it's just one data and one voice vlan, but no device to define option 242:
Solution 1 is to use LLDP capable switches all over. And make sure they know Avaya phones.
Solution 2 is to attach a new phone, press * to go to the craft menu and leave all values except the vlan ID. Set it as needed. Also set VLANTEST to 0. The phone will restart in the correct vlan and stays there. In this application, VLANTEST must be disabled as it would not help if the phone goes to untagged data lan.

 

[derfloh]

I only used LLDP with Avaya switches so far. They know Avaya phones but as far as I know the only need it for NAC and for ADAC (auto detection, auto configuration).

To simply tell the phone which VLAN to use the switch doesn't has to know Avaya phones. The switch only tells the phone that voice traffic hast to use a specific VLAN.