• V20: 3CX Re-engineered. Get V20 for increased security, better call management, a new admin console and Windows softphone. Learn More.

Anyone using AT&T IP Flex and geo redundant sites?

Status
Not open for further replies.

Bartman10

Joined
Aug 8, 2018
Messages
6
Reaction score
1
Firstly please excuse me as I'm a network guy and am going to butcher the phone speed in my description.

I'm working with AT&T and a 3CX contractor to come up with a HA geo redundant solution. I have have them at other locations but the "phone guys" handled the setup.

Does anyone using a 3CX system have AT&T IP Flex service with two geo redundant sites? The IP Flex service will detect if primary site is up or down. If down it will auto route the inbound call to the backup site. Backup site 3CX server has list of "dummy extension" and will route the call over the site-to-site VPN to primary site 3CX server which will ring the handset.
Contractor says they can get this setup.

Then I asked ok, what about outbound calls? How with the phone system at primary site handle outbound calls when the primary SIP server is down? How would it know to route the outbound call over the site-to-site VPN to the backup location's 3CX server and then out the backup location?

Sorry if I'm messing this up.. but I think you can follow what I'm looking at.

Do I need a SBC maybe that would detect links that are up and route the call? That way the 3CX systems would only need to know about the SIP connection to the SBC... right? or maybe I have no idea what I"m talking about.
 
Does anyone using a 3CX system have AT&T IP Flex service with two geo redundant sites? The IP Flex service will detect if primary site is up or down. If down it will auto route the inbound call to the backup site. Backup site 3CX server has list of "dummy extension" and will route the call over the site-to-site VPN to primary site 3CX server which will ring the handset.
Contractor says they can get this setup.

So the fact that you are using AT&T IP Flex doesn't really matter. As long as the provider can handle CFU (call forward unavailable) the setup is the same regardless of the provider. It's just a matter of configuration depending on whether they are doing a catch-all forward (any TN at the primary site forwarded to a single TN at the backup site), a 1-to-1 forward for each TN, or whether it's a fully redundant trunk setup where the call delivery is attempted to the primary site first and then the secondary site.

hen I asked ok, what about outbound calls? How with the phone system at primary site handle outbound calls when the primary SIP server is down? How would it know to route the outbound call over the site-to-site VPN to the backup location's 3CX server and then out the backup location?

I'm confused since you mention working with a 3CX contractor who said they could to the first part of the equation. Then you said you asked them about outgoing calls and then... So do they not know how to handle this part or are you so super excited you couldn't wait for their response so you asked here? :) Anyways the answer to this is super simple. In the outbound rules at the primary site you can set up to 5 different trunks or routes per rule. 3CX will try them in order. So if the primary site SIP connection is up it will go out there. If it is down, it will attempt to go out that way and fail and then try the second option, etc. Now you will likely incur a delay since AT&T is IP authenticated meaning 3CX will try and wait for the timeout since it won't receive an error message but otherwise calls will flow normally.
 
Ok let me try to explain again.

-I pick up my phone at primary site to make an outbound call.
-Route 1 is down
-Do I have to stay on the line 30 sec for 3CX to send the call to Route 2?
-Is this 30 sec fail over to Route 2 per call? Or is route 2 remembered. At this point the 3CX system would be receiving errors from Route 1 so it should know it's down.
-If Route 2 is rembemberd, how do we fail back to Route 1 when it becomes available again?

What can convergence be speed up? How can route 2 be selected faster.

30 sec seems a really long time to tell a user to stand there with the phone in their hand while the system keeps trying to send the call over Route 1.

Is my only option to assign a different outbound dialing digit to Route 2. For example users dial "9" for outside line. If that line does not work, try "8". 8 will use Route 2.

Just spoke with a friend at another company and his Avaya system will detect Route 1 is dead and send the call over Route 2 in seconds. Then when errors stop from Route 1, or Route 1 again provides a dial tone calls with resume over Route 1.

What am I missing?
 
3CX will use the first route that is on line, that means from the top down it will try each one until the call goes out. all this happens instantly.
 
Instantly... that's not what I'm finding in the documentation or what my contractor is saying.

https://www.3cx.com/docs/multiple-outbound-call-routes/
" This will result in an outbound call delay of 32s until route 2-5 is selected which cannot be overcome (or avoided)."

I assume you mean if the 3CX receives one of the listed error codes. What if it's totally dead and does not receive any code? How long will it take until it tries Route2?
 
3CX will use the first route that is on line, that means from the top down it will try each one until the call goes out. all this happens instantly.
let me clarify my last statement, this happens if you have a outbound rule with more than one outbound route
 

Attachments

  • out.PNG
    out.PNG
    4 KB · Views: 5
Instantly... that's not what I'm finding in the documentation or what my contractor is saying.

https://www.3cx.com/docs/multiple-outbound-call-routes/
" This will result in an outbound call delay of 32s until route 2-5 is selected which cannot be overcome (or avoided)."

I assume you mean if the 3CX receives one of the listed error codes. What if it's totally dead and does not receive any code? How long will it take until it tries Route2?

Ok so you are on here trying to find out if your contractor knows what they are talking about.

Just spoke with a friend at another company and his Avaya system will detect Route 1 is dead and send the call over Route 2 in seconds. Then when errors stop from Route 1, or Route 1 again provides a dial tone calls with resume over Route 1.


This has nothing to do with 3CX vs Avaya. The Avaya would behave exactly the same if the provided dialtone is the same. Because AT&T is providing you IP authenticated trunks the PBX (3CX or Avaya or anyone else) has no status of the trunk as far as up/down. It appears to always be up. So 3CX will send the outbound call attempt to AT&T (route 1) and wait for a response. If the down condition generates an error (SIP server is up but they can't route the call) then the next route will be tried with minimal delay. But if the SIP server is not responsive then this is where the delay comes from and this delay will always be there (per call). If you had registration-based trunks and the trunk was unregistered then the call would immediately fail over to the next route. The only thing Avaya might do differently is have a configurable delay on how long it will wait for but the steps would still be the same. So the option is to wait out the timeout or put a secondary dialplan (like leading 8) for calls to go out the backup route directly.
 
TBWD sorry mate.. you are 100% wrong. The 3CX document I linked says so and here's the reply from 3CX
"
  1. Does every call have to hit the 32 second timeout?
    1. Yes,
  2. Why can't we edit that timeout?
    1. This is a SIP standard per RFC 3261 (https://tools.ietf.org/html/rfc3261) and is not something that can be changed inside of 3CX,
  3. How is that 32 second timeout determined?
    1. See #2,
  4. What happens if we get no response from the SIP Provider?
    1. The outbound call will transfer to the second outbound route (after 32 seconds)
"

No idea what you are talking about but it's not what I'm describing.

So the question is how do others do it so easy? IP Flex is kind of a big service and I simply don't believe everyone sits on a dead call for 32 sec waiting for the call to fail over to the 2nd route.
I understand if 3CX get's a error back from Route1 it will fail over but we already had a case where this did not happen. We had to log in and manually change the route order.

I understand going with a different provider would have the option to authenticate with the SIP provider but I'm stuck with what I've been given for now.

One option I have proposed is if Route1 fails with no error and users could try dialing "8" to get an outside line on the alternative path. Management doesn't really love this idea.

I keep getting challenged with "how do others do this"? I have to admit I'm not finding that answer. AT&T does IP registration. The AT&T engineer I'm working with has experience working with other phone system instigators and says this can be made to fail over quickly.

I simply don't believe everyone using AT&T IP Flex tells their users to wait 32 sec to fail over to a backup connection in the even the primary is lost.
This is what I'm getting at.
 
I simply don't believe everyone using AT&T IP Flex tells their users to wait 32 sec to fail over to a backup connection in the even the primary is lost.

You are correct. This is not an AT&T issue, this is a PBX/PBX vendor issue. It's not AT&T's job or responsibility to decide how fast the PBX bails on a failed or potential failed call.

I keep getting challenged with "how do others do this"? I have to admit I'm not finding that answer. AT&T does IP registration. The AT&T engineer I'm working with has experience working with other phone system instigators and says this can be made to fail over quickly.

Again, I don't think you are comparing Apples to Apples and having spoken to several AT&T "engineers" I would take take what they say with a grain of salt. There's several possible work arounds for this but quickly is a relative thing. If I could tell you how to cut the current wait in half that would be amazing, but your end users would still bitch because 16 seconds is still a long time. But you may want to fact check your engineer. For one, AT&T use to provide Cisco routers and you would get two SIP servers (still IP authenticated). Now AT&T is doing Edgewater devices and you are given a local IP (typically 1.1.1.1 which is not bad on their part since that's a valid public IP and I've had to ask them to change that many times on data only circuits). So the experience may have changed with the equipment change. It could also be that these other PBXs he's referencing used a PRI hand-off and it could be that with the vPRI AT&T registers on the CPE which would give an immediate error response allowing failover. Lastly, the other PBX products might have a call time or some other mechanism to cut the call attempt short/trigger secondary routing outside of the SIP timer stated in the RFC. The only one that really matters to you from a 3CX perspective would be the PRI option.
 
I have to jump in here. Calls can absolutely be re-routed in seconds without the user knowing it. It all has to do with the call control and the Session Border Controller (SBC) and not the 3CX version which is really a VPN bridge. Do not flame me, it is what it is. If you want to re-route calls get an SBC like EdgeWater or AudioCodes they route calls using a route table to Proxy's which are usually other SBCs.

The key to making all of this work is the SIP OPTION message, it is the equivalent of the data PING. The OPTION message is used to determine if the other end is alive and if not then other conditions like re-routing can take place. For this to happen, you need another Proxy in the service providers network to re-route to; usually this is for high-end SIP trunk services because the service provider has to provision the trunk in two SBC's in their cloud. The proxy's are geographically separated like New York and Chicago, so if there is a failure the OPTION message will trigger a re-route based on the timers configured. Depending on the service, you can even round-robin calls between the proxy's. You absolutely will pay for these higher end redundant services. I did the SBCs for the VA Administration call centers all four of them and when there was failure the user doesn't even know it, it happens that fast.

You really don't want the PBX to be involved in the core network signaling per se. It is an end point to the core network, the SBC provides the security, mediation and routing control at the edge of the core network which is all another ball of wax on its own. The 3CX SBC serves a purpose and works well for small sites; it is not a SBC true Session Border Controller, as it tunnels everything more like a VPN. If you are putting the 3CX PBX in larger sites put in an SBC, by pass the firewall and save yourself a lot of support time.

I put an EdgeWater 2900E SBC in with every 3CX customer of mine. Why because it solves problems. I don't have QoS issues. I get alarms when something stops working. I don't have NAT issues. I can see what the problem is at the customer site remotely. This is not an endorsement but proof positive that SBCs work. I get asked all of the time do I really need this, my response is always "Yes, because I can't support you when you have a problem." For $25 a month, the customer isn't going drop a brick like they would over $2,000+ SBC plus installation.

The AT&T IP Flex with the EdgeMarc 4808 is more of a "static" routed end-point to the customer. Meaning it is SIP trunk on ramp without a lot of bells and whistles. AT&T can manage the device and provide the proxy and registration locally, you are not registering into their core cloud at least that is how I configured the last AT&T implementation I did. On larger sites I have setup, the SBC was pointing to proxy's in the AT&T network. On sites with the 4808, they have point to the local proxy on the 4808.
 
  • Like
Reactions: gd1
Status
Not open for further replies.
Get 3CX - Absolutely Free!

Link up your team and customers Phone System Live Chat Video Conferencing

Hosted or Self-managed. Up to 10 users free forever. No credit card. Try risk free.

3CX
A 3CX Account with that email already exists. You will be redirected to the Customer Portal to sign in or reset your password if you've forgotten it.