SIP Trunk+SIP PSTN

[sheraz butt]

Hi All,

I got 100 PSTN SIP lines i need to configure this any help.

Currents

CUCM 9-------->Voice Gateway 2801Router  with FXO line which we want to remove nowSo .

In new scenrio i got one SIP router from service provoder with 100 PSTN lines (SIP) and some information about ip address circut number and so on

So i make direct trunk with CUCM--->to SIP router or  CUCM--->Voice gateway===SIP TRUNK===>SIP PSTN (100) Router

Any configuration help

Regards

Sh

[Ayodeji Okanlawon · Accepted solution]

You need to do the ff

1. Configure a sip trunk from cucm to your gateway

2. Configure your gateway as a CUBE gateway..

http://www.cisco.com/en/US/products/sw/voicesw/ps5640/products_configuration_example09186a00808ead0f.shtml

For CUBE...

• Here are some thoughts for you

  1. Configure CUBE for media flow through. In this Mode CUBE acts as a true  B2BUA. Advantages you get include address hiding and security becaue CUBE  terminates and re-originate both signalling and Media. In this mode CUBE becomes  a point of demarcation from th external world.

  2. Configure CUCM to use Delayed Offer and CUBE to convert delayed offer to  ealry offer...This prevents the need for you to use MTP to send Early offer on  CUCM

  voice service voip

  early-offer forced

  3. Configure DTMF signalling method on sip trunk to "No preference" This  setting allows Unified CM to make an optimal decision for DTMF and to minimize  MTP allocation.

  4. Configure your CUBE to meet the requirements of your ITSP. Ask if they  have configuration templates or any specific configuration they like you to use.  This will save you time troubleshooting. Most of them dont use the default port  5060 because of security, confirm with your proivider what ports they use.

  voice service voip

  allow-connections sip to sip

  sip

  early-offer forced

  header-passing

  error-passthru

  5. Use SIP to SIP...Use end to end sip. CUCM---sip---CUBE--sip----ITSP

  6. Create a Trusted list of IP addresses on your CUBE is your CUBE IOS is  15.1 .2(T) and above.

  voice service voip

  ip address trusted list

    ipv4 203.0.113.100 255.255.255.255

    ipv4 192.0.2.0 255.255.255.0

  This is imprtant because sometimes your ITSP will send you a single ip  address for signalling and will then send media on a different IP adress. So get  all the IP address your ITSP is using and add them to the trust list as shown  above

  7. Configure your inbound and outbound dial-peer approriately

  Inbound Dial-Peer for calls from CUCM to CUBE (CUCM sending 9 +all  digits dialled to CUBE)

  dial-peer voice 100 voip
  description *** Inbound LAN side dial-peer  ***
  incoming called-number 9T
  session protocol sipv2
  codec  g711ulaw
  dtmf-relay rtp-nte

  Outbound Dial-Peer for calls from CUBE to CUCM (SP will be sending 10  digits inbound)

  
  dial-peer voice 200 voip
  description *** Outbound LAN side dial-peer  ***
  destination-pattern [2-9].........
  session protocol sipv2
  session  target ipv4:
  codec g711ulaw
  dtmf-relay rtp-nte

  Note: If more than 1 CUCM cluster exists, you will have to create multiple  such LAN dial-peers with “preference CLI” for CUCM redundancy/load balancing

  
  Inbound Dial-Peer for calls from SP to CUBE

  
  dial-peer voice 100 voip
  description *** Inbound WAN side dial-peer  ***------------------(catch-all for all inbound PSTN calls)
  incoming  called-number [2-9].........
  session protocol sipv2
  codec  g711ulaw
  dtmf-relay rtp-nte

  
  Outbound Dial-Peer for calls from CUBE to SP

  
  dial-peer voice 200 voip
  description *** Outbound WAN side dial-peer  ***
  translation-profile outgoing Digitstrip
  destination-pattern  9[2-9].........
  session protocol sipv2
  voice-class sip bind control source  gig0/1
  voice-class sip bind media source gig0/1
  session target  ipv4::XXXX (where XXXX is the port number your  provider is using if different from 5060)
  codec g711ulaw
  dtmf-relay  rtp-nte

  8. SIP Normalization:

  You may need to configure sip normnalization to modify sip headers, CLI etc.  A good example is during call forwarding. You may need to change your diversion  headers to  match the CLI your provider is expecting.. if your call forwarding  is failing during testing this may be the reason..We can help you with this.

  9. Media Resources

  Plan your solution properly. Consider if you will need Xcoders, MTP,  Conference bridge etc. You may avoid the need for xcoders if you confure your  regions properly and use voice class codecs on your sip profiles. It is  important to know if there are any endpoints in your network that do not support  dtmf relay rtp-nte. You can avoid the use of MTP if you configure your  dial-peers to have multiple dtmf types for thos phones that do not support  rtp-nte

  e.g

  dial-peer voice 1 voip

  session protocol sipv2

  dtmf-relay rtp-nte digit-drop sip-kpml (if your phones support kpml..then  this will be used)

  If in your environment you will need to do xcoding or CFB then ensure you  have PVDMS

  .10.FAX

  If you have FAX in your network, determine what fax protocol your sip  provider supports. Dont assume. Ask them and confirm in writing  what they  support. I have seen legal cases because of fax failures over sip trunks

  Configure your FAX devices in seperate device pools and use porefix to route  calls using G711 only. Even if you are using T38, ensure your fax use G711 to  establish the voice calls

  Finally

  11. Have a detailed and carefully planned TEST Plan. Test the FF:

  • Inbound and outbound Local, Long distance, International calls for G711  & G729 codecs (if supported by provider)
  • Outbound calls to information and emergency services
  • Caller ID and Calling Name Presentation
  • Supplementary services like Call Hold, Resume, Call Forward &  Transfer
  • DTMF Tests
  • Fax calls – T.38, modem pass-through--whichever one you decide to  use

Please rate all useful posts

"opportunity is a haughty goddess who waste no time with those who are unprepared"

[sheraz butt]

Hi ,

First of all thanks and impressive work.

1) this is what i understand    CUCM===Sip trunk===>CUBE===== SIP trunk==>PSTN sip router

2)we were using 2081 voice gateway as AA with tcl script because we have only 8 FXO line for now we have 100 or in future more so what you should suggest for AA like (IP-IVR) ? to handle multiple calls coming in.

correct me if i am wrong

Thanks in advance

[Ayodeji Okanlawon · Accepted solution]

Hi,

1. Yes you are correct..

2. You can use the 2801 as your CUBE and AA gateway. Your IVR tcl script can be used as you are using it now, you just need to change your config to match your voip dial-peer instead of your pots dial-peer

Please rate all useful posts

"opportunity is a haughty goddess who waste no time with those who are unprepared"

[jbollen]

Hi there,

With all due respect but I have a couple of comments:

1. "becaue CUBE  terminates and re-originate both signalling and Media". Are you sure about the media part?

As far as I saw in numerous wireshark traces, the RTP stream "before" and "after" the CUBE keep the same

SSRC and RTP parameters, only destination and source addresses change. When the CUBE needs to transcode,

you will have a "new" RTP session.

2 "Configure CUCM to use Delayed Offer and CUBE to convert delayed offer to  ealry offer...

This prevents the need for you to use MTP to send Early offer on  CUCM"

Well, the fact that CUCM is by default sending a Delayed Offer (DO) causes problems

for early media sent by the caller. E.g.: calls to an IVR requiring caller input before the

SIP dialogue is established. Changing the DO into and Early Offer (EO) on the CUBE is not

going to change that. I would recommend to configure EO all the way. Never understood why

CUCM is not doing that by default...

3. "voice-class sip bind control source  gig0/1

     voice-class sip bind media source gig0/1"

    I would recommend to use the IP address of a Loopback interface, it is

    a frequently seen practice when building SIP trunks to VoIP service providers.

    Imagine that you have multiple physical interfaces for redundancy, the failure

    of "Gig0/1" would render your SIP trunk unusable...

kind regards,

Jan

[Ayodeji Okanlawon]

1. "becaue CUBE  terminates and re-originate both signalling and Media". Are you sure about the media part?

If CUBE doesnt do this in media flow through, then it provides no point of demarcation. Yes I am 100% sure of that. CUBE terminates media and re-originate it. I dont know what wire shark traces you are referring to or looking, so I cant comment on that. Please refer to CUBE documentation for more information.

2. Why will you want to do that? Have you read the SRND guide? Enabling CUCM to do EO means that you will invoke MTP for each call leg. That in itself is not scalable and causes tons of problems..Eg fax etc

Early offer and Early media are two different things...We are talking EO here

3. Using loopback as nice as it sounds has its own issues. What happens if your ITSP doesnt have ip connectivity to your loopback ip..which in most cases is the case...It doesnt fit for all especially when WAN connectivity is involved

Please rate all useful posts

"opportunity is a haughty goddess who waste no time with those who are unprepared"

[jbollen]

Hi there,

Thank you for your reaction.

1. CUBE and media flow through. I suggest that you take a look at an RTP stream before and after

a CUBE: the only items changed are source & destination IP addresses and ports. Everything else

such as SSRC is the same *IF* the CUBE does not transcode. When it transcodes,

it does originate a new RTP stream of course. With respect to signaling, the CUBE does

create a complete new dialogue (different call-id a.o.) and acts - as you stated - as a B2BUA.

2. It is documented in the SRND, that the CUCM user cannot send his/her input (early media) when calling

an IVR in the case of a delayed offer. The IVR will need to receive an early offer before it can

process the early media from the user.

That loads of MTP's are causing tons of problems or scalability issues, is known for a long time.

It is a problem since the days of version 2.x hence I still wonder why CUCM still needs this

"suboptimal" configuration. If one does not expect scalability issues, I would recommend

using early offer and MTPs on IOS gateways.

3. Using the IP address of a loopback interface is very widely used for SIP trunks and BGP. Most service providers do it and recommend it.

kind regards,

Jan

[Ayodeji Okanlawon]

Perhaps there is a mis understanding here of what terminating media means...

Look at the sneario below...

ip phone--cucm--sip---cube---sip---itsp ----------------this represents our signalling paths

ip phone ---rtp---cube----rtp---itsp-----------------this represents our media path

Clearly media is terminated on CUBE..it doesnt flow directly to the endpoint from the ITSP..This is where CUBE is the demarcation point and also provides IP address hiding..the ITSP never knows about your internal network becuase its signalling and media terminates on the IP of the CUBE

It is a good idea to use loopback if you can..If you cant, there are other options. In My case, our ITSP cant route to my CUBE ip address..I am not saying its not a good practice I am saying that if you cant use it its not the eend of the world..I am not sure about the use of the word "most"

Please rate all useful posts

"opportunity is a haughty goddess who waste no time with those who are unprepared"