Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

PRI to SIP 5

Status
Not open for further replies.

tone70

Technical User
Dec 14, 2007
238
US
Hello to All:

Our dialtone provider, AT&T is coming on strong with moving us from PRI to SIP trunking. I'm doing some research on the pros & cons. I'd like to hear from those who have converted recently and those who have been using SIP for a while on the benefits, issues you've experienced, things of that nature. We're currently running CM 5.2.1, and I'd like to know also what other pieces are needed for the implementation. Any & all comments on the good and the bad are welcomed. Thanks In Advance.
 
First, expect any PRI to SIP project to last 2 years. AT&T is great for stability and call quality but their internal processes are difficult and they have nice, long lead times.

That said, they do offer competitive rates (if your spend is high enough to get the discounts) and Geo-redundant SIP trunks with automatic fail over.
If you have two or more data centers in different areas then you can have automatic routing of incoming calls to the other location for business continuity.
For example, if the SIP trunk in NYC drops then inbound traffic automatically routes over to LA.

Things you'll want:

Avaya Enterprise SBCs
These are part router, part firewall and help with SIP header manipulation. Depending on your size, you may want two servers for high availability. SBCs can handle multiple, simultaneous vendors so you do NOT need one per SIP trunk.

Avaya Session Manager / System Manager
Makes your life easier with setting up SIP routing. System Manager is a one-stop management for users, CM & other PBXs, and voicemail. Also used for Session Manager programming.

You can use an AT&T MPLS for SIP or an AT&T managed Internet circuit to save some money. They will install a managed router that will be the demarc - AT&T will monitor the circuit to the router. Any issues after that will be your responsibility.

Pro:
[ul]
[li]It's SIP! You'll be able to brag to upper management about cutting edge technology (Mileage at your company may vary).[/li]
[li]Remote offices can still use centralized trunks and have local billing and phone numbers (Calls route out LA or NYC but with Chicago numbers. Chicago to Chicago calls look local and billed as local).[/li]
[li]Disaster Recovery - Automatic fail over of DIDs from one location to another (The best reason, IMO).[/li]
[li]Load Balancing - Session Manager can utilize both SIP trunks to keep outbound traffic flowing.[/li]
[li]Trunk Management - Reasonably quick to add more bandwidth and concurrent calls. If you install a 50mb circuit but only purchase 100 concurrent calls (about 10mb) then adding more calls is easy.[/li]

[/ul]
Con:
[ul]
[li]Money - There isn't a whole lot of savings (if any). Consider it a wash.[/li]
[li]Faxes - The Avaya Media-Gateways can handle T.38 faxing over IP but your clients may not. Best to keep 1 non-SIP trunk for faxing. You may need to split out Fax numbers from your DID range.[/li]
[li]Time - See previous gripes about AT&T.[/li]
[/ul]

Other Considerations
If you do want to plan for fail over, you need to double the capacity of concurrent calls at each location.
For example, NYC and LA each have an average concurrent call volume of 25 and an average peak of 50. You should set your SIP trunks to 100 concurrent calls so that if LA does fail over to NYC then the concurrent call volume is there to handle all calls. You could purchase less concurrent calls and run the risk of a few dropped calls in the event of such a major outage, if your DR plan accepts such risk.
 
Great thread... Do they support running their SIP over another providers MPLS? We have a separate carrier for our data network and aren't going to replace or add a 2nd mpls network. Also, isn't there a single point of failure using ATT for both voice(sip) and data? I thought I had read somewhere where you'd want a totally different SIP service provider than data network. I'm getting the push from ATT and Verizon as well.
 
Thank you for your excellent answer!! One other question: Will we need to go to SIP phones or can I keep my installed base of TDM & H.323 IP phones?
 
Internal SIP is completely separate from provider SIP so you can keep your H.323 and older digital phones.

I don't think you can use AT&T SIP over another MPLS since AT&T will want to manage the router. I don't know for sure, however.
You can use a separate AT&T data circuit for backup Internet access and SIP trunking. It is technically not as robust as MPLS but since it is a managed AT&T circuit connecting from a router at your site to AT&T (via local carrier for last mile, of course) it is pretty stable. I haven't had any issues with SIP trunks that way.

The least preferred method would be to use an existing Internet provider and any other SIP provider. Voice traffic can be encrypted but there's no QOS in the Wild World Web. It's not horrible and lots of companies do this but I would not recommend it for Enterprise size.

Yes, if AT&T or the LEC has a serious outage then your voice and data drop but that happens with PRI, too. I had a multi-slot smart jack fail six PRI due to a faulty power supply - I never heard of it before then! The local outage issue can be mitigated with a Geo-redundant circuit in another office and state. The managed router can be a single point of failure, too.
 
Thank you for your excellent answer!! One other question: Will we need to go to SIP phones or can I keep my installed base of TDM & H.323 IP phones?
 
We added SIP trunks to our headquarters 2 years ago and have been porting numbers to it from PRIs and disconnecting them ever since. We have one Aura system with media gateways around the world. The service we got is USA only so we can get numbers from any location in the USA coming in the one SIP trunk.

There is a huge cost savings once you get over your initial buy. Typical PRI for us is $600+-/month (with taxes) with 20-40 DIDs. Moving those DIDs to SIP cost $1.50/month for each block of 20. You just need enough bandwidth and concurrent calls on the SIP trunk. We added 22 offices in those 2 years where our old standard was a local PRI. Lots of savings there. You do have to account for the extra bandwidth usage to remote locations that are being served by the central SIP trunk.

SIP trunks are connections to external world. SIP phones are internal. You can do one without the other however SIP phones do not have a cost savings unless you eliminate the phones and use soft phones. We still use 3500 4621 h.323 and 2420 digital phones. No SIP. It doesn't matter what your internal network vendor is. This SIP trunk will terminate on the phone system (SBC/SM/CM). it should not touch your network.

All SIP is not the same. Providers and system implementations vary and that can cause problems. We still have trouble making calls to companies that use Cisco (and SIP I believe). Calls to their auto attendants fail over to voicemail because of something in our SIP header that we cannot change without breaking other things. Cant remember the details now but there are threads about it.

Our vendor does not let us make calls out that do not use a DID they have assigned to the SIP trunk. So backing up other trunks does not work without faking the outgoing number to something on the trunk. The geo redundancy that we looked at was a backup SIP trunk that you paid for but did not use unless the primary failed.



 
We have our MPLS data side from a different vendor than our SIP trunks and it works just fine. You have to remember that the MPLS is only for calls that stay within our WAN and that only calls bound for destinations outside of our WAN go out the SIP trunking. If a SIP call is internal to the WAN all setup and communication between endpoints remains totally within the WAN and your WAN vendor should be totally agnostic towards your packets. If a call comes in via SIP trunking and is being sent to a SIP endpoint within your network the data circuit vendor should still be agnostic towards your packets, but if they are not you can manipulate the SIP header to eliminate any hint of the originating carrier (while researching vendors to replace your current data circuit supplier).

Another important reason for the Session Manager is that it is where your SIP endpoints register to. If you are moving to SIP phones you MUST have a Session Manager for that and a few other reasons.

In our case the cost savings have been very significant and ongoing.

There are different models for how the SIP trunking is charged out. Bandwidth, per minute of use basis, burst capabilities, amount of DID's carried, etc all can govern pricing.

You should also consider the availability of DID's from you SIP carrier for the local exchanges you may have a presence in. We found that a lot of the mom and pop local phone companies in our neck of the woods had a vise hold on all the available numbers for an area and until our SIP vendor actually got a POP for an area we couldn't get numbers for that area.

Another quite large consideration for SIP is since it is such an agile solution it becomes quite mobile and that raises 911 considerations that will likely figure into your overall cost of implementation and operation.
 
Granted I'm also the guy who just started a thread for a release 6 switch, but my experience with SIP has been pretty bad overall; call quality that's inconsistent at best, and issues with reliability come to mind.

Disaster Recovery - Automatic fail over of DIDs from one location to another (The best reason, IMO)

Out of curiosity, couldn't this just be done with conditional call forward at the LEC's switch, or something along those lines when the circuit fails?
 
AT&T has AERS, Alternate Enhanced Redirect Solution, that allows you to redirect phone numbers, but it isn't automatic - at least I have only seen it implemented manually. It allows you to point one DID to another. For example, 212 555 1212 can be redirected to 404 777 1234 if there is a problem.

You can set up as many pairs as you want but it is REALLY expensive and you have to dedicate the target numbers for the redirect so you end up losing the use of them for other purposes. Prior to SIP, I used that for "critical" numbers only. With SIP, the whole DID range can redirect.

 
My central office experience is pretty limited, but I know the 5ESS has among all the typical call forwarding parameters like busy, no answer and whatnot, an option to forward when something is simply non-reachable. I'd be willing to bet you could if nothing else, kluge it into applying that to a trunk group or a DID block if there's no other options.

Of course the bigger problem, aside from cost as you mentioned, is whether or not the company will even turn it on for you in the first place.
 
You can do a lot of that as it is today by setting up a secondary preference in the trunk group forms. I "Had" AT&T swing all ten spans in one PRI group to a backup trunk group made up with 12 POTS lines when the group was down.

Today I have these setup to go to one pots line with a coverage to a second line if its busy. This only happens when the main PRI groups is down.

As for the OP and SIP, I really do not see anything that great that would make me jump on it. If this place of business were part of an international company, I would already be there. But were not. I forget with the ROI was just for statewide trunking on SIP was. But I believe it was around 10 years to pay for the original investment. So like the others stated, it's probably a wash and more headaches.

When is the last time you helped someone, just because you were able to?

For the best response to a question, read faq690-6594


 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top