The Secrets of a Successful SIP Trunk Migration

By Andrew Prokop of SIP Adventures, a unified communications blog

Learning never exhausts the mind.
-- Leonardo da Vinci

I go to a fair number of tradeshows and industry conferences throughout the year, and as much as I love presenting at these events, I am just as thrilled to attend sessions given by my communications colleagues. It's rare when I don't pick up a tip or two, and there have been a number of sessions that have changed the way I view a product or a particular technology.

This year, I presented two breakout sessions at Avaya's IAUG Engage 2016 conference and attended at least a half dozen more. I would love to write about all my favorites, but for now, I will pick one.

For me, Larry Riba's "Secrets of a Successful SIP Trunking Migration" was a highlight. It's not that I haven't been living and breathing SIP migrations for the past several years. Believe me, I've paid my dues, time and time again. Still, hearing personal experiences allow me to see things that I might not otherwise see. Even the best of us fall into patterns and tend to repeat the same methodologies over and over again.

Larry began his presentation by discussing the benefits of SIP trunking. While most were not new to me (e.g. better redundancy and failover, cost savings, reduction in inbound and outbound trunks, etc.), a few were not on my personal list of "Why SIP?" For instance:

• Simplified call recording infrastructure with the capability to record in stereo
• Centralizing SIP trunks can often become the catalyst for improving WAN resiliency and increasing bandwidth, benefits that are then shared with non-voice traffic

Following the benefits of SIP, it was off to checklists for preparation, design decisions, choosing a SIP provider, planning, and avoiding pitfalls.

Everything begins with an inventory of the existing TDM infrastructure. This involves detailing voice circuits and DIDs and should be fairly easy to complete using configuration records and billing statements.

The next part is much harder. You may know all the DIDs owned by your enterprise, but determining how they are used takes an examination of overall usage patterns, current costs, specific or location usages, and busy hour requirements. The goal is to not only meet the current needs, but right-size the organization following a centralization of trunks. For example, the number of required trunks may drop substantially once they become a shared resource spread across a large geographical area and over many time zones.

My nerdy soul loved this part of the presentation. Here, Larry discussed:

• Designing for reliability, redundancy, and self-healing -- You need a really good reason not to deploy everything in high availability pairs. This means that everything has a backup. One SBC becomes an HA pair. That HA pair is backed up by another HA pair in a geographically distributed disaster recovery data center. The same holds true for WAN circuits, carrier connections, session managers, power, etc.
• Optimizing network regions -- What may have been considered best practices years ago needs to be reexamined for SIP.
• Implementing voice quality monitoring -- Be prepared to proactively solve problems before your users flood your helpdesk with complaints about lousy sounding phone calls.
• Examine or create bandwidth needs, codec requirements, and call admission control policies
• Determine the required number of DSPs -- The number may go up or possibly down after moving to SIP.

SIP brings along a lot of flexibility, but at some point you need to put a stake in the ground and choose a path. Some of the decision points include:

• Codecs -- G.711 vs. G.729. Do you sacrifice call quality for reduced bandwidth?
• How will branch survivability be implemented (if at all)?
• Shared vs. dedicated MPLS circuits
• Redundant carriers
• What do you do with all those analog devices -- fax machines, modems, etc.?
• What is the right SBC for your enterprise?
• Who does the work -- PBX vendor, reseller, or internal resources?

Choosing your SIP carrier should never be an afterthought. Although it's tempting to stick with your current TDM provider, this is the time when you need to consider the following:

• NPA NXX footprint -- Who delivers the services exactly where they are needed?
• Failover capabilities for active calls -- If resiliency is paramount, consider multiple links at every ingress/egress point. Understand how BGP is supported by the carrier.
• Don't make bursting a hard requirement -- Rather, do your homework ahead of time in determining your organization's busy call hour.

As a technologist, I hate having to think about money, but as a practical person, I can't ignore it. Here, the following should be considered:

• Non-recurring costs:
  • Initial Implementation (circuits, features, etc.)
  • DID porting
• Monthly recurring costs:
  • DID numbers
  • Toll-free numbers
  • Toll-free features
  • Toll-free talk paths
  • Talk paths for redundancy
  • Incoming Caller ID Name delivery

You've done your due diligence and determined what you want and from who you want it from. You now need to come up with a clear and concise plan on how to implement everything. This includes:

• A verification and test lab
• Create test DIDs
• Port individual DIDs for selected users (e.g. telecom team)
• Prepare a detailed implementation process, cut sheet templates, and test plan(s)
• Assign implementers and checkers. Make sure that every change is checked for accuracy
• Cut outbound local/toll at each site a week or more before DID port
• Determine support model -- onsite or remote
• Port toll-free traffic last

Even the best laid plans sometimes go awry. Here are a few potential problem areas that must be considered:

• Call recording -- often a nightmare if not carefully planned
• Fax
• Modems
• TDD
• Carrier screening of outbound Caller ID for certain call types
• 911
• Long carrier lead times on circuits and/or features
• Number porting
• DTMF recognition on inbound calls

You won't know if it works if you don't properly test everything. Areas of high concern include:

• Inbound
• DID to station
• Toll-free
• IVR and call vectoring
• DID with blocked or unavailable Caller ID
• DID to voicemail
• Outbound (include on-net locations and lots of DTMF)
• Tandem (extend to mobile, call forward, carrier take back and transfer, etc.)
• Fax (inbound and outbound)
• Failover
• SIP circuits
• SBCs
• Session managers

Life doesn't come with guarantees, and Murphy has a way of showing up when he's least expected, but the better the plan, the more likely you will have a positive outcome. The tips that Larry shared will take you a long way to achieving the results you want, but it's up to you to customize the approach to one that delivers the results that make sense for your organization.

Note: This blog post first appeared on No Jitter. 

Andrew Prokop writes about all things unified communications on his popular blog, SIP Adventures.