Question. How do you eat an elephant?
Answer. One bite at a time.
My piano teacher had those words taped to the wall of his studio and they are as applicable to a large communications project as they are to learning a new piece of music. You will wind up completely frustrated if you attempt to tackle either one without breaking them down into bite-sized pieces.
I was recently engaged to assist a major health organization in their move from TDM to SIP. While much of the project involved upgrading their carrier trunks, quite a bit flowed down to the users and their use of unified communications. They wanted the flexibility and cost savings of SIP trunks and the productivity of SIP endpoints.
Like that elephant, they would have been met with certain failure if too much was attempted too quickly. Although the desire to make all this happen yesterday was there, so was the need to ensure that things were done correctly. They were still reeling from a difficult technology rollout and no one wanted to suffer through yet another catastrophe. It was my job to help identify the tasks, quantify the risks, and create an implantation plan.
When attempting a major communications project, the very first thing you need to ask yourself is “How will you measure success?” Put another way, “What do you expect this thing to look like when it’s done?”
While that sounds obvious, you would be amazed at how many people never ask the question. They have an idea as to what they are attempting to accomplish, but they never fully visualize how tomorrow will be different from today.
Start at the Edge
The best place to start on a TDM to SIP project is with the existing TDM infrastructure. Who provides the carrier trunks? Is it more than one provider? How many trunks come in? How many trunks go out?
It’s essential that you understand any special services that might exist on those trunks. While every T1 might look like every other T1, each might have its own configuration and just as importantly, associated carrier offerings.
For example, I once worked with a company that utilized a network-based DNIS (dialed number identification service) routing application. A call would arrive at the carrier which would then alter the DNIS before sending it on to the customer. A little investigation on my part determined that that DNIS service was only available on their TDM trunks and was not present on the SIP trunks.
Also, what are all the uses an enterprise applies to their existing trunks? Some carriers are reluctant to allow you to run a predictive dialer on their SIP trunks. This is not something you want to discover after you’ve decommissioned your PRIs.
I recently wrote about 9-1-1 in 9-1-1 and SIP. While putting 9-1-1 calls on SIP is very doable, you cannot expect them to work without thinking through your configuration and potential use cases. Are you dealing with a single site? Are there remote offices? Are you implementing mobile SIP?
Document your 9-1-1 expectations early on to avoid being met with potentially devastating and tragic surprises.
Session Border Controllers
Although some people will attempt to implement SIP trunks without a Session Border Controller (SBC), this is not something I would ever recommend. I could go on and on about why an SBC is important, but for now I will just say security, SIP adaptation, flexible routing, network topology hiding, and resiliency. Trust me, you need them all.
Choosing the right SBC requires some thought, but it’s a relatively easy job once you identity your needs. Rather than repeating myself too much, please see my articles Choosing the Right SBC – It all Depends and Andrew’s Session Border Controller Checklist for my thoughts on how to make the right decision.
With TDM, you are limited to the number of calls you can carry on a T1. One T1 gives you at most 24 calls (23 if the D-Channel is used for signaling). If you need 25 calls, you buy another T1.
SIP, however, is limited by bandwidth and not the physical medium. That same 24-channel T1 can give me 40+ G.729 calls when I convert it to a data pipe.
Determining the amount of bandwidth you need for your SIP calls is crucial. Too much and you are wasting money. Too little and calls don’t go through.
I always begin by looking at the current call volumes. Some companies know their numbers off-hand, but more than likely it’s necessary to look at carrier reports to determine the number of calls along with peak, and off-peak times. I then follow the steps I laid out in Calculating Bandwidth for SIP trunks.
I recommend that during the initial phases you over provision bandwidth and scale it back once things are up and running. You do not want to chase bandwidth problems during rollout. Find and fix the architecture issues first.
SIP trunks are great, but the real sizzle comes from adding SIP users. This is especially true when it comes to mobile SIP devices and replacing those old, black telephones with unified communications endpoints and applications.
During this part of an engagement, I spend a lot of time understanding how a company works and how its employees would benefit from unified communications. It’s important that this be done at a business unit level. I don’t like trying to shove every employee into the same mold. The needs of the sales folks may be completely different from what accounts receivable specialists expect.
I would never recommend putting SIP trunks in without an SBC and my recommendation is even stronger when it comes to remote SIP users. Remember, you are exposing your network to the wild, wild Internet. It’s essential that you make sure that authorized SIP traffic flows in and out and unauthorized access is stopped dead in its tracks.
SIP is a protocol. Session management is a solution.
Session Management allows an enterprise to create its own private SIP cloud. Session management allows you to turn applications and trunks into pooled, shared resources. Session management gives resiliency and survivability to SIP endpoints.
As part of any SIP engagement, I look to see how session management might be utilized by an enterprise and help them make an intelligent decision as to how it should be implemented.
I hope this gives you an idea as to how you might approach a SIP transformation. Again, biting off too much at one time will most likely cause you to choke. Identify your elephant. Pick a place to start and start eating – one bite at a time.