Alright, if you haven't read Challenges of an Internet Retailer’s VOIP Implementation Part 1, please read it first. I think you'll find that the differences between Part 1 and Part 2 of the posts are pretty staggering.
We've finally made the transition from our digital-to-analog-to-digital lines over to full digital. To be clearer, we had eight phone lines coming into an AdTran over T1 that were converted from digital to analog. We then took each line and converted it back to digital using an Audiocodes MP-118 analog gateway. The tweaking required to make it work well was intense and not a project I ever want to repeat. In place of our analog lines, we've added a PRI line with 23 channels (plus one D channel), greatly expanding the number of lines available and drastically improving quality. Surprisingly, this change only increased our monthly bill by a few dollars.
We are still using a trixbox solution and I did a fair amount of research, mostly on the trixbox forums, to determine what PRI card would be best/easiest/least expensive to install and configure. In the end, I decided on the Sangoma A101DE. The reason I went with Sangoma, and not the, perhaps, more natural selection of a Digium card, was the wide spread vocal support in the VoIP community. Not only did users swear by Sangoma products, but the vendor's presence among the community was also readily apparent; so I knew I would be able to get help when I ran into a snag (which we all knew was going to happen). I used the same VoIP vendor that I had used in the past as I knew I could depend on them to get me the A101DE I needed.
Since my last phone system post, I have also found a Grandstream GXP-2000 firmware that seems to work well, but we have transitioned most of our call center users over to softphones. While I was waiting for the Sangoma card to arrive, a friend of mine that works for a large multinational corporation dropped off a huge box of Polycom phones. Apparently, his company has a very liberal "dead phone" policy and all of these were going to be junked. Picture this, a user says, "Hey IT! My phone doesn't work!" IT provisions a new phone, brings it to the user, and puts the old phone in the trash pile. He brought me the trash pile. More on this in a minute...
So, I received the Sangoma card within a few days of the order and proceeded to replace the production system with a temp system. I swapped out the systems and proceeded to install the card on the production system. Sangoma had recently posted a very simple installation walkthrough on the trixbox forums, so I just followed instructions found there. Piece of cake, right? Maybe not. At the same time, I was also coordinating with our T1 vendor on the PRI installation. Keep in mind, I have never been involved in a PRI build-out or a trixbox system that consumes PRI. I was flying in the dark, other than community documentation. I had already read a considerable amount about users who had installed this card on similar systems. I spent a fair amount of my time learning to match settings between the Sangoma setup and the PRI. The majority of the rest of my time was spent building out a scalable queuing solution that would support multiple customer service queues for separate ecommerce websites, business queuing for various business functions, and individual user assignments to queues and DIDs (Direct Inward Dial). There was also some up front maintenance in ensuring that the box was secured (no default passwords, alternate ports, etc.) and provisioned for our company's network.
Everything was ready to go, except for one glaring issue... the system could not be truly tested until our T1 was live! The months of research, preparation, planning and building would all come to an exciting "everything works" climax or a dismal "nothing works" thud! When I mentioned the impending go live to our development department team leader, he asked, "How long have you been preparing?" I replied, "A few Months." To which he replied, "Well, nothing ever goes 100% right on a project that you're months deep in without live testing." My confidence was obliterated. Was he right? What would I do if something went wrong? I only had a window of a few days to get it to full functionality once our T1 was live. Would I be able to get the necessary support from Sangoma or from the trixbox community in that short window? Would I be forced to buy really expensive support time from someone much smarter than me? Had I done enough research to have a clear understanding of what I was doing? I panicked a bit. But, I calmed down quickly, since I knew that there was nothing I could do until testing.
On testing day, I arrived early, checked all systems to ensure readiness for testing and waited for the telecom tech to arrive. I paced nervously, knowing that I was helpless and alone until my hour of glory or defeat... At 7:00AM, on the dot, he arrived, Customer Premises Equipment in hand. He came in straight away and began the work of testing our new PRI. As he tested the equipment that he had provisioned the night before, he noticed something that prevented the go live. So I waited as he called his project manager to ensure that everything was set up appropriately. I hovered over his shoulder to see what was going on. I'm sure I annoyed him enough to make him want to punch me in the face. As I peered into the screen of the battered field laptop he was using to telnet into the gateway, I could see all of the settings used to prepare my connection. I could see everything I need to confirm that my setup was appropriate! So, I continued to hover, elated as I watched him work through all of the settings. Everything I had set appeared to be correct. By 10:00AM, he had worked through his bugs and we were ready to get the party started. So, we did. We plugged it in, turned it on, and I pulled up a softphone on my laptop. I placed my first test call, a local call, to my cell phone... it worked! I placed another test call from my cell phone to my softphone... it worked! I went down the list of test calls, international, emergency, information, toll-free, etc. I tested queues, DIDs, ring groups, caller ID data, reporting... everything worked! The first time! Was I dreaming? Did I escape from this project unscathed? The answer was an assured "YES!"
The only thing that was really left to do was to get a phone for everyone who didn't have one in the past. Thanks to my very generous friend, I had a pile of Polycoms to go through. So, I brought one home, tested it out and used trixbox's endpoint manager to make provisioning the phones a snap. Eight of the first eleven phones were (nearly fully-) functional! A few had non-functional speakers or broken microphones, but they worked well enough for most users. I even spent the three extra minutes it took to build a customized logo.
My development team leader had put doubt in me, and I knew he was likely right, that not everything would go as planned. The phone system has been such a thorn in the side of the IT department from our humble beginnings with a two-line Vonage solution, to our analog hunting solution, to our digital-to-analog-to-digital line solution, to this. I did not want to see another half-completed phone system that barely meets the needs of the organization. I wanted to build it out to its fullest potential and I wanted it to be scalable and I wanted it to be inexpensive. I think I accomplished that. This one feels good.