Network Simplification
- 1. NETWORK SIMPLIFICATION USE CASE This paper discusses a usecasewhere after an acquisition,the company needed to meet the synergy targets they had set as pa rtof the acquisition agreement. This is justone projectof many, but demonstrates an architectural process.Thespecific vendors and company arenot included to avoid violatingany intellectual property rights and non-disclosureagreements. THE CHALLENGE After acquisitions of three companies over the course of 2 years,the CTO issued the followingchallenges to the Network Architecture team to achieve the synergy targets of the acquisitions: 1. Reduce Network Costs 2. Reduce Operational complexity/simplify the Network The target was to start(and hopefully complete) this objectivewithin the year and to be Net Present Value (NPV) positivewithin one year of beginningthe project. THE PROCESS As the Principal Architectfor the Voice Network, I began lookingfor opportunities to meet the CTO’s challenges.One of the areas that I began to look at was removing the Class 5 CircuitSwitches.We had a pristineVoIP Network prior to the acquisitions.As partof the acquisitions,we inherited closeto 100 Class 5 CircuitSwitches from various vendors providingdifferentservices. Step 1: How to reduce Network Costs Step 2: How to simplify theNetwork by completely removing the CircuitSwitches. The followingsections detail theapproach taken for the two steps. REDUCE NETWORK COSTS Looking at the ways to reduce the Network Costs,I identified two opportunities that could happen independently. One opportunity was to optimize the interconnections to the Local Exchange Carriers (LEC) and Inter-Exchange Carriers (IXC).The other, smaller, opportunity was to replace the 911 circuits to the Selective Routers. In workingwith the Network Planningteam, we decided that the removal of the LEC and IXC trunks should occur before the 911 circuits.Thequickest win was to move all the PSTN bound traffic to the VoIP Network to let it Least Cost Route the traffic.Sincethere were still someEqual Access agreements in place,we needed to set up routingon the Class 5 CircuitSwitch to identify the non-Equal Access traffic thatit could route to the VoIP Network. We then set up trunks from each Class 5 CircuitSwitch to the VoIP Network and started routingthe PSTN bound traffic,both Local and Long Distance,across thosetrunks. The next step involved moving the inbound traffic fromthe PSTN to come into the VoIP Network that would route itto the proper Class 5 CircuitSwitch.Where there was sufficientLocal trunks on the VoIP Network, this was a matter of rehoming the Class 5 Circuit Switch’s Local Routing Number (and 10k blocks) to pointto the VoIP Network as its homingswitch in the LERG. Prior to the step, we set up the routingof the NPA-NXX of the LRN and 10k blocks on the VoIP Routing Engine to point to the circuits weset up for the outbound PSTN traffic.We had sized the VoIP to Class 5 CircuitSwitch trunks to handlethis additional traffic when we did the outbound to PSTN call flow. As we rolled the solution out across thenetwork, we gradually removed the trunks to the LEC from the Class 5 CircuitSwitches.Because of the better interconnect agreements and the efficiency gains,we were ableto reduce the Network Costs by roughly 20%. Unfortunately, we needed to spend Capital to buy the ports on VoIP that connected to the Class5 CircuitSwitch.However, the monthly costsavings paid for the one time capital expenses after justtwo months, so the projec twas still Net Present Value (NPV) positivewithin the year, so it achieved its goal.Another sideeffect of this project was that the Class 5 CircuitSwitches were servicingsomecities/RateCenters that the VoIP Network did not. In those instances,we needed to add these new Rate Centers to the VoIP footprint. This added some Capital Expense,but allowed the Product and Sales teams to sell VoIP
- 2. services to Rate Centers they previously couldn’t.In some instances where there wasn’tenough traffic to justify the build out, Product exited those Rate Centers, which resulted in some small revenue loss,but not enough to offset the build costs. The next step was to replacethe 911 circuitson the Class 5 CircuitSwitch with VoIP 911 Circuits.For this project, there wasn’t a lot of Network Expense to save, but itsimplified the operations job of managingthe 911 circuitsby combiningthem all to be on VoIP. For this project, we used a method we were sellingto Customers as a VoIP PositioningCenter (VPC) solution thatal lowed for non- geographic telephone numbers to receive proper 911 service(see https://en.wikipedia.org/wiki/Enhanced_9-1- 1#VoIP_enhanced_911 for more information).The solution entailed settingup an internal setof Emergency Services Routing Numbers (ESRN) on the Class 5 CircuitSwitch that itsent to the VoIP Network. The VoIP Network used the ESRN to route the ca ll to the proper Selective Router to get it to the proper Public ServiceAnsweringPoint for the geographic location of the Calling Telephone Number. Operations called this projecta success,becausethe VoIP Network offered 911 circuitredundancy that the Class 5 CircuitSwitch 911 circuits did not,so they sawfewer outages and fewer FCC reportable incidents. NETWORK SIMNPLIFICATION This step required more justification,so I began by gathering information,such as: Switch Location Number of Customer T1s Lease costs,where applicable,and expiration date(used for prioritization) Utilities costs Maintenance Costs I used this information to calculatethe savings gained and the replacement costs incurred for moving the circuitsto the VoIP Network. I did this on a switch by switch basis to showpositiveand negative overall savings/costs.I would showthe savings/costs over a year. The Capital Costs were one-time costs,whereas the savings were month over month. By showingthe savings versus the costs over 12 months, it showed when/if the switch removal went positive.At the end of the exercise, I could show that, at the Network-level, the project was a positivecostsavings,even if certain switches were not. The next step after showingthe project was worth doing was to convinceProductthat we could not/did not want to replicatea ll the services thatwere on the Class 5 CircuitSwitch,such as Equal Access or sub-T1 services/lineservices.Instead,we wanted to go after the majority of Customers (>90%) that were basic Primary RateInterface (PRI) Customers that did not require any Class 5 Features. Fortunately, I had been working a different project with the Voice Product team on how to differentiate ourselves from our competitors on the SIP Trunking product. One of the ways I had architected for them was a way to allowPRI Customers to use the SIP Trunking product (I know itsounds oxymoronic to say SIP Trunking with PRI handoff, but it was the only way for Product to get their heads around the idea). This simply meant that to move the Customers, we would sell them the new Producton trunks to the VoIP Network and turn off the trunks to the Class 5 CircuitSwitches.This was where the project stalled for the followingreasons: Product’s willingnessto losesome of the embedded revenue basethrough disconnects or lower rates on the new service o As further incentive, I quantified for them the likely risk of when the Class 5 CircuitSwitches would startfailing and in some cases,we had no way to replacethe parts, so they could losethe Customer revenue regardless. ConvincingCustomer Activation groups to do it, becauseit was not new revenue on which they were rated Waitingfor Product to prioritizethe IT work needed to enable this new flavor of SIP Trunking At the time of this writing,this step had not been started, but ithad been planned to happen in 2016 after I left the compa ny. CONCLUSION Although I cannot claimtotal victory,I did make significantprogress towards the goals the CTO laid out. There were other projects that had similar partial success for many of the same reasons listed above. Depending on the organization,there will always becases where an Architect cannot achievean entire plan.However, the job of the Architect is to show where savings,new revenue, or simplicity existand work with teams to execute on the ones on which the organization chooses to move. For me, as longas I am working on solvingchallengingproblems,I amsatisfied with whatever parts of the plan the organization does execute.