A great way to support tens of thousands of agents and other knowledge workers in a unified communications (UC) environment is to use SIP Trunking: abstracting your communication services to a software level to provide easy provisioning and new collaboration capabilities to your workforce. But like anything, the more SIP Trunks you have, the more expensive it gets. And if you have multiple data centers and want to load balance your SIP Trunks for better service and redundancy, that gets expensive, too.
A major home goods wholesaler was running into the issue of having too many SIP Trunks, too many data centers and too much cost. They wanted to see if they could consolidate the number of SIP Trunks they needed to support their operations (including more than 20,000 agents and knowledge workers), and they wanted to explore the possibility of load balancing their SIP Trunks between two of their own data centers.
SIP Trunk Load balancing is traditionally handled by service providers, and although there are infrastructural tools to support this in theory, no OEM officially supports it or has published any validated solutions matching the desired requirements. The big challenge was risk: trying this on their own, even just to explore the possibility, would require huge investments into hardware and skillsets just to try and prove out something that may not, in the end, even be possible. That’s where we come in.
After a series of briefing sessions and workshops hosted at our Advanced Technology Center (ATC), we performed a proof of concept (POC), putting together a high-level architectural design and test plan which would represent their voice environment across redundant data centers and remote branches. The proposed solution would provide a full SIP proxy load balancing, including SDP (Session Description Protocol) and RTP (Real-Time Protocol).
We built a functional simulation of their entire voice network to prove out the concept: they could actually dial “branch to branch” and “service provider to branch” within the simulated topology to perform real-world validation to ensure the SIP trunks were being load-balanced correctly.
We partnered with both F5 and Citrix to integrate their load balancing technology into the designed environment, providing our customer insight into two separate functional solutions to learn the benefits and challenges of both. While performing the proof of concept tests, we evaluated alternative solutions that would provide the same value while still using their existing environment, simplifying the solution and saving the customer costs down the road.
Leveraging the ATC, we were able to provide a multi-vendor, integrated POC and educate the wholesaler on additional data center solutions and configuration possibilities. At the end of the tests, we determined that F5 met all of the wholesaler’s requirements and would be the load-balancing framework should that path be selected (Citrix could load balance SDP but not RTP, while F5 could load balance both protocols).
Not only were we able to prove out the possibility of load balancing SIP Trunks at the enterprise layer, but we were able to show it using two different technologies. This means the possibility of hundreds of thousands of dollars in savings for the wholesaler, all while still providing necessary capabilities and redundancy.
By showing the feasibility of this kind of solution in their existing environment, we helped the wholesaler not only prove the load balancing component, but also identify other areas for better SIP configuration and management to save costs given their current setup. As a result of the POC, the home goods wholesaler can now make informed, cost-saving decisions and investments in the unified communications space going forward.