When you hear “redundant” you probably think of something that’s no longer needed. Something that might have a better alternative or has a replacement. Well you’re half right. There’s a secondary definition in the technology field; data redundancy. It describes a piece that can be omitted without loss of meaning or function. Having data redundancy is extremely important in telecommunications; a phone system needs to be online at all hours of the day to ensure no calls are missed. Imagine all the revenue you could miss or the customer outrage if your company stopped answering calls and inquiries. In real world conditions where we rely on downstream and upstream carriers and “Bring your own Bandwidth” we rely on many different systems. Here’s a peek at how we do it:
Strategic diversification of server roles. Most smaller companies and carriers subscribe to a model that a phone server must do all things in one server. While this is much less complex, it unfortunately leaves much to be desired when it comes to reliability. If that one server goes down, EVERYTHING ceases to function. What Audian has done is break out each different role into multiple servers. This first layer of redundancy means that should a registration server go down, perhaps phones can’t make new calls, but inbound and existing calls will continue. Though more complex, Audian uses 4-5 servers in a cluster to replace what many do on just one.
Server role redundancy and load balancing. Now that we have broken things into different roles, our next redundant layer is to provide multiple servers WITHIN the datacenter. What this does is prevent any one server role from going down. As an example, each data center has 2-3 media servers that handle the actual voice mediation. If one of these servers were to go down, calls fail over instantly (you aren’t even disconnected!) and seamless redundancy is achieved. This allows us to sustain multiple failures within the datacenter without seeing an interruption in service. In addition, we are able to perfectly load balance all of our calls and data. All of our servers run in active-active roles, which means rather than having a primary and backup server which is a waste of resources, we use all available servers load balanced, and can fail over instantly.
Different data centers per geographic location. Last but not least, we maintain multiple geographic data centers. By doing this, we can ensure that even with a large outage or natural disaster, all calls will simply fail over to another location completely. By doing this, we also gain the ability to have calls route to the closest centers, so an east coast client will route to a different data center than an east coast client. This leads to less call lag and a stronger more reliable connection. Since there are servers all over the USA it means you’ll always have the best possible connection no matter where you go!