Business Policy Routing to Maintain a Good User Experience
We need to set the scene to properly explain business policy routing. Traditionally routing traffic across a network involves a routing protocol, calculating an efficient route to take it from its source to its destination. The route is based on a technical perspective and has very little to do with the context of the sort of traffic that is being sent, and all the traffic goes in one particular way. The methodology is reliable and resilient, it will fail-over when you run into an issue, via an alternative path, but there’s nothing within the traffic itself that impacts the route it takes.
Things are much more complex now, we are becoming more diverse in terms of the sort of applications that we use; the network is no longer just being used for browsing the internet or data applications, it’s being used for voice and video, and this is a much richer medium.
This is where the issues arise; different applications need different things, added to which you are in an environment where your applications aren’t all in the same place that they used to be and it becomes very distributed. For instance, you might not have to go from your office to your data centre, you might have to go from your office to your data centre that’s in the public cloud, or you might be using a software as a service application such as Google Hangouts. In this distributed world every application has got its own set of requirements. Business policy routing is the ability to be able to categorise applications and treat them in a specific way, both in terms of the route that is taken over the network, from its source to its destination, and also based on the minimum criteria that you need in order to ensure or, at least, make a reasonable effort of having a good user experience.
Business policy routing is a core aspect of SD-WAN. You look at the application and the things the application needs and then make a choice about the best route to send that traffic, in real-time based on the characteristics that you’ve got. It’s fundamentally different to traditional routing because it addresses one of the issues that we have, that is, you don’t have to ‘be down’ to have a bad experience. At Gyrocom we use the expression “degraded is the new down.” I can have a network link that’s up, but if the voice traffic is such that you can’t have a good conversation, the experience is poor. That’s where business policy routing really comes into things – it really takes us forward. It’s the next way of being able to treat traffic over the network. It’s becoming much more appropriate, much more prevalent, and it’s all about ensuring user experience.
The way networks still operate today is that you’ve got to have some sort of hard failure; a connection has got to go down and you’ve got to be able to actually break the thing in two in order for it to be able to failover. That’s not the experience that we get, we have situations where things aren’t quite working. The reality is that there are so many things that are involved in that, we need the ability to be able to create a policy that says; “I need this minimum set of criteria in order for this sort of application to work. And if it doesn’t, I need to look for options about where else I can get those criteria from.” What you don’t want is instability, you don’t want traffic constantly flipping over from one connection to another because there’s inefficiency in that. You want it to stay in as long as you’re making a set of criteria. That criteria might vary slightly, but as long as it’s within parameters where the end-user experience is okay, you don’t want it to flip backwards and forwards. For instance, an example of a business policy route would be, “When I’m supporting real-time communication, I don’t want the link latency over which I’m sending that traffic to exceed, 30 milliseconds. If it does, I need to look for another connection.”
In addition to that, “If I can’t find a connection that does that, I want to start remediating traffic” What we mean by remediating traffic is that there are certain techniques from a technology perspective that SD-WAN can start injecting into the environment. Forward error correction, packet order correction, and jitter buffering (an imperceptible delay that gets inserted into a conversation). In real-world speech, you and I talking to each other online, you can’t really perceive that there’s a delay. That delay actually gives both ends of the conversations, from a technology perspective, enough time to be able to catch up if there’s any sort of delay in ones and zeros being transitioned across the line because what you want is consistency of conversation. So, even if I just delay it by half a millisecond, it gives the technology time to catch up.
What you want above all else is to be reliable and consistent, and things are not always reliable and consistent. We have got lots of different mediums, especially when you’re increasingly talking about wireless connectivity, you’re not always in control of the characteristics of the connection. For example, with a 4G connection if it’s raining really heavily the signal between you and the radio masts that you’re connecting can be impacted by that. It can be impacted by a truck coming in between you and the line of sight of the signal.
The whole idea is to create an environment where you’ve got options. You don’t want to put yourself in a situation where you’ve only got one option in terms of a route. You always want to maximise the best option for what you’re trying to do. That can be based on cost, it can be based on characteristics or it can be based on medium. You want options because certain things that affect one option don’t affect another and so options are much lower cost.
Managing Director at Gyrocom - Helping customers build better networks.