How many links in your chain?

 

By Simon Woodhead

TLDR: We’ve added even more directs so most of your UK fixed, mobile and non-geographic calls will route directly to their destination, improving quality, speed of connection and resilience to failure.

We’ve just brought up yet another direct interconnect to a major mobile operator and are in the final stages of the last one. This gives us the full deck of major operators. This compounds the numerous other bilaterals we have with fixed line operators and number hosts. It has taken years and years, no small amount of arm twisting, but we’re about done. So what does it mean?

Some people think voice services, particularly wholesale voice, are a commodity to be compared on price alone, and we occasionally get drawn into the naive “beating some carriers up” for the same traffic they’ve dangled in the last several rounds of supposed procurement. Simultaneously though, voice is the single most quality sensitive service there is – who cares if their porn is two seconds delayed, or their cat picture downloads out of sequence behind the scenes. Prima-facie sensitive services like streaming video allow a generous buffer so you won’t notice any packet-loss anyway. But you can’t buffer Aunt Mabel, delays of more than 160ms are really disruptive to conversations and heaven forbid things arrive out of sequence.

If the whole industry were still simply reselling the incumbent, then yes, it’d be a commodity because the underlying service is the same and whether you have reseller A’s logo on the bill or reseller B’s logo on the bill doesn’t really matter. And nobody knows or cares about reseller C, D, E and F who are in the chain apparently to make the economics of putting a logo on a bill work! But that isn’t where the world is, particularly the world of IP Voice.

There are thousands of resellers in the UK market, although many will be reselling each other, and there’s 400 odd assignees of phone numbers according to Ofcom data. In reality though, those assignees are hosted by a very small group of which Simwood is no small part. Then add in the obvious self-hosted networks such as the big mobile operators and the real functional market is not small, but far smaller than it first appears to a casual observer.

Within, there are those who are “sweating” TDM assets bought 30 odd years ago, those who started life as a student project 20 odd years ago and have spent lots on cars but zero on infrastructure since, those who thought Huawei was a sensible thing to deploy everywhere and then thankfully a core of sensible operators with their own strengths and weaknesses that we consider peers. Oh, there’s also a boatload of spivs who’ll tell you they sit in this group but in reality are hosted by or reselling someone else and, erm, lying. This includes big sounding international operators who claim a global network and are long on schmooze but in reality are just reselling or have single-point of failure interconnects to the incumbent.

A chain is only as strong as its weakest link and if you think of your call from one end to the other, how many links are there? There’ll be more than you’re told of that I’m pretty certain, but you have no control over what spiv, archaic architecture, Chinese switches, or single point of failure is in the path and no visibility of it whatsoever. Privacy, resilience and of course call quality are undeniably affected. Do you care or do you genuinely believe that all those mouths can be fed for cheaper rates, with no compromise anywhere else?

An alternative approach is to completely eliminate all that and ensure that as much traffic as is possible flows directly end-to-end. That means if you hand Simwood a call for Vodafone, it’ll go from Simwood to Vodafone and vice versa. If another 10 operators that otherwise would be in the chain fail, who cares, the call isn’t touching them. Extend that out to all major operators in the market and that is pretty much where Simwood is today with the majority of calls flowing directly.

Another really exciting aspect of this is that when routing directly, there’s only two parties to the discussion about such things as codecs. There’s no transcoding at every hop between systems, or no dropping back to a common 50 year old standard that suits everyone in the chain to avoid doing so. Want a gucci VoLTE optimised codec directly from the handset into your application? That’s possible – more later. Need to turn up masses of traffic from network A to an application on network B for a Government department during a pandemic? We did that years ago.

We’ve always believed that “better” trumps “cheaper”, but recognise we’re a relatively lone voice. However, it turns out that if you cut the other dozen links out the chain, all with shareholders to satisfy and debt to repay, there’s a cost benefit. So I would argue that not only is routing through us “better”, it is also going to represent the “best value” for those who recognise it.

Still want “cheap”?