By Frazer Barnett
There are some of the more esoteric aspects of the Simwood service which I feel do not get enough attention. Channels, while viewed by many as a commodity that is identical across providers, is one of part of our portfolio I think warrants some explanation, especially as we haven’t really talked about it since 2016.
Simwood provides channels, which is a unit of voice connectivity to make a single concurrent call. We operate an N+many architecture; each Simwood availability zone is sized and specified to be able to run the entire Simwood network on its own, and your trunks are available in all simultaneously. That’s something else I feel we do not talk enough about, but shall be a blog for another time.
Unlike some carriers, we do not insist that you pick, in advance, which of our availability zones you are going to interconnect with or how much capacity to procure at each. Instead, we simply ask that you size your interconnect on an N+1 basis, using sound engineering principles, and let the Simwood architecture take care of the rest.
Let’s say for arguments sake that Simwood Slough suffers an outage; you don’t need to worry about whether or not you have a second “SBC” configured in Simwood Volta, or even Simwood Los Angeles, that is taken care of, and heaven forbid, should it happen, we hope you would not even notice. If, simultaneously, we lost two zones, again, you should not notice, whereas that scenario with some other providers could be catastrophic.
We also do not want to be the guys that give your customers a busy tone because there’s a popular act in the X-Factor final. We do not impose hard channel limits in the normal operation of the network; we are more than happy for transient use of headroom in our network by our customers – maybe you took on a customer that has more calls than you anticipated, or one of your other carriers had a brief outage – you have an unexpected extra call that needs to be conveyed, we can convey it, why would we not?
The quid pro quo is that we expect our customers to size their interconnects with us properly. There’s a lot of debate about what good engineering practice is; personally, my starting point is channels = peak concurrent calls / 0.4 , giving N+1 and 20% headroom for unexpected transient peaks, but no two of our customers are the same, and the right answer will be a matter for their own design teams.
Simon put it like this recently – “approach your sizing right, and we’ll flex around the odd issue you experience or miscalculation, but don’t take the mick.”
Of course, ‘don’t take the mick’ is not a term you’ll find in a contract, so I’ll outline what Simwood considers to be the start of the mickey-taking spectrum on channels – it’ll be a burst on two separate days, in a given rolling 30 day period, of more than 50% of the channels purchased.
Why 50%? We have primary and back-up channels. When you buy channels from us, half are placed in the primary pool, and half in the back-up pool – that’s what gives you the N+many architecture of your interconnect with Simwood as standard. If you’re over 50% utilisation on the total channels bought, then you’re using your back-up as your main interconnect, and in the event of network distress, that’s bad.
The vast majority of our customers are professionals and size their interconnects correctly. We reserve the right to charge mickey-takers that persistently view the headroom we offer as free capacity though. If any of you want to have a chat about what this means in practice for you, give me a call!