I’m pretty focused on A-Team’s “Business & Technology of Low-Latency Trading” events coming up in London in less than two weeks, and in New York City at the month end. We’ll have Brennan Carley of Spread Networks on a panel in NYC, and so I’m reminded that Spread has put out a couple of interesting news releases in the past week or so, that warrant some attention.
First off is the news that Spread has reduced the latency of its managed wave service, which runs from Carteret, NJ (across the road from Nasdaq) to 350 East Cermak Road in Chicago (the world’s largest data centre, where many futures trading firms locate their systems).
The reduction: a full millisecond – from a guaranteed 15.75 milliseconds round-trip down to 14.75 milliseconds. That’s a huge reduction and is not down to some R&D tinkering. Rather, it’s the result of a fundamental product configuration change for commercial reasons. Spread’s underlying connectivity – its dark fibre service – clocks in at an industry leading 13.33 milliseconds round-trip, by dint of the very direct route that Spread constructed. It’s for customers who require the lowest latency, are prepared to pay for it, and who can handle the integration of network gear at each end.
The wave service – based on the same dark fibre – offers an ethernet interface with bandwidth as low as 1 gigabit/second chunks, at a slower speed, at a lower price. And it’s been deliberately engineered to do that – a different product for a different type of customer. I’ve asked Spread a couple of times how they are slowing down the wave service but they aren’t saying – “technical reasons” is the best answer I’ve got. My best guess is they’ve purposely put in place an awful lot of excess fibre to do it, but perhaps it’s some kind of configuration of the network interfaces at either end (the vendor of which is also not public – does anyone know?).
Whatever, it’s clearly possibly to alter the latency of the wave service significantly with little effort, as commercial requirements require. In other news, Spread has concluded a reseller deal with network specialist CFN Services. CFN – which creates custom networks for customers – can now include Spread’s wave service in 1 gigabit/second and sub-1 gigabit/second bandwidths in its portfolio.
In fact, it has exclusivity over the sub-1G offering. For Spread, the deal allows it to address a segment of the market that’s beyond its primary focus in a cost-effective way – without spreading its own sales resources too thin, as it were. While for CFN, it provides a latency boost in its ability to service customers and offer continual latency improvement.
Asked what the market for ‘sub-1G but still low latency’ is, CFN executives suggest a few sweet spots for it, including where data traffic is more transactional – such as order routing applications – than market data oriented, and where traffic to a particular execution venue is low. It’s also popular for trading firms that previously used third party execution management services, and who are looking to move to direct market connectivity. So latency and bandwidth do not always go together.
Oh, and finally … according to recent announcements, Spread has extended its NJ reach to established data centres in Secaucus, Weehawken and Newark. It’s working with another metro network provider for those, though it’s not saying who. Now it’s back to the events. Come and hear from Brennan, who is on a panel hosted by ADVA Optical Networking. Maybe you can get some answers that I can’t 🙂