Author: Scott Feagans, Co-founder of NetXpress
Looking around, it is evident that our competitors have adopted a more traditional network architecture which relies on a series of network layers for exchange access and multi-client distribution architecture. With larger deployments, the distribution network becomes even more complex, adding significant latency to your trading network. This model typically has several switch hops (we’ve seen 4-5 in some cases…ouch!) As a result, even if a competitor has deployed low latency switches for each of these hops, such as Arista 7150s, the latency of your trading stack is increased by 760ns -1400ns.
At NetXpress, we take a completely different approach to network architecture and ensure the lowest latency possible as a managed service provider. By deploying client servers with direct connectivity to our layer 1 exchange switches, NetXpress delivers a deterministic latency profile of 5-95 ns (the exact latency profile varies per exchange venue).
HOW HAVE WE DONE THIS?
Our Secret Sauce (well, some of it anyway!)
- It’s really quite simple. We removed all the unnecessary hops (mentioned earlier) to provide our clients with the flattest architecture available by managed service providers today: NetXpress Ultra. The combination of a proprietary solution and Metamako’s ultra-low latency layer 1 switches, delivers native market access to NetXpress Ultra clients in 95 nanoseconds or less (dependent on exchange venue). Additionally, we utilize the latest Cisco Nexus 3500x series switches to ensure the fastest, full layer 3, inter-server communication in the industry today.
ARE THERE ANY OTHER BENEFITS?
Network Stability at its Finest
- Because NetXpress has eliminated the unnecessary switch hops in our architecture, we’ve dramatically reduced points of failure.
- As your servers connect directly to the exchange at the layer 1 level, you will never be impacted by any other client’s changes. No more unnecessary change notifications!
- In our competitor’s traditional layer 3 based architecture, everyone could be dramatically impacted by a wrong change to the community switch. NetXpress is built for scale and as a client, you are only impacted by what happens on your port.
We Futureproofed It – You Are in Control
- When your trading stack is deployed in a managed service provider’s network, you are subject to the switches they’ve deployed for distribution. Real talk: who knows what year those switches were deployed? Could they possibly be the same day they started business? Once a service provider begins deploying multiple clients on their shared switches, they are reluctant to upgrade. Distribution switch migrations are expensive, disruptive and an overall nightmare to swap out. Their reluctance impacts your trading latencies. Unfortunately, you won’t be able to do anything about it.
- With the only network devices being the client’s backend server switches, our Ultra clients can be in the driver’s seat when it comes to upgrades. If a newer, faster switch comes out, our Ultra clients can simply swap out those devices on their schedule rather than our ours.
You Really Can Have It All
- You might anticipate that our vastly improved technology and architecture would be too unbearably expensive to consider. Don’t worry; you really can have the best of both worlds. Our pricing is extremely competitive to other providers that are still operating under a slower, legacy model. How? It costs us less. Our flat, uncomplicated network requires less to operate and maintain. We pass those operational savings onto our clients.
- With exchange fees increasing year over year, we’re all in the same boat. As a managed service provider, our clients benefit from sharing exchange connectivity at a fractionalized price. If you could take the same connectivity at a latency profile that is either the same or better than if you were to do it yourself, why wouldn’t you?
For more information, please reach out to the NetXpress team: firstname.lastname@example.org or 312-871-3190.