Dedicated virtual servers
Dedicated cores and dedicated NVMe. No shared CPU, so your tail latency stops being a function of what somebody else is doing on the same host.
Who this suits
Production workloads where p99 matters more than average, which for anything trading is most of them. If you have ever watched a decoder run fine for an hour and then miss a burst for no reason you could name, this is the tier that fixes it.
What changes
Your machine sits in the same facility as the relay in Frankfurt, so the path between the feed and your process stops crossing networks neither of us controls. The feed itself is identical. What changes is how much of your own infrastructure is between it and you.
Getting the catalogue
Configurations are built to the workload and hardware pricing moves, so we send the current list rather than publishing one that goes stale. Tell us what you are running and roughly what it needs.
Email sam@shredstream.sh