Size Direct Connect ports with 50% redundancy headroom and price port hours plus data transfer out.
AWS resiliency guidance for Direct Connect is to provision so that peak traffic still fits after a failure, which for a two-port design means running each port at no more than 50% utilisation. Doubling the requirement and dividing by the port speed gives the port count, and the bill combines port hours — charged whether or not traffic flows — with a per-GB data transfer out rate that is materially cheaper than internet egress. A Direct Connect port takes weeks to provision through a partner, so discovering the second port is missing during an outage is not a problem you can fix on the day.
Direct Connect Bandwidth
provisioned = port speed × ports; resilient capacity = provisioned × 50%; ports needed = max(2, ceil(required × 2 ÷ port speed)).
provisioned = port speed × ports; resilient capacity = provisioned × 50%; ports needed = max(2, ceil(required × 2 ÷ port speed)). AWS resiliency guidance for Direct Connect is to provision so that peak traffic still fits after a failure, which for a two-port design means running each port at no more than 50% utilisation. Doubling the requirement and dividing by the port speed gives the port count, and the bill combines port hours — charged whether or not traffic flows — with a per-GB data transfer out rate that is materially cheaper than internet egress.
A Direct Connect port takes weeks to provision through a partner, so discovering the second port is missing during an outage is not a problem you can fix on the day.
This calculator takes 7 inputs: Peak bandwidth required, Port speed, Ports provisioned, Price per port-hour, Data transfer out per month, Data transfer out price per GB, Billable hours per month. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.
Because the surviving port has to absorb the failed port's traffic. At 80% per port a single failure leaves you 60% oversubscribed and every application feels it; at 50% the failover is invisible. AWS's maximum resiliency model goes further with separate devices and locations.
Yes for most designs. A Site-to-Site VPN over the internet is inexpensive standby capacity, though at about 1.25 Gbps per tunnel it will not match a 10 Gbps port, so size expectations for degraded rather than equal service.