SAP-C02 Practice Question 22: Direct Connect Maximum Resiliency
AWS Certified Solutions Architect – Professional (SAP-C02) · Test 3 · question 2 of 10
A payments processor connects its primary data centre to AWS with two 10 Gbps dedicated AWS Direct Connect connections. Both connections were ordered at the same Direct Connect location, and during a recent maintenance window an outage at that facility took both of them down at once. Traffic did not fail over anywhere and the payment gateway was unreachable for 40 minutes.
The remediation plan states that the hybrid connection must be eligible for the highest service commitment that AWS publishes for Direct Connect, that the design must survive the loss of an entire Direct Connect location as well as the loss of a single Direct Connect device, and that bandwidth must not drop below 10 Gbps during any single failure. The network team will run BGP on every connection and is willing to order additional circuits.
Which configuration should the team order?
Show answer and explanation
Answer:
The AWS Direct Connect Resiliency Toolkit publishes three models, and only one of them is eligible for the 99.99% SLA: maximum resiliency, which uses separate connections that terminate on separate devices in more than one location. That is the only model that protects against device failure, connectivity failure, and the complete loss of a Direct Connect location at the same time, which is precisely the failure the postmortem described. Option D is that model, and with two 10 Gbps circuits surviving in the healthy location the 10 Gbps floor still holds.
Option A is the development and test model scaled up: separate devices, but still one location. It survives a device failure and would have done nothing during the facility outage that caused the incident. Bundling the four circuits into one LAG makes it worse, because a LAG must terminate on a single Direct Connect device; that single device becomes the failure domain.
Option B is the high resiliency model, two single connections to two locations. It does protect against a location failure, but it is documented as eligible for the 99.9% SLA, not 99.99%, because a single device failure in one location removes that entire site's capacity. It also drops the link to 10 Gbps of total capacity with no headroom.
Option C leaves the real problem in place: both Direct Connect connections still share one location. A Site-to-Site VPN is a legitimate backup pattern, but it rides the public internet, its throughput per tunnel is far below 10 Gbps, and VPN capacity is not covered by the Direct Connect SLA at all.
Reference: AWS Direct Connect Resiliency Toolkit