AWS Certified Advanced Networking – Specialty (ANS-C00) — Question 364

A company has applications running in a single AWS Region and its on-premises data center in a hybrid mode. The company has a 1 Gbps AWS Direct Connect connection from the data center to AWS that is 65% utilized. The company has an AWS Enterprise Support plan.
The company is planning to deploy a new critical application on AWS that will connect with existing applications running in the data center. The application SLA requires a minimum of 99.9% network uptime between the data center and AWS.
What is the MOST cost-effective way to meet this SLA requirement?

Answer options

• A. Create a second virtual interface (VIF) on the existing Direct Connect connection, and terminate this VIF in the existing VPC. Use BGP for load balancing between the VIFs in active/active mode.
• B. Purchase an additional 1 Gbps Direct Connect connection from AWS in a different cross-connect location terminated in the associated Region. Provision a new virtual interface (VIF) to the existing VPC, and use BGP for load balancing.
• C. Set up two new hosted Direct Connect connections of 500 Mbps each through an AWS Direct Connect partner. Provision two virtual interfaces (VIFs) to the existing VPC on both Direct Connect connections, and use BGP for load balancing. Terminate the existing 1 Gbps Direct Connect connection.
• D. Purchase an additional 1 Gbps Direct Connect connection from AWS in the existing cross-connect location. Ask AWS to terminate this new connection in a different router. Provision two virtual interfaces (VIFs) to the same VPC on both Direct Connect connections, and use BGP for load balancing.

Correct answer: B

Explanation

To meet the 99.9% uptime SLA, AWS recommends a high-availability architecture utilizing dual Direct Connect connections terminated at different locations to prevent a single point of failure. Option B is the most cost-effective way to achieve this redundancy because it leverages the existing 1 Gbps connection and adds a second one at a different location. Option A and D do not provide true location redundancy, while Option C is invalid because 500 Mbps connections would fail to handle the existing 65% (650 Mbps) traffic load if one link goes down.