3.11. Direct Connect Resilience

What does Direct Connect architecture look like?

Diagram

What are the three main physical components in a basic Direct Connect architecture?

AWS region, DX location (data center), and customer premises.

What is a DX location in the context of AWS Direct Connect?

A third-party data center that houses AWS DX routers and optionally customer or communications partner routers.

How is the AWS region connected to DX locations?

With multiple high-speed, highly available network connections.

What is a cross connect in AWS Direct Connect?

A physical cable connecting the AWS DX router to a customer or provider DX router inside the DX location.

Is a single Direct Connect setup inherently resilient?

No, a single Direct Connect has multiple single points of failure and is not resilient by default.

What are some single points of failure in a basic Direct Connect setup?

DX location, AWS DX router, cross connect cable, customer DX router, extension cable to premises, customer premises, customer premises router.

Why is assuming Direct Connect is resilient a common misconception?

Because it's made of physical components, each of which can fail unless resilience is architected.

How can resilience be improved in Direct Connect architecture?

By provisioning multiple DX ports, routers, and extensions across redundant locations.

What does ordering two Direct Connects into the same DX location provide?

Separate DX routers for each connection, adding a layer of hardware redundancy.

What are the remaining SPOFs (single points of failure) in a dual-DX-router single-premises architecture?

DX location failure or customer premises failure.

What hidden SPOF might exist in a dual-connection setup?

Both extension cables might share the same physical path (e.g., sidewalk), creating a potential single point of failure.

How can you avoid SPOFs from cable routing in Direct Connect?

Use two physically separate customer premises and DX locations to ensure separate cable paths.

What does a fully resilient Direct Connect architecture look like?

Two DX locations, each with two DX ports and routers, connected to two separate customer premises, each with dual routers.

What kind of failures can a fully resilient Direct Connect architecture tolerate?

DX router failure, DX location failure, customer premises failure, extension cable failure.

Why is Direct Connect not resilient by default?

Because it relies on single physical components unless multiple connections are architected.

What is a good backup solution for Direct Connect?

A Site-to-Site VPN.

From an exam and real-world perspective, what is critical to understand about Direct Connect?

It must be explicitly architected for resilience; single DX setups are not resilient.

What is an improvement on resilience?

Multiple devices on DX and customer premises

What’s a second-level improvement?

Multiple customer premises

What’s a third-level improvement?

Multiple DX and Customer premises locations