Making A Direct Connect Connection Resilient

Does not cover choosing the connection model, virtual interface and BGP setup, reaching many VPCs, or encryption. Those are separate references.

Execute commands using the AWS MCP server when connected (sandboxed execution, audit logging, observability). Fall back to the AWS CLI otherwise. The Direct Connect console is regional; pass the customer’s working --region on every aws directconnect command.

Table of Contents

• Overview
• Decision: resiliency model
• The single-device trap
• Resiliency model vs failover speed
• Failover test and VPN backup
• Troubleshooting
• Procedure
• Additional Resources

Decision: resiliency model

Constraints:

• You MUST match the customer’s uptime target to the resiliency model that delivers it before any connections are ordered, because the higher targets require a specific multi-location layout that cannot be retrofitted cheaply.
• You MUST use the Resiliency Toolkit wizard for Maximum and High Resiliency, since it prevents terminating redundant connections on the same device.

The single-device trap

Constraints:

• You MUST NOT let the customer order two connections and assume they are redundant; terminating both on the same Direct Connect device leaves a single point of failure. The toolkit prevents this.

Resiliency model vs failover speed

These are two different questions. The toolkit sets the topology; it does not set how fast failover converges.

Constraints:

• You MUST set the convergence-time expectation: with default BGP timers, failover can take around 90 seconds, while enabling Bidirectional Forwarding Detection (BFD) and tuning the BGP hold timer brings it under a second.
• You SHOULD offer BFD and hold-timer tuning when the customer needs fast failover, rather than implying the resiliency model alone determines failover speed. When the customer asks “is my failover 3 seconds or 90 seconds,” this is the answer.

Failover test and VPN backup

Constraints:

• You MUST close with the toolkit’s failover test, which brings down the BGP session to confirm traffic routes to the redundant virtual interface, so the first real test is not an incident.
• You SHOULD offer a Site-to-Site VPN as a backup path for customers who need resilience beyond what the connections alone provide; note the MTU drops to 1500 during VPN failover.

Troubleshooting

Production is on a single connection and an outage took it all down

A single connection carries a 95% service level agreement with no redundancy. Move to High or Maximum Resiliency using the toolkit.

Two connections did not protect against a device failure

Both terminated on the same Direct Connect device. Use the Resiliency Toolkit, which prevents this.

Failover is much slower than expected

Default BGP timers leave convergence around 90 seconds. Enable BFD and tune the hold timer for sub-second failover.

Failover was never tested before an incident

Run the toolkit failover test, which brings down the BGP session to verify rerouting, before the next maintenance event.

Procedure

Overview

This procedure matches the uptime target to a resiliency model, builds it through the toolkit, tunes failover speed when needed, runs the failover test, and surfaces the console link.

Parameters

• uptime_target (required): The service level the customer needs (maps to a resiliency model).
• location_count (required): How many Direct Connect locations are available.
• fast_failover (optional): Whether sub-second failover (BFD) is needed.
• vpn_backup (optional): Whether a Site-to-Site VPN backup path is wanted.

Constraints for parameter acquisition:

• You MUST establish the uptime target and available locations before choosing a model.

Steps

1. Verify dependencies

Constraints:

• You MUST confirm credentials with aws sts get-caller-identity.
• You MUST use ephemeral IAM credentials (e.g., AWS SSO, an instance profile, or aws sts assume-role) rather than long-lived IAM user access keys for Direct Connect management operations.
• You MUST confirm how many Direct Connect locations the customer can use, since Maximum Resiliency needs more than one.

2. Build the resiliency model through the toolkit

Constraints:

• You MUST use the Resiliency Toolkit connection wizard for Maximum and High Resiliency so redundant connections do not land on the same device.
• You MUST NOT mix bandwidths within one Resiliency Toolkit configuration.

3. Tune failover speed when needed

Constraints:

• When the customer needs fast failover, you MUST enable BFD and tune the BGP hold timer, and set the expectation that default timers leave convergence around 90 seconds.

4. Run the failover test

Constraints:

• You MUST discover the virtual interfaces and BGP peer addresses on the connection before starting the test:

  Copy

  aws directconnect describe-virtual-interfaces --connection-id {connection_id} \
    --query 'virtualInterfaces[].{Id:virtualInterfaceId,PeerAddress:bgpPeers[0].customerAddress}' \
    --output table --region {region}

  Capture the virtualInterfaceId and customerAddress from the response.

• You MUST run the BGP failover test and confirm traffic moves to the redundant path:

  Copy

  aws directconnect start-bgp-failover-test --virtual-interface-id {virtual_interface_id} \
    --bgp-peers {peer_address} --test-duration-in-minutes 180 --region {region}

• You MUST poll for test completion:

  Copy

  aws directconnect list-virtual-interface-test-history \
    --virtual-interface-id {virtual_interface_id} \
    --query "virtualInterfaceTestHistory[0].status" --output text --region {region}

• For early termination (if the customer needs to abort):

  Copy

  aws directconnect stop-bgp-failover-test \
    --virtual-interface-id {virtual_interface_id} --region {region}

• You SHOULD recommend CloudWatch alarms on Direct Connect connection state and virtual interface BGP status metrics, and confirm CloudTrail is capturing directconnect API calls with log file validation enabled and the trail encrypted with a KMS key, and any CloudWatch Logs log groups receiving these events or alarm state data encrypted with a KMS key, so state changes trigger alerts and configuration changes are audited with assured log integrity and confidentiality rather than relying on manual detection.

• You SHOULD ensure any SNS topics receiving Direct Connect alarm notifications are encrypted with a KMS key and that subscriptions are restricted to authorized operations personnel.

• You SHOULD remind the customer that the redundant Direct Connect connections are not encrypted in transit by default, and point them to the encrypting-traffic reference if the workload requires encryption.

5. Surface the console link

Constraints:

• You MUST present the Direct Connect connections console link, filling {region}, and tell the customer to confirm the redundant connections and their states:

  Copy

  https://console.aws.amazon.com/directconnect/v2/home?region={region}#/connections

Security Considerations

• Not encrypted by default. Direct Connect does not encrypt traffic in transit. You MUST treat encryption as a separate, deliberate step (MACsec or a private IP Site-to-Site VPN) before regulated or sensitive data crosses the redundant connections. See the encrypting-traffic reference.
• Ephemeral credentials. You MUST use ephemeral IAM credentials (e.g., AWS SSO, an instance profile, or aws sts assume-role) for Direct Connect management operations rather than long-lived IAM user access keys.
• Least-privilege IAM. You MUST scope IAM permissions for directconnect API actions to the specific actions and resource ARNs each principal needs, and MUST NOT grant directconnect:* on resource * or attach any *FullAccess managed policy.

Additional Resources

• Direct Connect connection options (AWS Direct Connect User Guide) (opens in a new tab)
• AWS Direct Connect Resiliency Toolkit (AWS Direct Connect User Guide) (opens in a new tab)
• Resilience in AWS Direct Connect (AWS Direct Connect User Guide) (opens in a new tab)
• Enabling BFD for a Direct Connect connection (AWS re:Post) (opens in a new tab)
• AWS Direct Connect Service Level Agreement (opens in a new tab)