AWS Security ChangesHomeSearch

AWS wellarchitected: Rewrite ADVREL05-BP01 as multi-AZ RTB Fabric bid gateway guidance

Service: wellarchitected · 2026-11-11 · Documentation medium

File: wellarchitected/latest/video-streaming-advertising-lens/advrel05-bp01.md · Type: network

Summary

Replaces the generic 'routine fault tolerance evaluation' best practice with concrete guidance to deploy RTB Fabric partner-facing bid gateways across at least two Availability Zones: requesting the 'Availability Zones per Gateway' service quota, provisioning gateways in separate subnets with adequate IPv4 capacity, configuring per-AZ security groups to allow HTTPS (TCP 443) inbound based on requester/responder role, and monitoring subnet AvailableIpAddressCount via ec2:DescribeSubnets and CloudWatch. Adds desired outcomes, anti-patterns, implementation steps, and related links.

Security assessment

The change is primarily reliability/resilience guidance (multi-AZ gateways, subnet IP capacity, quota requests), but it does document security-relevant configuration: per-AZ security groups permitting only HTTPS/TCP 443 inbound, scoped by the workload's requester/responder role direction, plus the IAM permission ec2:DescribeSubnets used by a monitoring Lambda. It describes network exposure control and least-privilege-ish traffic direction rather than fixing a specific vulnerability, so it is documentation of hardening practice, not a vulnerability remediation.

Evidence

  3. **Configure security groups per Availability Zone:** Configure security groups in each Availability Zone to allow HTTPS (TCP 443) inbound per your role (requester or responder).

Diff

diff --git a/wellarchitected/latest/video-streaming-advertising-lens/advrel05-bp01.md b/wellarchitected/latest/video-streaming-advertising-lens/advrel05-bp01.md
index dffca8e..aac3e91 100644
--- a//wellarchitected/latest/video-streaming-advertising-lens/advrel05-bp01.md
+++ b//wellarchitected/latest/video-streaming-advertising-lens/advrel05-bp01.md
@@ -7 +7 @@
-Implementation guidanceKey AWS servicesResources
+Implementation guidanceResources
@@ -9 +9 @@ Implementation guidanceKey AWS servicesResources
-# ADVREL05-BP01 Perform routine evaluation of your workload's fault tolerance capabilities
+# ADVREL05-BP01 Deploy partner-facing bid gateways across multiple Availability Zones
@@ -11 +11,34 @@ Implementation guidanceKey AWS servicesResources
-Resiliency evaluations should not be considered a one-time effort, but a continuous part of any workload's lifecycle. 
+RTB Fabric defaults to single-AZ gateway deployment. To survive an Availability Zone failure, you must deploy gateways in at least two Availability Zones, colocate with the Availability Zones where your DSP and SSP compute runs. Each subnet must have sufficient IPv4 capacity for gateway ENIs. 
+
+**Desired outcome:**
+
+  * RTB Fabric gateways are provisioned in at least two Availability Zones, each in a separate subnet with adequate available IPv4 address capacity. 
+
+  * Gateways are colocated with the compute fleet they serve, avoiding unnecessary cross-AZ latency. 
+
+  * An Availability Zone failure leaves bid connectivity intact through the gateways in the surviving Availability Zones, with no manual intervention and no sustained drop in bid win rate during or after the impairment. 
+
+
+
+
+**Common anti-patterns:**
+
+  * Deploying all gateways in a single Availability Zone, so an Availability Zone impairment removes all bid connectivity. 
+
+  * Provisioning a gateway in an Availability Zone that has no compute fleet. 
+
+  * Subnets with insufficient available IPv4 addresses, which causes gateway ENI provisioning to fail during scaling or failover. 
+
+
+
+
+**Benefits of establishing this best practice:**
+
+  * Bid connectivity survives a single-AZ impairment without manual intervention. 
+
+  * Co-locating gateways with compute keeps bid-path latency within budget while adding redundancy. 
+
+
+
+
+**Level of risk exposed if this best practice is not established:** High 
@@ -15 +48,5 @@ Resiliency evaluations should not be considered a one-time effort, but a continu
-Your workload, as well as the environment (both regulatory and partner) in which it operates, is constantly changing. Make resilience a regular part of your feature delivery and operational cadence throughout a workload's lifetime. Create a living document to track evolving processes, expectations, and improvements. Use AWS Gamedays, Well-Architected Framework Reviews, and Support Countdown engagements to improve reliability of advertising workloads. Coordinate with your various advertising partners and stakeholders to perform successful failover testing. 
+RTB Fabric provisions gateways in one Availability Zone unless you request more. That default assumes a single zone stays healthy for the life of the workload. An Availability Zone impairment breaks that assumption, and because every gateway terminates partner traffic, the loss is total. 
+
+A gateway in an Availability Zone without compute adds a cross-AZ hop to every bid and provides no resilience benefit. Colocate gateways with compute so each zone can both receive bid traffic and act on it locally. 
+
+Multi-AZ plans often fail on subnet capacity. Each gateway attaches network interfaces that draw IPv4 addresses from its subnet, and scaling or a failover into a surviving zone raises that draw exactly when a stall is unaffordable. A tight subnet looks fine at steady state and then refuses new interfaces under load. 
@@ -17 +54 @@ Your workload, as well as the environment (both regulatory and partner) in which
-## Key AWS services
+The service controls how many zones a single gateway can span, and that limit comes from a quota requested before provisioning. Your role determines which direction of connectivity each zone must permit, depending on whether you act as requester or responder. Request the quota first, before any gateway exists, and derive each zone's rules from your role. 
@@ -19 +56 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [AWS Well-Architected Tool](https://aws.amazon.com/well-architected-tool/)
+### Implementation steps
@@ -21 +58 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [Fault Tolerance Analyzer Tool](https://github.com/aws-samples/fault-tolerance-analyser) is an open-source tool that focuses specifically on identifying potential fault tolerance issues across different AWS services 
+  1. **Request the Multi-AZ gateway service quota increase:** Request the "Availability Zones per Gateway" service quota increase for Multi-AZ gateway deployment through the AWS Service Quotas console before provisioning. 
@@ -23 +60 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [AWS Gamedays](https://aws.amazon.com/gameday/)
+  2. **Provision gateways in at least two Availability Zones:** Provision gateways in ≥2 Availability Zones within your VPC, each in a separate subnet with adequate available IPv4 addresses. Confirm gateways are deployed in the same Availability Zones as your compute fleet. A gateway in an Availability Zone with no compute provides no resilience benefit and adds unnecessary cross-AZ latency. 
@@ -25 +62,3 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [Support Countdowns](https://aws.amazon.com/premiumsupport/aws-countdown/)
+  3. **Configure security groups per Availability Zone:** Configure security groups in each Availability Zone to allow HTTPS (TCP 443) inbound per your role (requester or responder). 
+
+  4. **Monitor subnet IPv4 capacity:** Create a scheduled Lambda function that periodically calls ec2:DescribeSubnets and publishes the AvailableIpAddressCount as a custom CloudWatch metric, enabling you to set alarms before IP exhaustion affects gateway scaling. 
@@ -32 +71,19 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [AWS Countdown](https://aws.amazon.com/premiumsupport/aws-countdown/)
+**Related best practices:**
+
+  * [ADVREL03-BP03 Deploy stateful dependencies across multiple Availability Zones](./advrel03-bp03.html)
+
+  * [ADVREL05-BP02 Monitor Availability Zone-level health and detect asymmetric failures](./advrel05-bp02.html)
+
+
+
+
+**Related documents:**
+
+  * [AWS RTB Fabric - Setting Up](https://docs.aws.amazon.com/rtb-fabric/latest/userguide/setting-up-aws-rtb-service.html)
+
+  * [AWS RTB Fabric - Service Quotas](https://docs.aws.amazon.com/rtb-fabric/latest/userguide/rtb-fabric-quotas.html)
+
+
+
+
+**Related services:**
@@ -34 +91 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [Build Your Own Game Day to Support Operational Resilience](https://aws.amazon.com/blogs/architecture/build-your-own-game-day-to-support-operational-resilience/)
+  * [AWS RTB Fabric](https://aws.amazon.com/rtb-fabric/)
@@ -36 +93 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [Best practices for handling EC2 Spot Instance interruptions](https://aws.amazon.com/blogs/compute/best-practices-for-handling-ec2-spot-instance-interruptions/index.html)
+  * [Amazon VPC](https://aws.amazon.com/vpc/)
@@ -38 +95 @@ Your workload, as well as the environment (both regulatory and partner) in which
-  * [Using the Fault Tolerance Analyzer Tool to Identify Potential Issues](https://aws.amazon.com/blogs/mt/using-the-fault-tolerance-analyser-tool-to-identify-potential-issues/index.html)
+  * [AWS Service Quotas](https://docs.aws.amazon.com/servicequotas/latest/userguide/intro.html)
@@ -49 +106 @@ To use the Amazon Web Services Documentation, Javascript must be enabled. Please
-Failure management 
+RTB Fabric resilience
@@ -51 +108 @@ Failure management
-ADVREL05-BP02 Create disaster recovery (DR) runbooks, and regularly test documented backup and restoration processes
+ADVREL05-BP02 Monitor Availability Zone-level health and detect asymmetric failures