Reliability & Delivery · Fault Handling

Bulkhead

Partition resources and workloads into isolated pools so one overloaded tenant, connector or dependency cannot consume all available capacity or spread failure across the entire solution.

isolationcapacityfault-containment
The problem
How can a failure or capacity shortage in one part of a system be prevented from exhausting every other part?
GateSift summary based on Microsoft Azure Architecture Center guidance. The wording, visualization, Azure mapping and analyzer context are original GateSift material.
Original GateSift visualization

How Bulkhead works

Workloads use isolated capacity pools so exhaustion or failure in one pool does not consume every resource.

Incoming workloads
Resource isolation
Critical pool
Independent pool
1

Identify workloads, tenants or dependencies that should not share the same failure boundary.

2

Allocate separate queues, workers, connection pools or runtime instances.

3

Monitor each pool independently and preserve capacity for critical flows when another pool is saturated.

GateSift explanation

What this pattern helps you decide

Partition resources and workloads into isolated pools so one overloaded tenant, connector or dependency cannot consume all available capacity or spread failure across the entire solution.

Which failure mode or capacity limit is this pattern intended to contain?
Which thresholds, deadlines, state or compensation data must remain durable?
How will operators know the pattern activated and whether recovery succeeded?
Common Azure implementations

Where you may see it

  • Separate Service Bus queues and worker pools per workload class
  • Dedicated Function Apps or Container Apps for critical flows
  • Per-backend connection pools and concurrency limits
GateSift relevance

How the analyzers can surface it

  • Many unrelated flows sharing one endpoint or worker boundary
  • No separation between critical and best-effort processing
  • Single BizTalk host or handler carrying unrelated adapter workloads

Pattern detection is contextual. GateSift should present these as architectural signals, not claim a pattern is implemented solely because one policy statement or adapter exists.

Source guidance and attribution

This page links to Microsoft Azure Architecture Center for the source architecture guidance. GateSift summaries, diagrams, Azure examples and analyzer signals are original. No endorsement by Microsoft is implied.

Back to pattern library