Summary
The Sequential Convoy pattern processes a set of related messages in a defined order without blocking the processing of other message groups. It addresses the challenge in distributed systems of needing ordering guarantees (FIFO) while still wanting to scale out to handle heavy load. Messages are classified into categories, and locking in the queuing system keeps each listener to one message at a time within a category. In Azure Service Bus this uses Message Sessions, and strict FIFO requirements limit throughput.
Key Points
- Purpose: Ensure that messages belonging to the same group (category) are processed in the order they arrived, while different groups can be processed in parallel.
- How it works:
- Classify incoming messages into "categories" (for example: by Order ID).
- Use locking in the queuing system so that each listener takes and processes only one message at a time within the same category.
- In Azure Service Bus, this is done through Message Sessions.
- Benefits:
- Maintains strict ordering for interdependent data (for example: create order -> payment -> delivery).
- Allows scale-out by processing many different "convoys" on different workers.
- Considerations:
- Throughput: Strict FIFO requirements limit scalability. Not suitable for extremely high-load scenarios (millions of messages per second).
- Network latency: Messages may arrive out of order because of the network. Consider using sequence numbers to check.
- Bottleneck: The initial queue processor (ledger processor) can become a bottleneck if it is not designed well.
A convoy is one category, often an order id, and a listener holds only one message from that category at a time. Azure Service Bus does this with message sessions, so other orders keep moving on other workers. Inside one order the chain stays FIFO, such as create, then payment, then delivery. Strict FIFO caps throughput and is a poor fit for millions of messages per second. The first ledger processor can bottleneck the rest. Network delay can reorder arrivals, so sequence numbers are the check.
When to Use
- When messages must be processed in the order they were sent.
- When messages can be categorized so the categories serve as the unit of scale-out for the system.
Relationships
- Competing Consumers Pattern: Sequential Convoy improves on this pattern by adding an ordering constraint for groups of related messages.
- Azure Service Bus Sessions: The main Azure tool for implementing this pattern.
References
See also: Competing Consumers Pattern · Priority Queue Pattern · Publisher Subscriber Pattern