Service Bus queues, topics, and subscriptions
Microsoft Azure Service Bus supports a set of cloud-based, message-oriented middleware technologies including reliable message queuing and durable publish/subscribe messaging. These "brokered" messaging capabilities can be thought of as decoupled messaging features that support publish-subscribe, temporal decoupling, and load balancing scenarios using the Service Bus messaging workload. Decoupled communication has many advantages; for example, clients and servers can connect as needed and perform their operations in an asynchronous fashion.
The messaging entities that form the core of the messaging capabilities in Service Bus are queues, topics and subscriptions, and rules/actions.
Queues offer First In, First Out (FIFO) message delivery to one or more competing consumers. That is, receivers typically receive and process messages in the order in which they were added to the queue, and only one message consumer receives and processes each message. A key benefit of using queues is to achieve "temporal decoupling" of application components. In other words, the producers (senders) and consumers (receivers) do not have to be sending and receiving messages at the same time, because messages are stored durably in the queue. Furthermore, the producer does not have to wait for a reply from the consumer in order to continue to process and send messages.
A related benefit is "load leveling," which enables producers and consumers to send and receive messages at different rates. In many applications, the system load varies over time; however, the processing time required for each unit of work is typically constant. Intermediating message producers and consumers with a queue means that the consuming application only has to be provisioned to be able to handle average load instead of peak load. The depth of the queue grows and contracts as the incoming load varies. This capability directly saves money with regard to the amount of infrastructure required to service the application load. As the load increases, more worker processes can be added to read from the queue. Each message is processed by only one of the worker processes. Furthermore, this pull-based load balancing allows for optimum use of the worker computers even if the worker computers differ with regard to processing power, as they pull messages at their own maximum rate. This pattern is often termed the "competing consumer" pattern.
Using queues to intermediate between message producers and consumers provides an inherent loose coupling between the components. Because producers and consumers are not aware of each other, a consumer can be upgraded without having any effect on the producer.
To quickly learn how to create a queue, then send and receive messages to and from the queue, see the quickstarts for each method. For a more in-depth tutorial on how to use queues, see Get started with Service Bus queues.
For a working sample, see the BasicSendReceiveUsingQueueClient sample on GitHub.
You can specify two different modes in which Service Bus receives messages: ReceiveAndDelete or PeekLock. In the ReceiveAndDelete mode, the receive operation is single-shot; that is, when Service Bus receives the request, it marks the message as being consumed and returns it to the application. ReceiveAndDelete mode is the simplest model and works best for scenarios in which the application can tolerate not processing a message if a failure occurs. To understand this scenario, consider a scenario in which the consumer issues the receive request and then crashes before processing it. Because Service Bus marks the message as being consumed, when the application restarts and begins consuming messages again, it will have missed the message that was consumed prior to the crash.
In PeekLock mode, the receive operation becomes two-stage, which makes it possible to support applications that cannot tolerate missing messages. When Service Bus receives the request, it finds the next message to be consumed, locks it to prevent other consumers from receiving it, and then returns it to the application. After the application finishes processing the message (or stores it reliably for future processing), it completes the second stage of the receive process by calling CompleteAsync on the received message. When Service Bus sees the CompleteAsync call, it marks the message as being consumed.
If the application is unable to process the message for some reason, it can call the AbandonAsync method on the received message (instead of CompleteAsync). This method enables Service Bus to unlock the message and make it available to be received again, either by the same consumer or by another competing consumer. Secondly, there is a timeout associated with the lock and if the application fails to process the message before the lock timeout expires (for example, if the application crashes), then Service Bus unlocks the message and makes it available to be received again (essentially performing an AbandonAsync operation by default).
In the event that the application crashes after processing the message, but before the CompleteAsync request is issued, the message is redelivered to the application when it restarts. This process is often called At Least Once processing; that is, each message is processed at least once. However, in certain situations the same message may be redelivered. If the scenario cannot tolerate duplicate processing, then additional logic is required in the application to detect duplicates, which can be achieved based upon the MessageId property of the message, which remains constant across delivery attempts. This feature is known as Exactly Once processing.
Topics and subscriptions
In contrast to queues, in which each message is processed by a single consumer, topics and subscriptions provide a one-to-many form of communication, in a publish/subscribe pattern. Useful for scaling to large numbers of recipients, each published message is made available to each subscription registered with the topic. Messages are sent to a topic and delivered to one or more associated subscriptions, depending on filter rules that can be set on a per-subscription basis. The subscriptions can use additional filters to restrict the messages that they want to receive. Messages are sent to a topic in the same way they are sent to a queue, but messages are not received from the topic directly. Instead, they are received from subscriptions. A topic subscription resembles a virtual queue that receives copies of the messages that are sent to the topic. Messages are received from a subscription identically to the way they are received from a queue.
By way of comparison, the message-sending functionality of a queue maps directly to a topic and its message-receiving functionality maps to a subscription. Among other things, this feature means that subscriptions support the same patterns described earlier in this section with regard to queues: competing consumer, temporal decoupling, load leveling, and load balancing.
Create topics and subscriptions
Creating a topic is similar to creating a queue, as described in the previous section. You then send messages using the TopicClient class. To receive messages, you create one or more subscriptions to the topic. Similar to queues, messages are received from a subscription using a SubscriptionClient object instead of a QueueClient object. Create the subscription client, passing the name of the topic, the name of the subscription, and (optionally) the receive mode as parameters.
For a full working example, see the BasicSendReceiveUsingTopicSubscriptionClient sample on Github.
Rules and actions
In many scenarios, messages that have specific characteristics must be processed in different ways. To enable this processing, you can configure subscriptions to find messages that have desired properties and then perform certain modifications to those properties. While Service Bus subscriptions see all messages sent to the topic, you can only copy a subset of those messages to the virtual subscription queue. This filtering is accomplished using subscription filters. Such modifications are called filter actions. When a subscription is created, you can supply a filter expression that operates on the properties of the message, both the system properties (for example, Label) and custom application properties (for example, StoreName.) The SQL filter expression is optional in this case; without a SQL filter expression, any filter action defined on a subscription will be performed on all the messages for that subscription.
For a full working example, see the TopicSubscriptionWithRuleOperationsSample sample on GitHub.
For more information and examples of using Service Bus messaging, see the following advanced topics: