Message queue

转载：https://en.wikipedia.org/wiki/Message_queue

In computer science, message queues and mailboxes are software-engineering componentsused for inter-process communication
(IPC), or for inter-thread communication within the same process. They use a queue for messaging – the passing of control or of content. Group communication systems provide similar kinds of functionality.   

The message queue paradigm is a sibling of the publisher/subscriber pattern, and is typically one part of a larger message-oriented middleware system.Most
messaging systems support both the publisher/subscriber and message queue models in their API, e.g. Java Message Service (JMS).

Overview

Message queues provide an
asynchronous communications protocol, meaning that the sender and receiver of the message
do not need to interact with the message queue at the same time.
Messages placed onto the queue are stored until the recipient retrieves them. Message queues have implicit or explicit limits on the size of data that may be transmitted in a single message and the number of messages that may remain outstanding on
the queue.

Many implementations of message queues function internally: within an operating system or within an application. Such queues exist for the purposes of that system only.[1][2][3]
Other implementations allow the passing of messages between different computer systems, potentially connecting multiple applications and multiple operating systems.[4] These message queueing systems typically provide enhanced
resilience functionality to ensure that messages do not get "lost" in the event of a system failure. Examples of commercial implementations of this kind of
message queueing software (also known as message-oriented middleware) include IBM WebSphere MQ (formerly MQ Series) and Oracle Advanced Queuing (AQ). There is a Java standard called Java Message Service, which has several proprietary
and free software implementations.
Implementations exist as proprietary software, provided as a service, open source software, or a hardware-based solution.
Proprietary options have the longest history, and include products from the inception of message queuing, such as IBM WebSphere MQ (formerly MQ Series), and those tied to specific operating systems, such as Microsoft Message
Queuing. There are also cloud-based message queuing service options, such as Amazon Simple Queue Service (SQS), StormMQ and IronMQ.
There are a number of open source choices of messaging middleware systems, including
Apache ActiveMQ, Apache Kafka, Apache Qpid,[5] Beanstalkd, HTTPSQS[6]JBoss Messaging, JORAM, RabbitMQ, Sun Open Message Queue, and Tarantool.
In addition to open source systems, hardware-based messaging middleware exists with vendors like Solace Systems, Sonoa / Apigee and Tervela offering queuing through silicon or silicon/software datapaths.
Most real-time operating systems (RTOSes), such as VxWorks and QNX,
encourage the use of message queueing as the primary inter-process or inter-thread communication mechanism. The resulting tight integration between message passing and CPU scheduling is attributed as a main reason for the usability of RTOSes
for real time applications. Early examples of commercial RTOSes that encouraged a message-queue basis to inter-thread communication also include VRTX and pSOS+, both of which date to the early 1980s. The Erlang programming language uses processes to provide
concurrency; these processes communicate asynchronously using message queuing.

Usage

In a typical message-queueing implementation,
a system administrator installs and configures message-queueing software (a queue manager or broker), and defines a named message queue. Or they register
with a message queuing service.

An application then registers a software routine that "listens" for messages placed onto the queue.

Second and subsequent applications may connect to the queue and transfer a message onto it.

The queue-manager software stores the messages
until a receiving application connects and then calls the registered
software routine. The receiving application then processes the message in an appropriate manner.

There are often numerous options as to the exact semantics of message passing, including:

Durability - messages may be kept in memory, written to disk, or even committed to a DBMS if the need for reliability indicates a more resource-intensive solution.
Security policies - which applications should have access to these messages?
Message purging policies - queues or messages may have a "time to live"
Message filtering - some systems support filtering data so that a subscriber may only see messages matching
so-me pre-specified criteria of interest
Delivery policies - do we need to guarantee that a message is delivered at least once, or no more than once?
Routing policies - in a system with many queue servers, what servers should receive a message or a queue's messages?
Batching policies - should messages be delivered immediately? Or should the system wait a bit and try to deliver many messages at once?
Queuing criteria - when should a message be considered "enqueued"? When one queue has it? Or when it has been forwarded to at least one remote queue? Or to all queues?
Receipt notification - A publisher may need to know when some or all subscribers have received a message.
These are all considerations that can have substantial effects on transaction semantics, system reliability, and system efficiency.

Standards and protocols

Historically, message queuing has used proprietary, closed protocols, restricting the ability for different operating systems or programming languages to interact in a
heterogeneous set of environments.
An early attempt to make message queuing more ubiquitous was Sun Microsystems' JMS specification, which provided a Java-only abstraction of a client API.
This allowed Java developers to switch between providers of message queuing in a fashion similar to that of developers using SQL databases. In practice, given the diversity of message queuing techniques and scenarios,
this wasn't always as practical as it could be.

Three standards have emerged which are used in open source message queue implementations:

Advanced Message Queuing Protocol (AMQP) – feature-rich message queue protocol
Streaming Text Oriented Messaging Protocol (STOMP) – simple, text-oriented message protocol
MQTT (formerly MQ Telemetry Transport) - lightweight message queue protocol especially for embedded devices

These protocols are at different stages of standardization and adoption. The first two operate at the same level as HTTP, MQTT at the level of TCP/IP.

Some proprietary implementations also use HTTP to provide message queuing by some implementations, such as Amazon's SQS. This is because it is always possible to
layer asynchronous behaviour (which is what is required for message queuing) over a synchronous protocol using request-response semantics. However, such implementations are constrained by the underlying protocol
in this case and may not be able to offer the full fidelity or set of options required in message passing above.

Synchronous vs. asynchronous

Many of the more widely known communications protocols in use operate
synchronously. The HTTP protocol – used in the World Wide Web and in web services – offers an obvious example where a user sends a request for a web page and then waits for a reply.
However, scenarios exist in which synchronous behaviour is not appropriate. For example, AJAX (Asynchronous JavaScript and XML) can be used to asynchronously send text or XML messages to update part of a
web page with more relevant information. Google uses this approach for their Google Suggest, a search feature which sends the user's partially typed queries to Google's servers and returns a list of possible full queries the user might be in the process of
typing. This list is asynchronously updated as the user types.
Other asynchronous examples exist in event notification systems and
publish/subscribe systems.

An application may need to notify another that an event has occurred, but does not need to wait for a response.
In publish/subscribe systems, an application "publishes" information for any number of clients to read.
In both of the above examples it would not make sense for the sender of the information to have to wait if, for example, one of the recipients had crashed.
Applications need not be exclusively synchronous or asynchronous.
An interactive application may need to respond to certain parts of a request immediately (such as telling a customer that a sales request has been accepted, and handling the promise to draw on inventory), but may queue other parts (such as completing
calculation of billing, forwarding data to the central accounting system, and calling on all sorts of other services) to be done some time later.

In all these sorts of situations, having a subsystem which performs message-queuing (or alternatively, a broadcast messaging system) can help improve the behavior of the overall system.

Implementation in UNIX

UNIX implements message passing by keeping an array of linked lists as message queues. Each message queue is identified by its index in the array, and has a unique
descriptor. A given index can have multiple possible descriptors. UNIX gives standard functions to access the message passing feature.[7]

msgget(): This system call takes a key as an argument and returns a descriptor of the queue with the matching key if it exists. If it does not exist, and the IPC_CREAT flag is set, it makes a new message queue with the given key and returns its descriptor.
msgrcv(): Used to receive a message from a given queue descriptor. The caller process must have read permissions for the queue. It is of two types.[8]
Blocking receive puts the child to sleep if it cannot find a requested message type. It sleeps till another message is posted in the queue, and then wakes up to check again.
Non-blocking receive returns immediately to the caller, mentioning that it failed.
msgctl(): Used to change message queue parameters like the owner. Most importantly, it is used to delete the message queue by passing the IPC_RMID flag. A message queue can be deleted only by its creator, owner, or the superuser.