According to software engineer Constantine Aaron Cois, the web is changing. “The web used to be about consumption: Viewing web pages, watching videos, looking at pictures of cats. Of course, its still about pictures of cats, but the web has become more about interaction,” Cois stated. “Users around the world want to interact with each other, and they want to do it in real time. Chat, gaming, constant social media updates, collaboration — each of these features requires real time communication between users, clients, and servers across the web.”
“What’s more,” says Cois, “this real-time communication needs to happen at massive scale, supporting hundreds, thousands, even millions of users.”
Node.js uses an event-driven, non-blocking I/O model that makes it lightweight and efficient and perfectly suited for building data-intensive real-time applications running across distributed devices.
Node.js uses an event-driven, non-blocking I/O model that makes it lightweight and efficient and perfectly suited for building data-intensive real-time applications running across distributed devices. The Node development team offers a simple “hello world” web server example to show that many client connections can be handled concurrently, since each connection is only a small heap allocation.
“Node tells the operating system (through epoll, kqueue, /dev/poll, or select) that it should be notified when a new connection is made, and then it goes to sleep,” a Node rep explains. “If someone new connects, then it executes the callback.”
Contrast this to the more common concurrency model that employs OS thread-based networking, which is often relatively inefficient and very difficult to use, being plagued by process blocks. With Node, nothing blocks, so less-than-expert programmers are able to develop fast and reliable systems.
“Node will show much better memory efficiency under high-loads than systems which allocate 2mb thread stacks for each connection,” the rep clarified. “Furthermore, users of Node are free from worries of dead-locking the process — there are no locks. Almost no function in Node directly performs I/O, so the process never blocks.”
According to its developers, Node is influenced by and similar in design to systems such as Ruby’s Event Machine or Python’s Twisted, but takes the event model further by presenting the event loop as a language construct instead of as a library.
For example, while other systems may issue a blocking call to start the eventloop, defining behavior through callbacks at the beginning of a script and at the end to start a server through a blocking call such as EventMachine::run(), there is no such start-the-event-loop call in Node.
Other nods to the web can also be found in Node, which treats HTTP as a first class protocol, with a library that has grown out of experiences in developing and working with web servers. For example, while streaming data through web frameworks is often impossible, Node’s HTTP parser and API attempt to correct these problems, and when coupled with Node’s event driven infrastructure, it makes a solid foundation for libraries and web frameworks.
For developers concerned about multiple-processor concurrency and using threads to scale programs to multi-core computers, with Node, new processes can be started via the child_process.fork(), with these processes being scheduled in parallel. For load balancing incoming connections across multiple processes, the cluster module provides a solution.