This is similar to the pipelining performed by most modern microprocessors: multiple instructions are pipelined through different execution units of the processor to achieve the best performance. – link
Facebook does a great job outlining their use of BigPipe and I even found the opensource project on github – check it out here. Check out some of their performance results using a traditional approach versus pipelining the content:
As most know, Facebook is mostly written in PHP. The open source project on gitHub has sample implementations also in PHP so it is very easy to follow.