Coroutine Gotchas – Dispatchers | Weblog | bol.com

This takes roughly 2,7 seconds to complete, fairly a latency to gather some information. On this instance, each RestTemplate name blocks the thread it’s operating on till the decision returns some information after which palms over the thread to the subsequent RestTemplate name. It really works like this as a result of RestTemplate, by nature, is a synchronous, blocking operation. If you wish to understand how blocking and non-blocking operations work, there’s a detailed clarification right here.

After studying this, Tracey decides that she will make use of among the idle threads on her system to make a number of blocking calls on the identical time. It’s good to filter out a false impression right here: that is referred to as parallelism, not concurrency.

GOTCHA: Parallelism and concurrency are usually not the identical factor

Even when this isn’t a coroutine particular matter, there’s a widespread false impression relating to parallelism and concurrency. These phrases are sometimes used interchangeably, however they characterize distinct ideas.

Let’s take a quick journey into the realms of sequential blocking, concurrency, and parallelism by imagining a bustling workplace state of affairs, the place a diligent clerk is actively aiding shoppers.

Sequential/Blocking (No Concurrency or Parallelism):

In our imaginary workplace, this conduct can be seen because the clerk serving to one consumer at a time. Whereas the primary consumer fills within the kind, the clerk waits and would not serve every other consumer.

A single-threaded program works in a sequential/blocking method, the place duties are executed one after one other with none concurrent or parallel execution.

Parallelism:

This entails the clerk calling in a colleague to assist. Now, each shoppers are being served concurrently by totally different clerks. Every clerk works independently on their assigned job, resulting in true parallel execution.

In a multi-process or multi-threaded program, duties are genuinely executed concurrently on a number of processors or cores. For instance, two RestTemplate calls could be made on the identical time, however they’ll use and block totally different threads, not sharing their sources.