RxJava 2: Parallel Multiple Network Calls Made Easy

Picture by Jonathan Chng on Unsplash

In my previous blog, I have shown that how we could connect multiple network call together easily using RxJava 2. But there’s a drawback with the previous approach, as they are still called sequentially.

Let see how we could make them parallel.

Connecting 2 networks call

The slower version

When we use mergeWith as below to combine two network call…

val disposable = firstNetworkCall()
    .mergeWith(secondNetworkCall())
    .subscribeOn(Schedulers.io())
    .observeOn(AndroidSchedulers.mainThread())
    .subscribe { it -> doSomethingWithIndividualResponse(it) }

The time required would be the total time needed by firstNetworkCall and secondNetworkCall as shown in diagram below.

However, since firstNetworkCall and secondNetworkCall are not related to each other, why can't we make them parallel?

The Parallel Version

To make it parallel, for each of the Observable, put a subscibeOn on it

So the code looks like below

val disposable = firstNetworkCall().subscribeOn(Schedulers.io())
    .mergeWith(secondNetworkCall().subscribeOn(Schedulers.io())
    // .subscribeOn(Schedulers.io()) // not needed anymore
    .observeOn(AndroidSchedulers.mainThread())
    .subscribe { it -> doSomethingWithIndividualResponse(it) }

However, this looks a little confusing. We could make it slightly clearer instead of using mergeWith, use Observable.merge

val disposable = Observable.merge(
        firstNetworkCall().subscribeOn(Schedulers.io()),
        secondNetworkCall().subscribeOn(Schedulers.io()))
    .observeOn(AndroidSchedulers.mainThread())
    .subscribe { it -> doSomethingWithIndividualResponse(it) }

The result will look something like below where the time is just the maximum of 2 two call (and perhaps with some slight overhead)