Member-only story
Why Reactive Streams are the SECRET to CUTTING your monthly cloud bill!
In the world of software development, we are FOREVER battling two competing forces: bigger demands vs. more powerful machines.
Now this may SOUND like NOT an actual competition. In fact, it may sound like perfect alignment, because wouldn’t the more powerful machines meet the bigger demands?
Amazingly…no.
Because what REALLY happens is a little like this:
- Demands for more users and more data come in.
- We get bigger/more servers.
- We catch up to the demands of step (1).
- In the time span of steps (2) and (3), the demand increases another 25%, which simply repeats this vicious loop.
In fact, this is a KEY reason cloud computing took off like crazy. In previous decades, the very thought of growing your own data center was expensive, time consuming, and laborious. There was legit pushback when someone would say “we need more servers!”
But call up your favorite cloud vendor to say “we need 25% more than last month”, and instead of dread you’ll hear excitement!
Of course your cloud bill is going up 25% as well. But the capacity you need is readily available!
That’s great…if 25% more server power yields 25% or more sales. But what if it doesn’t? What if your cost-to-profit ratio isn’t like that? A growing monthly cloud bill doesn’t always translate to enough profit. And thus, you need to start looking for ways to CUT COSTS.
What if I told you…that you’re already wasting precious time. (And in the cloud, time is literally money.)
You have idle threads sitting around, waiting for a response. And when you scale to any appreciable size, the idle cost will kill you.
10,000 server processes, each with 100 threads sporting some appreciable amount of lingering idle time…well let’s just say it doesn’t have to BE that way!
Reactive Streams beats this crazy problem of over-consuming an underutilized resource by introducing: non-blocking.
The idea is simple: instead of holding up a thread waiting for a response, instead let that thread go do some…
