The μ-services rant

The economy is powered by the bytes today and in byte economy, the cost of storage, network and compute power is rapidly declining. This trend has made quite a few feats viable in the recent years. Some of them aren’t completely new. “Big data”, “Fast data”, “IoT” and “Deep learning” are few to mention. Yet another trend byte economy created is: it rejuvenated the age-old idea of building smaller, loosely coupled, reusable piece of software that does one thing and one thing well for quicker time to market and cheaper cost of change and gave it a fancy name – Microservices (hereinafter μ-services).

In a technical sense, it’s a bunch of “frontrunners” in the tech world going – ‘A: Guys, compute, storage and especially network is dirt cheap today, why can’t we build and deploy functionalities as tiny, independent full stack software? ..and to build apps developers can choose the functionalities desired and compose network calls? B: hmmm.. so whats new? A: Nothing new, the same old concept just replace all in-memory function calls or shared library calls with remote network calls, but the key is in promoting it well to organizations and developer communities. The message is: they can now independently build, change, deploy and scale them with different teams who don’t have to know the existence of other teams. B: Awesome … C: Sure, but independent full stack software sounds too verbose, lets come up with a name (preferably a name that can generate some hype)? say… Microservices…….A: Sure but if you use the term ‘services’ in the name we need to follow that up with how it compares to SOA, how the μ-services interaction works: say choosing between Orchestration vs Choreography… C:  isn’t Orchestration vs Choreography is a simple Synchronous (req/res) vs Asynchronous (Event-driven) argument…..A:  Yes, but let’s complicate it, ESB is least we killed it…….. B: Sure that can be arranged, and don’t worry let me complicate the DevOps world with this whole μ-services thing, possibly combine this with containerization and so on….A: Great guys, here is the target: end of next year each of these concepts should have a Gartner quadrant of its own, cheers…………….  (and the rant continues)

(I am not saying this is how it happened, but it could very well be the case and finally no offense to anyone.)