At what point does a collection of microservices become a monolith that uses http instead of a bus 🤔
My favorite is seeing developers directly reach out to a DB of another microservice because “it’s just easier to pull the data from there”. One of the few times I’ve literally said “bruh”
Even more often the monolith bits are pulled out but continue to share the same monolithic database. The horizontal scanning is void, because while you can make any number of stateless service instances their operations performance is mostly databound and the (relational) database cannot horizontally scale.
Oh wow, yikes. Yeah, that's misguided.
A properly architected and implemented microservice architecture optimizes work throughput while minimizing risk. In practice its architecting in such a way that no part can take down the whole individually - the very opposite of a monolith where everything is inseparably interdependent at some level.
Problem is, most organizations don't know how to properly architect for and integrate microservice architectures into their environments and work process. Most think that a crew of former sysadmins can just spin up a few saas services, slap some autoscaling on it if they're feeling spicy, segment along traditional monolith "frontend/backend" lines for "security," and call it a day. They then spend time and money learning and/or fighting this system, only to see minimal (if any) improvement in work capacity/quality and instead end up with an outsized cloud bill.
Problem is, most organizations don’t know how to properly architect for and integrate microservice architectures into their environments and work process.
Which is exactly the same thing said about OOP, or structured programming, or Agile, or a number of other paradigms. Organizations aren't doing it right, but it works if it's "properly architected". Yet, this "proper architecture" doesn't seem to exist anywhere in the real world.
In fact, I've argued before that if OOP is fundamentally about objects ending messages to each other, then microarchitectures are just an approach to OOP where the communication channel is over network sockets.
If you understand things like encapsulation and message passing, then the difference doesn't come down to microarchitecture vs monolith. It's self-contained components vs not, and the method of passing messages is only important from a feasibility and performance standpoint.
The reality is, that hardly any projects actually need or benefit from micro services.
Most applications would scale just fine as a monolith, micro services seem to be rather an organizational tool to separate modules, because you can't come up with a proper architecture.
INSTANTLY
Developers tend to think in extremes.
This practice is bad sometimes? Avoid it at all costs!
This practice is good sometimes? Use it all the time without question!
Yeah, you need to have experiences with the intermediate architecture sizes and need enough information about the concrete use-case, to be able to deliberate something that's not an extreme. Both of those are a rarity with programming...
I do microclouds in the service.
BaaS (Bong as a Service).
Sounds really tight, dawg. You know what's up.
Neither will the trendy new language, distro, or standard
perl 5.10 for life!
Python, which I hate the syntax of with an actual passion, legit offers more library support though, so is this a lost cause at this point? Especially with junior devs, who to the extent they think they know how to program at all, do so in (well, JavaScript mainly iirc, but afa back-end specifically) Python?:-(
Sure they do. They just come with exciting, new problems
Sometimes a change is as good as a rest
'Nailing down the definition of a story point to "1 developer day" oughta do it '
Tbh mostly microservices solves two issues for me:
- You can scale resource-heavy operations without an impact of scaling everything else.
- You kind of ”force” people to keep things contained, instead of making everything interdependant. Which helps containing issues with updates to certain functionality to only have an impact on that functionality.
You could do #2 in a ”monolithic” setup as well (by doing modules or libraries or something), it just never works out that way in practice
Let's say there is a user lmicroservice. I'm on a UI team. I don't get to tell the user service team what, or when, to implement any features.
I'm tasked with making a page displaying all the users who have a birthday this month.
User API service can only search by user id, email, display name, or nickname.
Now instead of just querying the goddamn database, a one line fucking SQL statement, I have to deal with the user team, getting them to first off even admit that my use case is valid, convince them to work on the feature, coordinate with them to make sure the query works, sorts the data the way I need, etc, et. al, blah blah blah.
They already have the next 3 sprints full so I'm sitting on my ass for the next month before I can test.
Meanwhile they decide they're gonna implement a super generic thing, and so despite me working on code that we talked about using an interface we talked about, they implement something else so i have to throw out half my work anyway.
Then when I finally start using it I find, oh, it doesn't support a sort, only returns 100 results max with no pagination, so if there's 200 this month with a birthday fuck the 2nd hundred they don't show up because they're implementing bare minimum and the rest is slated for another sprint.
And it was then, your Honor, I grabbed the lead dev for the user microservice and tossed him off the 9th story of the building.
/sarcasm
Now instead of just querying the goddamn database, a one line fucking SQL statement, I have to deal with the user team
Exactly, you understand very well the purpose of microservices. You can submit a patch if you need that feature now.
Funnily enough I'm the technical lead of the team that handles the user service in an insurance company.
Due to direct access to our data without consulting us, we're getting legal issues as people were using addresses to guess where people lived instead of using our endpoints.
I guess some people really hate the validation that service layers have.
Neither will the YAML
DEFINITELY not YAML.
Yelling at my laptop
XML: Xtremely Masochistic Language
Neither will containers
The real win is encapsulation. Whether that's many modules running on the same server or many is not where the juice is
B-b-but the salesperson said that i-i-it...
