Hidden Secrets to Making a New Microservices Cluster a Success

Published on , 570 words, 3 minutes to read

Mara is hmm
<Mara>

So everyone is talking about microservices these days. I've been wondering what the best way to make a fully end-to-end continuously deployed multi-cloud federated global microservices cluster is. Do you have any advice? My product team needs this done quickly so we can meet an artificial deadline set by someone that doesn't understand the problems we face at all.

Cadey is coffee
<Cadey>

Don't.

Mara is happy
<Mara>

But based on all the artificial requirements this looks like it's the solution to all our needs! We can set up Anthos on GKE to combine our EKS clusters with our legacy onprem Kubernetes cluster so that we can have a fully hyperscaled coherent user experience across multiple continents!

Cadey is coffee
<Cadey>

Do you like hurting yourself? You are not Google. You do not need to be hyperscale. You can probably get away with running this as one big monolith on a server you rent from a datacenter. Two servers if you want redundancy, but that's most likely not relevant until you get above several tens of thousands of daily active users. Single servers can do more than you think. Most of what you pay for in a server is idle compute. Use it.

Mara is hmm
<Mara>

But we need to have the ability to horizontally scale to meet the latency requirements that marketing made up so that our time to first byte is minimized, we need to reap that sweet, sweet SEO optimization so that people can connect and live with eachother while having the highest possible quality experience so they come back soon.

Cadey is coffee
<Cadey>

If keeping customers is really your goal, make your product more addictive than porn, drugs and TikTok. Either that or make a cult. Those are great ways to abuse parts of the human baser instincts to make sure that you can maintain users over longer amounts of time.

Mara is hmm
<Mara>

But how do I make the cluster though?

Cadey is coffee
<Cadey>

Don't.

Mara is wat
<Mara>

But the SRE team I'm on can take care of the requisite setup.

Cadey is coffee
<Cadey>

Lol, no it can't. You will be managing more servers than man has any right to care about. You will be subjecting yourself to a torture that the Geneva Convention has not banned yet because the sheer philosophical depths of depravity have not been fully understood yet. You will become an expert in YAML and then find out the truth of how much it will take out of your mortal lifespan. You will be fighting people trying to use HTML templating languages to create YAML and you will be terrified to find out that this strategy works at all. You will burn out so hard that you will never want to touch a computer again.

Do you really want to do this to yourself? Do you really want to subject yourself to a level of misery that will define the rest of your life? Do you want to have to explain what the torture nexus of AWS is to therapists? Do you want to create this reality? Doing this will do all of those things and more.

MIGROSERBICES

[Remainder of content expunged by the order of O5-1]


Facts and circumstances may have changed since publication. Please contact me before jumping to conclusions if something seems wrong or unclear.

Tags: microservices, k8s, deployment, cicd, hashtag, swag, gke, eks, anthos, satire