Scaling to millions of users is easy, just put some simple microservice behind and lb and add an off the shelf database. Google doesn't even use a real database, it's all flatfiles. The only reason scaling is hard is if you gave ridiculous overcomplicated code. When insta was acquired it had like 5 people. There is nothing magical about scaling it's just not making stupid mistakes.
"Google doesn't even use real databases" Lol okay. Spanner is an incredible feat of engineering.
It is the kind of thing that people read the press release for and say “Well that can’t be true, that’s not how databases work” but then.... it is
Spanner is not even real nosql
OP: Are you a new grad? Serious question
Google was built by new grads
Nothing against new grads. I’ve worked with some really brilliant new grads. The brilliant ones usually don’t think scaling is easy.
Try scaling an exchange keeping latencies at 4 9s below 150mics :)
NASDAQ is nowhere near that unless you are nextdoor
Everyone who cares is close
Spoken like someone who has never worked at scale.
OP works on crud apps for a small startup with stateless services. Either that or they're trolling
I built the elk stack for the world's largest retailer...
How do you scale avocado toast for all the engineering team ?
Use self served toilet avocados
Are you a dentist? https://en.m.wikipedia.org/wiki/Scaling_and_root_planing
Your definitely new... Scaling is hard. How do you make the sessions in your microservice resilient ?
Built in off the shelf frameworks like spring mvc and possibly caching. No website is run at anywhere near that quality, the locking would be horrendous and nothing complicated ever works. Having a user logged in 100 devices at same time is neither useful nor practical.
Clearly OP is noob noob
Oh my sweet summer child
Op you are totally right, you are the smart dude in the fucking planet, we should worship our new god, please write a post on medium or other platform.
Medium is just a view on to a database. Not sure why anyone cares about it! -op
Slashdot ran on like 3 boxes showing the same cached news story over and over again and that was sufficient to destroy the internet
Interested how you would design a CI CD system? Assume it to be used with Github repos
Simple Jenkins job that is super primitive and only calls a bash script. Gradlew run and just rsync a jar. Everything self contained, config injected at compile time via spring yaml
^ clearly you're not on an infra team. There's nothing scalable about Jenkins and hacked together bash scripts.