Scaling: from 0 to 25 million users by Josip Stuhli
A story of how our infrastructure evolved over time to accommodate an increasing number of users - from on-premise to cloud and back down.
How does one make an infrastructure to handle more than a couple of users? How do you go from 100 to 1000 to 100,000 to tens of millions? What happens when due to popular demand hundreds of thousands of users hit your servers at the same time?
I'll tell you a story of how a small team of people managed to move software and services from one server to two, and then to dozens on cloud and then back to on-premise. What we encountered on the way, where we failed, and how we solved it.
Talk Questions
Questions moderation
All questions have to comply with our Code of conduct. So if you don't see your question right after sending it it's because either it has not been moderated yet or it's a question that does not comply with our CoC.
Question 744
You mentioned that you do 1 million queries/s,how big is your postgres cluster to support this?Question 755
So you somehow just built your own cloudQuestion 749
With all this caches, have you ever had to get a full cold start, with no caches filled in?Question 750
How does observability look at that scale ? Do you log/trace everything ?Question 747
You were using Varnish enterprise?Question 746
Which cache tool for frontend are right now using?Question 748
Which cache policy are you using?Question 751
If you have a single varnish cache node for a given subset of urls. How do you manage HA (and requests to backend if this cache dies)?Question 745
Do you use priority queues?Question 752
which php framework are you using, if any?Question 756
Do you were using Varnish enterprise to coordinate the cache? If not, how did you handle cache discrepancies?Question 757
Is varnish running in k8s?Question 753
How do you decide on what to purge from the cache, does it happen during feature creation?Question 758
How have you achieved multi DC arcitecture using postgress?