I remember reading that one of the benefits of microservices architecture is that you expose your DBs and logic through APIs so each team has its own technology choices...
But a certain level of standards is always beneficial, otherwise, you have knowledge silos inside the company because one team wouldn't want to even look at another's team service because it has completely different technology choices.
And looking at Pinterest at the time of 3 engineers, that seems very dangerous. If one engineer implements something in a technology that the others are not familiar with, they wouldn't jump into working on it.
Great post Leo, thanks for finding and simplifying good content :)
Thanks! At the time, ElasticSearch was still newer so it was seen as less mature/reliable. Tbh, off the top of my head, I don’t remember exactly why they switched from CloudFront either but it may have to do with pricing or redundancy.
Very interesting read! I'm curious about that latest jump (11M to 22M users, 11 to 40 engineers). I'm guessing it had more to do with developing features, that maintaining the infrastructure?
I mean in the 1-6 engineers days, they probably worked just to keep everythng running, without any innovation.
I believe so! As they found product market fit (as shown by their user growth), they started hiring much more aggressively in order to build to ur the product faster.
Initially, Pinterest was quite simple, allowing them to move forward with a small team.
> Limited, proven choices was a good thing
I remember reading that one of the benefits of microservices architecture is that you expose your DBs and logic through APIs so each team has its own technology choices...
But a certain level of standards is always beneficial, otherwise, you have knowledge silos inside the company because one team wouldn't want to even look at another's team service because it has completely different technology choices.
And looking at Pinterest at the time of 3 engineers, that seems very dangerous. If one engineer implements something in a technology that the others are not familiar with, they wouldn't jump into working on it.
Great post Leo, thanks for finding and simplifying good content :)
Very simple to understand, excellently written.
I am curious in knowing why Cloudfront was replaced with akamai, and elastic search with Apache Solr.
Thanks! At the time, ElasticSearch was still newer so it was seen as less mature/reliable. Tbh, off the top of my head, I don’t remember exactly why they switched from CloudFront either but it may have to do with pricing or redundancy.
Very interesting read! I'm curious about that latest jump (11M to 22M users, 11 to 40 engineers). I'm guessing it had more to do with developing features, that maintaining the infrastructure?
I mean in the 1-6 engineers days, they probably worked just to keep everythng running, without any innovation.
I believe so! As they found product market fit (as shown by their user growth), they started hiring much more aggressively in order to build to ur the product faster.
Initially, Pinterest was quite simple, allowing them to move forward with a small team.
Great article! With the embrace of denormalization, how did the team handle data consistency across tables/rows?