What server architecture solutions do startups use?

1 points by echo_nevada1 8y ago ↗ HN

I am a primarily Python developer who works with fairly larger clients. We use AWS as our main infrastructure for load-balancing, networking, app servers, celery/redis servers, Postgres on RDS, etc

For full blown Test/Stage & Scalable Prod deployments for even our small/med projects this is costing between $1-3k a month.

I'm curious how startups handle these costs. You can spin up a Digital Ocean droplet for ~$25, Heroku for something similar. Do people do this and just cram Django/Celery/Redis/Postgres/Node (or their equivalents) all on the same server and go with it?

2 comments

[ 2.4 ms ] story [ 19.1 ms ] thread

marenkay 8y ago ↗

From personal experience: a large DO instance safely gets you through for the first paying customers to the point where you need more power. You can make your First million with a 200$ hosting bill.

The decisive point isn’t resources anyway, but reaching a size we’re compliance laws and quality standards force you to invest in large scale architecture

techdragon 8y ago ↗

I use the same stack most of the time in my professional work and I'm currently looking heavily at kubernetes to reduce the "size" of my smallest working deployment for complex systems.

Instead of rolling several AWS autoscale groups or elastic beanstalk deployments, I'll simply deploy one autoscaling kubernetes cluster using the well supported tools for doing so, and then leverage the best of what AWS provides such as RDS, EBS, and most importantly for future rapid prototyping EFS, adding EFS as a shared flat file system across multiple automatically scaling docker containers is extremely effective for testing out ideas without the complexity of using S3. The unix philosophy of "it's all files" but at massive scales.