Earlier quoted context omitted.
similarly , making half that and using dedicated servers. When you re a solo dev you dont really have time to risk learning unproven tech or tech that doesnt scale. And as a dev i like writing code that does new and interesting things, not learning tools and other people's APIs. The amount of services this guy has to manage is mind-boggling, and unfortunately i m simple and stupid. I guess i m a hermit dev but thankf…
I found the list under miscellaneous to be similarly mind-boggling. There are approximately 56 moving pieces to this "boring" one-person setup. All the things they mentioned: Ubuntu, PostgreSQL, Elasticsearch cluster, Redis, RabbitMQ, Django / Python3, uWSGI, Nginx, Celery, Celery Beat, Supervisord, React + Redux + Webpack + ES, Amazon S3, Cloudfront, react media player, Ansible, Datadog, PagerDuty, Rollbar, Slack, P…
FWIW, this component:
PostgreSQL, Elasticsearch cluster, Redis, RabbitMQ, Django / Python3, uWSGI, Nginx, Celery, Celery Beat, Supervisord
All of that is a very typical Django stack with React bolted onto the frontend. Whether he runs into trouble managing RabbitMQ, Redis, Postgres or any of the Python services is another story, but at least
if he did there are many, many others using the exact same stack, so he should be able to easily find answers to people experiencing the same problems. All of these technologies have several year track records now, too, so (hopefully) they are stable and reliable in production. My biggest concern would be managing the ES cluster.
On the other hand, I disagree with his points about docker being overly complex. Docker images are simple. Kubernetes, once you get the hang of it, is great on GKE and gives you automatic SSL via Google or let’s encrypt, and load balancing and auto-scaling just works. It’s probably more expensive than managing your own servers, at this level, but maybe not since you could pack more services into fewer compute instances.