How would you deploy a taxi platform on a small budget?

November 16, 2026 4 answers
Asked by _a57~spoolladle.6

I'm building a ride-hailing backend with FastAPI, PostgreSQL, Redis, Kafka, and Docker. It's already split into several services, including ride matching and location tracking, with Nginx in front.

Before launch, would you put everything on one VPS with Docker Compose, split it across a couple of servers, or pay for managed services? I want to keep costs down without ignoring security and recovery.

Answers

2d~mosaicladle_571

A couple of small VPS instances can be a reasonable starting point. Run Nginx and the application containers on one, and put PostgreSQL on another with firewall rules allowing only the app host to reach it. Keep Redis and Kafka off the public internet too.

That separates the database from the public-facing host, but it doesn't give you high availability. Either machine can still fail, so budget for off-server backups, test a restore, and monitor disk space and service health. Add TLS and keep the servers patched.

9m+clampfolio_462:

Separate machines help with isolation, but check the networking and backup costs as well as the VPS prices. The cheapest setup on paper isn't always the cheapest one to operate.

Reply to this answer
b.761!parcelpiston

I'd take another look at the service boundaries before paying to host all of them. A modular application can keep identity, rides, payments, and matching separate in the code without making each one a separately deployed service.

Kafka is worth questioning too. If you don't yet need its messaging features, a simpler approach could remove a stateful system from the deployment. That doesn't mean microservices are always wrong. They can make sense when teams or scaling needs genuinely call for independent services.

_a57~spoolladle.6:

The services are already built separately, so I'd want to weigh the consolidation work against the operational savings.

Reply to this answer
6f~latchmural_843

Spend the comparison time on the stateful pieces, not just the app containers. Managed PostgreSQL may cost more than a small server, but included backups and less database administration could be worth it. Check the backup retention and try restoring before relying on the service.

For Kafka, compare managed pricing with the work of running it yourself, and consider whether a lighter messaging option meets the workload. Redis can remain a cache and lookup layer while PostgreSQL holds the authoritative records.

For deployments, use the same built image across environments, pass configuration separately, and run database migrations as a distinct step. That makes it easier to see what changed when a rollout goes wrong.

Reply to this answer
_k67+parcelgauge.3

Check whether you're eligible for cloud startup credits before settling on a provider. They could make managed database and networking options affordable while you prepare to launch, but compare the normal prices too. Credits expire, and you'll need a plan for the bill afterward.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions