How much does it cost to maintain a SaaS 24/7: API, PostgreSQL, Redis, and workers budget

Published on 2026-11-14 by frfrand

SaaS cost model running all month, with explicit assumptions, hidden costs, and comparison between PaaS and VPS.

How much does it cost to maintain a SaaS 24/7: API, PostgreSQL, Redis, and workers budget

The cost of publishing a website is not the cost of operating a SaaS. A product that needs API, PostgreSQL, Redis, workers, storage, email, backups, and monitoring can continue to generate expenses even if it barely receives visits.

1. Define a scenario and avoid misleading figures

This example is hypothetical, not a commercial offer: a Node.js or Django API with a permanently active process, PostgreSQL, Redis, a worker, 20 GB of data, 100 GB of monthly egress, and basic backup operations. We will calculate based on a 730-hour month. CPU peak, memory size, and real traffic can completely change the sizing.

ComponentMonthly assumptionWhat to evaluate
API730 h activeCPU, RAM, replicas
PostgreSQL730 h + 20 GBMemory, disk, backup, connections
Redis730 hRAM, persistence, operations
Worker730 hCPU, memory, queues
Network egress100 GBAllowances and price per GB
Support and securityVariable hoursPatches, alerts, incidents

2. Three hosting models

Consumption-based PaaS: you pay for services and resources according to the provider's rules; in return, you reduce part of the infrastructure administration. Review execution, network, and storage costs separately.

Instance-based managed services: you choose independent plans for API and databases. Spending is more predictable per instance, although adding components can increase the total.

Fixed-capacity VPS: you can share CPU and RAM between API, PostgreSQL, and Redis; the quota is usually stable until you need to expand resources. But you become responsible for updates, isolation, restoration, monitoring, and incident response. Do not confuse fixed quota with unlimited performance.

3. Formula that actually works for comparing

Total mensual = cómputo API + base + Redis + workers
              + almacenamiento + egress + backups
              + monitorización + soporte + impuestos aplicables
For variable costs, use hours × price/hour, GB × price/GB, and possible per-request charges. Simulate normal scenario, traffic peak, and growth. Include migration amortization and the hours spent maintaining the platform.

4. Capacity and availability

A single VPS can run all four components economically, but it also represents a shared failure domain. For a critical platform, consider a secondary instance, external backups, tested restoration, and possible database replication. In a PaaS, also verify what level of redundancy the purchased plan actually offers.

5. Choose according to your stage

For an MVP without operations staff, PaaS simplicity can compensate. With workers running permanently and stable consumption, compare the cost of a VPS. When the product requires high availability or elastic scaling, decide based on load tests and recovery objectives, not on the initial quota.

Explore ArduMaker vs Railway, ArduMaker vs Render and the available infrastructure services.

6. Reproducible numerical example (hypothetical rates)

For the budget to serve as a decision-making tool, apply fictitious comparison rates, not prices published by a provider. The example is a 730-hour month:

ResourceInvented assumptionIllustrative subtotal
API 24/7USD 0.02/h × 730 hUSD 14.60
PostgreSQLUSD 0.03/h × 730 hUSD 21.90
RedisUSD 0.01/h × 730 hUSD 7.30
WorkerUSD 0.02/h × 730 hUSD 14.60
Disk20 GB × USD 0.10/GBUSD 2.00
Egress100 GB × USD 0.08/GBUSD 8.00
Backups and monitoringIllustrative reserveUSD 10.00
Total illustrative PaaSUSD 78.40/month

Now imagine a shared VPS whose plan cost USD 30 monthly and other USD 10 for copies and monitoring. The infrastructure subtotal would be USD 40. If you assign 3 monthly hours of administration valued internally at USD 20/h, the full cost becomes USD 100/month. Under these hypotheses, the VPS is cheaper as a hosting bill, but more expensive if you include operations. The result reverses when you need more resources or the team already manages servers.

The amounts are only didactic variables. They are not a quote from ArduMaker, Railway or Render; replace them with the rates of the actual plans and measure capacity before calculating savings.

Formula to use in a spreadsheet

coste_total = cuota_fija + sumatoria(horas_recurso × tarifa) + transferencia + almacenamiento + backups + horas_operativas + impuestos

If high availability is required, add a second instance, external backup storage and recovery tests. Compare normal, peak and failure scenarios, not just the average day.

Frequently asked questions

### Can I use PostgreSQL and Redis on the same VPS? Yes, with limits, storage and appropriate backups, if capacity and risk allow.

Is it worth paying for a permanent worker with little traffic?

It depends on whether it needs to react immediately and on the billing model.

Where is the most common hidden cost?

In not counting backups, outbound traffic, support, replicas and operational time.

Note: no provider figure is presented as a current price; verify plans and terms before contracting.

  • SaaS
  • costes
  • hosting
  • PostgreSQL
  • Redis

On ArduMaker

  • PostgreSQL: Install PostgreSQL and manage databases, users and permissions without the command line.
  • Redis: Install Redis, tune memory and persistence, and search, edit and delete keys from the panel.

Keep reading

All articles · VPS plans