アーキテクチャ・意思決定比較表
Heroku と Dokku のデータ構造、運用コスト、ライセンスリスクの違いを詳細に分析します。
For engineering leads and financial planners, evaluating developer velocity against infrastructure overhead is a critical balance when managing cloud budgets. While Heroku popularized the frictionless, “git push” deployment workflow, its structured pricing tier updates and lack of a free tier make it a prime target for cost optimization via open-source alternatives like Dokku.
Heroku Official Pricing Plans (2026)
Heroku operates strictly on a paid model, charging per dyno (container) per month. Below is the active pricing structure:
| Plan | Pricing | Highlights | Best For |
|---|---|---|---|
| Eco | $5 / dyno / month | 1,000 shared dyno hours/month; sleeps after 30 minutes of inactivity | Personal projects, basic experimentation, and learning |
| Basic | $7 / dyno / month | No sleeping (always on); custom domains; automated SSL certificate management | Simple web apps, small hobby projects, and prototype APIs |
| Standard (Production) | $25 / dyno / month | 512 MB RAM (Standard 1X); horizontal scaling; preboot for zero-downtime deploys; metrics | Production-grade web applications, microservices, and APIs |
The Hidden Costs of Heroku
Relying solely on base dyno costs to calculate a budget will lead to significant forecasting errors. Heroku’s ecosystem charges heavily for supporting infrastructure:
- Database and Cache Add-ons: A functional production app requires data storage. Heroku Postgres starts at $5/month for minimal configurations and scales up rapidly into hundreds of dollars for production-tier storage. Heroku Data for Redis starts at $3/month and climbs quickly.
- Third-Party Add-On Marketplace: Essential services for monitoring, logging (e.g., Papertrail), and email delivery are billed separately. While convenient, their pricing curves are much steeper than buying direct or self-hosting.
- Team and Collaboration Overhead: Managing enterprise-grade permissions and compliance requires Heroku Team or Enterprise tiers, which introduce per-seat costs and minimum commitments not present in the self-hosted ecosystem.
- API and Resource Limits: Standard dynos are limited to 512 MB of RAM. Exceeding this limit forces an upgrade to Performance-tier dynos, which quickly scale to $250+/month per dyno, severely punishing memory-intensive modern frameworks.
Total Cost of Ownership (TCO) Analysis: Dokku (The Heroku Free Alternative)
Dokku is an MIT-licensed, Docker-powered, self-hosted PaaS. It allows teams to deploy applications to their own virtual private servers (VPS) using the exact same Git-based workflow as Heroku. While the software itself is entirely free, self-hosting introduces infrastructure and maintenance overhead.
Cost Trade-offs at a Glance
1. Hosting & Server Resource Estimation
Unlike Heroku’s strict limits, a single modern VPS can host dozens of Dokku applications and database instances simultaneously:
- Small Teams: A single $10–$20/month VPS (e.g., DigitalOcean, Hetzner, AWS EC2) with 2–4 GB RAM can comfortably host 5–10 small staging and production apps alongside lightweight local databases.
- Medium Teams: A $40–$80/month VPS (8–16 GB RAM) can run a robust development pipeline, multiple production apps, and caching layers with room to spare.
- Large Teams: A high-availability setup utilizing multiple Dokku servers, object storage, and automated backups ranges from $150–$400/month in raw cloud infrastructure.
2. Maintenance & Engineering Support Estimation
The real “cost” of Dokku is engineering labor. Organizations must budget for setup, OS patching, backup verification, and occasional troubleshooting:
- Small Teams: ~1–2 hours of developer time per month ($100–$200 internal cost).
- Medium Teams: ~4–6 hours of DevOps/SysAdmin time per month ($400–$600 internal cost).
- Large Teams: ~8–12 hours of specialized infrastructure engineering per month ($1,000–$1,500 internal cost).
Comparative TCO Table (Monthly SaaS Fees vs. Self-Hosted Infrastructure)
| Team Profile | Heroku Estimated Cost (Dynos + DBs + Add-ons) | Dokku Estimated Cost (VPS + Engineering Time) | Monthly Savings with Dokku |
|---|---|---|---|
| Small (5 Users, ~10 apps) | $250 - $400 | $120 ($20 VM + $100 maintenance) | $130 - $280 |
| Medium (20 Users, ~30 apps) | $1,200 - $2,000 | $580 ($80 VM + $500 maintenance) | $620 - $1,420 |
| Large (100 Users, ~80 apps) | $6,000 - $12,000 | $1,800 ($300 VMs + $1,500 maintenance) | $4,200 - $10,200 |
Scenario-Based Cost Analysis
Scenario A: The 5-User Startup
- Heroku: Deploying 5 active applications and 3 staging environments using Standard Dynos ($25 each) and Basic Postgres/Redis databases brings the monthly bill to roughly $280/month.
- Dokku: Running all 8 apps on a single $24/month VPS with local Dockerized Postgres and Redis instances. Adding 1 hour of maintenance overhead ($100/hour), the total cost is $124/month.
- Verdict: Dokku cuts costs in half while providing 4x the RAM and CPU headroom compared to Heroku’s 512MB limits.
Scenario B: The 20-User Engineering Team
- Heroku: Scaling to 25 microservices, QA pipelines, and high-availability production databases. At this scale, Heroku costs escalate to $1,800/month due to database scaling tiers and memory upgrades.
- Dokku: The team provisions two $80/month nodes (one for staging/CI, one for production) and schedules automated backups to S3. Factoring in 5 hours of monthly DevOps maintenance ($500), the total cost is $680/month.
- Verdict: Dokku saves over $1,100/month ($13,200+ annually), freeing up budget for direct business-building assets.
Scenario C: The 100-User Enterprise Division
- Heroku: Operating 80+ web processes and background workers requires entering Heroku Enterprise negotiations. Costs typically start at $8,000+/month when accounting for high-volume add-ons and support agreements.
- Dokku: The organization runs a clustered group of dedicated bare-metal or cloud servers costing $500/month. A semi-dedicated engineer manages updates and monitoring, consuming roughly 15 hours a month ($1,500). Total cost: $2,000/month.
- Verdict: At scale, self-hosting on Dokku yields massive savings (~$72,000+/year), though the organization must weigh this against the operational risks of managing its own servers.
When Does Paying for Heroku Actually Save Money?
Despite Dokku’s financial advantages, paying the premium for Heroku is the most cost-effective path under specific conditions:
- Zero DevOps Expertise: If your engineering team consists entirely of front-end or product-focused developers with no Linux command-line or Docker experience, the overhead of managing a server will result in costly downtime or security vulnerabilities.
- Strict Compliance Requirements: If your organization requires immediate HIPAA, PCI-DSS, or SOC 2 compliance out of the box, Heroku’s Shield Private Spaces provide instant compliance paths that would cost tens of thousands of dollars to audit and configure on self-hosted VMs.
- Critical Focus on Time-to-Market: When launching a brand-new MVP where every hour spent on infrastructure administration is an hour lost on product development, Heroku’s zero-maintenance model maximizes speed.
Final Purchasing Recommendation
- Choose Heroku if: You are a pre-revenue startup, have a small product-only team with no infrastructure skills, or require instant compliance certifications where engineering hours must be 100% focused on software development.
- Choose Dokku if: You are a growing mid-sized team looking for a high-performance heroku free alternative, want to escape arbitrary RAM/CPU limits, already run Docker in your workflow, or want to reduce staging/QA infrastructure costs to near zero. Setting up Dokku once provides a highly scalable platform that keeps cloud budgets flat even as deployment volume scales.
Cost and pricing analysis verified as of 2026-06-25. Self-hosting costs are estimates based on standard cloud providers.
よくある質問
How much does a basic always-on application with database storage cost to run on Heroku compared to Dokku?
A basic, always-on Heroku dyno costs $7/month, but you must also pay hidden costs like Heroku Postgres starting at $5/month and Heroku Data for Redis starting at $3/month, bringing the starting cost to at least $15/month. Because Dokku is an open-source, MIT-licensed Go/Bash tool, you can run your application and database containers on a single self-hosted server without any of these per-service or per-dyno subscription fees.
If I need to scale beyond Heroku's $5 Eco or $7 Basic plans, what are the financial implications of moving to the Standard tier instead of Dokku?
Upgrading to Heroku's Standard tier costs $25/dyno/month to get features like 512 MB RAM (Standard 1X) and horizontal scaling, which can become exceptionally expensive as you add more dynos. Hosting with Dokku avoids these per-dyno costs entirely, allowing you to deploy your Docker-powered applications and scale workloads using the raw capacity of your own single-server infrastructure.
関連する比較記事
機能と価格データは公式ドキュメントと料金ページを出典としており、最終確認日は 2026年6月25日 です。 誤りを見つけましたか?お知らせいただければ修正します。