Heroku

Heroku is the platform that taught the rest of the industry what "git push to deploy" should feel like. It's been around since 2007, owned by Salesforce since 2010, and somehow still the easiest way to deploy a Rails or Django app.

Some folks wrote Heroku off years ago. They were wrong. Heroku quietly remained the path of least resistance for hobby projects, MVPs, and surprisingly large production apps.

If you've evaluated Render or Fly.io and found yourself missing certain ergonomics, you're rediscovering what Heroku gets right.

What Heroku actually does

Heroku runs your app in containers it calls "dynos." You push your code, Heroku detects the language, builds it, and runs it. Databases, queues, and add-ons attach with one click or one CLI command.

The philosophy is "we'll handle the boring infrastructure stuff, you focus on shipping." It works because Heroku has been refining that pitch for almost two decades.

The buildpack model

Heroku detects what kind of app you have, applies a buildpack, and produces a runnable image. You can use the official buildpacks for Ruby, Python, Node, Go, Java, PHP, and others, or write your own.

Who Heroku is for

Solo developers shipping side projects. Startups that don't want to hire a DevOps engineer in year one. Teams that have already built on Heroku and don't have a reason to migrate.

It's not the right call for cost-sensitive scale-up workloads. Once you're running serious traffic, Heroku's pricing curve gets steep relative to bare-metal or AWS.

The "I just want to deploy" user

Heroku still wins on time-to-first-deploy. You can take a fresh repo to a public URL in under five minutes. Most competitors haven't matched that, despite years of trying.

Pricing breakdown

Eco and Basic dynos for hobby use. Standard and Performance dynos for production. Heroku Postgres has its own tier list, as do Redis and Kafka add-ons.

The free tier is gone. That stung. The Eco plan replaced it with a low-cost shared-pool option, but it's not free anymore. Hobbyists noticed.

The add-on marketplace

Anything you'd want to attach (Postgres, Redis, search, monitoring, error tracking) is available as a one-click add-on with usage-based billing. Convenience tax is real but real.

Standout features of Heroku

The CLI. `git push heroku main` still feels magical fifteen years later. Logs, scaling, config vars, and rollbacks are all one command away.

Pipelines and review apps. Every PR gets its own staging environment automatically. This was state-of-the-art in 2014 and is still better than most modern PaaS approaches.

Postgres in particular

Heroku Postgres is a genuinely good managed database. Followers, forks, and dataclips give you operational tools that take serious effort to build elsewhere.

Honest tradeoffs with Heroku

It's expensive at scale. Past a few medium dynos and a Postgres tier, you're paying for convenience in ways that start to add up.

The platform's pace of innovation slowed under Salesforce. New features arrive but not at the rate competitors ship. Heroku is mature, which has upsides and downsides.

Heroku vs alternatives

Heroku vs Render: Render is the spiritual successor. Cheaper, modern UI, similar buildpack-style deploys.

Heroku vs Fly.io: Fly is more of a global container platform. Heroku is more of a hand-holding PaaS.

Heroku vs Railway: Railway is the closest current pretender to Heroku's throne for indie devs. Better pricing, similar UX.

For broader views, see the best PaaS platforms or browse Heroku alternatives and Heroku vs Render.

When Heroku wins

You want to ship today. You don't want to learn Kubernetes. You're already on Heroku and migration cost outweighs savings.

Bottom line on Heroku

Heroku is the boring, reliable, expensive choice that ships your app. It's not the cheapest, it's not the trendiest, but it works.

If you're a startup with funding and you want to focus on product, Heroku is still defensible. If you're cost-sensitive or you've outgrown the dyno model, look at Render or Railway. See tools for developers for related picks.

Why Heroku still matters in 2026

The PaaS category exists because Heroku invented it. Render, Railway, Fly, and Vercel all owe Heroku their existence. The git-push-to-deploy paradigm is everywhere now, but Heroku is where it started.

Maturity matters. Heroku's edge cases are documented. Its failure modes are known. The community has been writing tutorials for fifteen years. When something goes wrong at 2am, the answer is usually one Stack Overflow post away.

The Salesforce ownership question

Yes, Heroku is owned by Salesforce. No, that hasn't visibly killed the product. Innovation is slower than competitors but the platform is still improving. The fear of "they'll shut it down" has been around for years and hasn't materialized.

Heroku in modern stacks

Rails apps still feel native here. Django apps work cleanly. Node, Go, and Java are all first-class. Newer stacks like Bun and Deno are workable but require some configuration.

The platform handles SSL, scaling, logging, and config management. You don't think about any of it. That's the value proposition, and it's still hard to beat.

Heroku Connect for Salesforce shops

If your company uses Salesforce, Heroku Connect lets you bidirectionally sync Postgres tables with Salesforce objects. This is a real differentiator that no competitor matches.

Common Heroku questions

Is Heroku still free? No, the free tier ended. Eco dynos start at a few dollars per month. Does Heroku support custom domains? Yes, with automatic SSL. Can you scale down to nothing? Eco dynos sleep but other tiers don't.

For more options, see tools for Rails developers and Heroku vs Railway.

Final take on Heroku

Heroku is the PaaS that earned its reputation honestly and has kept it through changes most products wouldn't survive. For startups that want to ship today and not think about infrastructure, it's still defensible. The premium is the price of zero-friction shipping.

Heroku's add-on ecosystem in detail

The add-on marketplace covers databases, caching, search, monitoring, error tracking, queues, scheduling, and more. Each add-on attaches to your app with one CLI command and provisions on your behalf. Billing rolls into your Heroku invoice.

This convenience is part of why Heroku is sticky. You're not separately negotiating contracts with five vendors. The platform aggregates the operational overhead. The downside is a markup on each service compared to going direct.

Heroku Postgres specifics

Heroku Postgres tiers run from Hobby to Premium and Private. Followers (read replicas) and forks (point-in-time copies) are first-class operations. Dataclips let you share read-only SQL queries with stakeholders without giving them production access.

Heroku in Salesforce-adjacent stacks

Heroku Connect bidirectionally syncs Postgres tables with Salesforce objects. For companies running Salesforce as their system of record, this is a unique capability. You can build apps on Heroku that read and write Salesforce data without API gymnastics.

This is a niche but important use case. Salesforce shops with engineering teams often default to Heroku precisely because of this integration depth. Outside the Salesforce ecosystem, this matters less.

The buildpack flexibility

Buildpacks detect your language and configure the build process. Custom buildpacks let you handle non-standard stacks. The buildpack system is open-source and well-documented. Most languages have community buildpacks that work without modification.

For weird stacks, you might need a Dockerfile-based deploy, which Heroku supports through Container Registry. The CLI handles the workflow even when you're shipping containers rather than buildpack-detected apps.

Performance dynos and scaling

Performance dynos give you dedicated resources, faster boot times, and consistent performance. For production apps that need predictable behavior, the upgrade is worth it. Eco and Basic dynos work for development and low-traffic apps but performance varies.

The migration question

Migrating off Heroku usually means moving to Render, Railway, Fly, or AWS. Render is the closest spiritual successor. Railway has the indie-dev appeal Heroku used to own. Fly is for global container deployments. AWS is for teams ready to absorb operational complexity in exchange for cost savings.

The migration cost depends on how deeply you've used Heroku-specific features. Apps that just use dynos and Postgres migrate cleanly. Apps that lean on Connect, Pipelines, and specific add-ons require more rework.

Heroku wrap-up

The platform's longevity is itself a feature. Fifteen years of operational learning have hardened Heroku's edge cases. Newer competitors will eventually accumulate similar maturity, but they don't have it yet. For mission-critical apps, that maturity matters.

The Salesforce ownership remains a quiet question. Innovation has slowed compared to indie competitors. The pricing structure has stayed mostly the same for years. The fear of platform shutdown has lurked for a decade without materializing. Reality has been calmer than the headlines.

The "should I migrate" framework

If your Heroku bill is climbing past what alternatives would cost, run the migration math seriously. Render and Railway are the closest replacements with similar UX. Fly is for global container deployments. AWS or GCP are for teams ready to absorb operational complexity. The right answer depends on your team's appetite for ops work, not just on cost.