<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: João Paulo Gomes</title>
    <description>The latest articles on DEV Community by João Paulo Gomes (@johnowl).</description>
    <link>https://dev.to/johnowl</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F233843%2Fdb0da06b-838c-4904-944a-7b5812958dad.jpeg</url>
      <title>DEV Community: João Paulo Gomes</title>
      <link>https://dev.to/johnowl</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/johnowl"/>
    <language>en</language>
    <item>
      <title>What if your service could ask for its own database?</title>
      <dc:creator>João Paulo Gomes</dc:creator>
      <pubDate>Fri, 20 Nov 2026 14:52:30 +0000</pubDate>
      <link>https://dev.to/johnowl/what-if-your-service-could-ask-for-its-own-database-1mgf</link>
      <guid>https://dev.to/johnowl/what-if-your-service-could-ask-for-its-own-database-1mgf</guid>
      <description>&lt;h3&gt;
  
  
  Meet Joomiz, the thing I've been building in my free time
&lt;/h3&gt;

&lt;p&gt;So, you are a developer and you need a database for a new service. You open a ticket. Someone from the platform team picks it up a few days later and creates the database by hand. If you work in a bigger company, maybe they run a pipeline instead. Then the password reaches you somehow: a chat message, an email, a file in a shared folder. Later you need the same thing for production, and the whole story starts again.&lt;/p&gt;

&lt;p&gt;I've seen this in some companies. Not because the platform teams are slow, but because they do the same work again and again for every team that asks.&lt;/p&gt;

&lt;p&gt;For the last months I've been working on a project to fix this. It's called &lt;strong&gt;Joomiz&lt;/strong&gt;, and today I want to show you what it does.&lt;/p&gt;

&lt;h3&gt;
  
  
  One file, next to your code
&lt;/h3&gt;

&lt;p&gt;The idea is simple. A feature team describes what a service needs in a file called &lt;code&gt;joomiz.yaml&lt;/code&gt;, in the service's own repository. The platform team decides what can be in that file by configuring the Golden Path. A Golden Path is the supported way to build a service in your company: the languages, the resources and the rules that the platform team offers and maintains.&lt;/p&gt;

&lt;p&gt;This is a small example, a Spring Boot jar running on Cloud Foundry with a PostgreSQL database:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="na"&gt;application&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;inventory-api&lt;/span&gt;
  &lt;span class="na"&gt;golden-path&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;java-web-application&lt;/span&gt;

&lt;span class="na"&gt;environments&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;test&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;deployments&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
      &lt;span class="p"&gt;-&lt;/span&gt; &lt;span class="na"&gt;id&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;inventory-api-test-deployment&lt;/span&gt;
        &lt;span class="na"&gt;type&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;cloud-foundry-app&lt;/span&gt;
        &lt;span class="na"&gt;artifact&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;inventory-api-jar&lt;/span&gt;
        &lt;span class="na"&gt;target&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;test-foundation&lt;/span&gt;
        &lt;span class="na"&gt;spec&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
          &lt;span class="na"&gt;instances&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;
          &lt;span class="na"&gt;memory&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;1G&lt;/span&gt;
          &lt;span class="na"&gt;env&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="na"&gt;DB_PASSWORD&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"${{managed-resources.inventory-api-db.outputs.password}}"&lt;/span&gt;

    &lt;span class="na"&gt;managed-resources&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
      &lt;span class="p"&gt;-&lt;/span&gt; &lt;span class="na"&gt;id&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;inventory-api-db&lt;/span&gt;
        &lt;span class="na"&gt;type&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s"&gt;postgres&lt;/span&gt;
        &lt;span class="na"&gt;spec&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
          &lt;span class="na"&gt;version&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"16"&lt;/span&gt;
          &lt;span class="na"&gt;storage-gb&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;20&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;I removed some lines to keep it short. Look at &lt;code&gt;DB_PASSWORD&lt;/code&gt;. There is no password in the file, only a reference to the output of the database. The password stays in your vault. The file can go to a public pull request and nothing leaks.&lt;/p&gt;

&lt;p&gt;There is much more in this file than I can show here: the team that owns the service, the artifact that the build produces, where it runs, and the resources that another team owns, like a Kafka topic. I'll explain the structure of the file in the next article.&lt;/p&gt;

&lt;h3&gt;
  
  
  Three commands
&lt;/h3&gt;

&lt;p&gt;If you used Terraform before, this part will look familiar. I borrowed the plan and apply idea on purpose, because it works.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;joomiz validate
joomiz plan --environment &lt;span class="nb"&gt;test&lt;/span&gt;
joomiz apply --environment &lt;span class="nb"&gt;test&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;validate&lt;/code&gt; checks the file against the rules of your platform. &lt;code&gt;plan&lt;/code&gt; shows what would change, and changes nothing. &lt;code&gt;apply&lt;/code&gt; makes the change, and only from a plan.&lt;/p&gt;

&lt;p&gt;This is the real output of a plan for the file above (I cut the ids):&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;plan for joomiz.yaml in environment test
  managed-resource/postgres: succeeded — Planned 1 change for managed-resource/postgres.
    inventory-api-db (created): succeeded — Would create 'inventory-api-db' (managed-resource/postgres) in environment 'test'. Nothing was changed.
  deployment/cloud-foundry-app: succeeded — Planned 1 change for deployment/cloud-foundry-app.
    inventory-api-test-deployment (updated): succeeded — Would update 'inventory-api-test-deployment' (deployment/cloud-foundry-app) in environment 'test'. Nothing was changed. 1 spec value is not known until apply.
      env.DB_PASSWORD: (known after apply)
plan succeeded
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The database does not exist yet, so its password does not exist yet either. The plan says exactly that: &lt;code&gt;(known after apply)&lt;/code&gt;. And the database comes first, because the deployment depends on it.&lt;/p&gt;

&lt;p&gt;Pretty nice! You see the whole environment before anything exists.&lt;/p&gt;

&lt;h3&gt;
  
  
  So, it's just Terraform?
&lt;/h3&gt;

&lt;p&gt;No, and I get this question a lot. (=&lt;/p&gt;

&lt;p&gt;Terraform is a tool for describing infrastructure in code. Joomiz sits one level above. A feature team does not write Terraform, and does not need cloud credentials. It writes a short file with the things the platform team already offers: a database, a Kafka topic, an endpoint in Kong, a deploy pipeline.&lt;/p&gt;

&lt;p&gt;Behind each one there is a &lt;strong&gt;provisioner&lt;/strong&gt;. A provisioner is a small service that follows an OpenAPI contract. The platform team can wrap a Terraform module, a script or a call to a cloud API in it. The control plane only talks to the provisioners, and it checks every value in the file against the catalog before any call. If your platform team says a database can have at most 100 GB, a file asking for 500 GB fails in &lt;code&gt;validate&lt;/code&gt;, not in production.&lt;/p&gt;

&lt;p&gt;Some other things that I care about:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Each team acts only on its own applications.&lt;/li&gt;
&lt;li&gt;Only one apply runs at a time for an application and an environment. The database enforces it.&lt;/li&gt;
&lt;li&gt;The audit trail records who applied what, and when.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;plan&lt;/code&gt; asks each provisioner what really exists, so a change made by hand shows up in the plan as drift.&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  And the AI agents?
&lt;/h3&gt;

&lt;p&gt;A few weeks ago I wrote here about &lt;a href="https://medium.com/@johnowl/are-agents-good-at-testing-3493465dc706" rel="noopener noreferrer"&gt;how coding agents test code&lt;/a&gt;. Agents are getting better every month, and many teams let them open pull requests already.&lt;/p&gt;

&lt;p&gt;Now think about an agent that can also run commands against your cloud. One wrong change can reach production, and nobody can say who approved it.&lt;/p&gt;

&lt;p&gt;With Joomiz the agent works the same way a developer does. It changes the &lt;code&gt;joomiz.yaml&lt;/code&gt; file. The credentials stay with the provisioners. A person reviews the plan in the pull request, and the apply can run from your pipeline. The agent gets no shortcut, and that's the point.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where it is today
&lt;/h3&gt;

&lt;p&gt;Joomiz is in beta, and I'm still working on it every week. The control plane, the CLI and the language server for VS Code (it gives completion and error messages in the &lt;code&gt;joomiz.yaml&lt;/code&gt; file) are working. There are real provisioners for Cloud Foundry apps and Azure DevOps pipelines, and I'm working on Azure right now.&lt;/p&gt;

</description>
      <category>platformengineering</category>
      <category>java</category>
      <category>springboot</category>
      <category>database</category>
    </item>
  </channel>
</rss>
