Your First Oracle Autonomous Database on OCI Always Free

The single best way to keep your Oracle skills sharp without a license, a server, or a finance conversation is OCI Always Free. It gives you a real, fully managed Oracle database — Autonomous Database, running the current 23ai release — that stays free indefinitely. It's the ideal sandbox: a place to test commands, try new-release features, and generate real screenshots, all on infrastructure that's yours and costs nothing. Here's how to stand one up in a few minutes, what's genuinely worth doing with it, and the one catch that catches people out. One ground rule first: do all of this on a personal OCI accoun, never an employer's tenancy. The short version. Sign up at oracle.com/cloud/free, create an Autonomous Database with the Always Free toggle on, and connect through the browser via Database Actions → SQL — no client install. You get a real, managed Oracle 23ai database with AI Vector Search included, free forever. The one catch: it stops after 7 days of inactivity and is reclaimed after ~90 days stopped, so log in occasionally and it's yours to keep. What "Always Free" actually includes OCI's Always Free tier is genuinely free forever — not a 30-day trial — and the database piece is the star: 2× Autonomous Database, each with 1 OCPU and 20 GB of storage — the focus of this post. Arm Ampere A1 compute: up to 4 OCPUs + 24 GB RAM total — enough to also run Oracle Database Free yourself on a VM, if you want the self-managed side too. Object storage, networking, and a handful of other services. The catch worth knowing up front: an Always Free Autonomous Database stops automatically after 7 days of inactivity (your data is preserved), and if it stays stopped for 90 cumulative days it can be reclaimed and permanently deleted. "Activity" means an actual connection running SQL — so just log in and run something every week or two and it's yours indefinitely. Set a calendar nudge; that 90-day clock is how most people lose their sandbox. Step 1 — Create a personal OCI account Sign up at oracle.com/cloud/free with a personal email. Identity verification asks for a card, but Always Free resources are never charged — the card is for identity and for if you later opt into paid resources (you won't need to). Pick a home region close to you; if you also want the free Ampere VM, choose a region with A1 capacity. Your home region is permanent, so choose deliberately. Step 2 — Provision the Autonomous Database In the Console, go to Oracle Database → Autonomous Database → Create Autonomous Database: Workload type: Transaction Processing (ATP) or Data Warehouse (ADW). ATP is the natural default for a general-purpose learning sandbox — it behaves like the OLTP databases you meet day to day. Pick ADW only if you're specifically playing with analytics/columnar workloads. (JSON and APEX flavors also exist; ignore them for now.) Always Free: toggle it on. This is the switch that matters — it's easy to miss, and without it you're provisioning a paid instance. Set a strong ADMIN password, leave everything else at defaults, and click Create. It's ready in a minute or two. Step 3 — Connect (no client install needed) The fastest path needs nothing on your laptop. From the database's detail page, open Database Actions → SQL — a full browser-based SQL worksheet (SQL Developer Web). Log in as ADMIN and you're querying: -- prove you're on a current release SELECT banner_full FROM v$version; SELECT database_role, open_mode FROM v$database; Enter fullscreen mode Exit fullscreen mode v$version confirms exactly which 23ai build your Always Free ADB is on the day you provision it — worth checking, because Oracle keeps the managed service current for you. For a desktop client or an application, download the wallet (the mTLS credentials bundle) from the database page and point your tool at the connection strings inside it. But for learning and quick tests, Database Actions in the browser is all you need. Step 4 — Put something in it A sandbox is more useful with data. This runs as-is under the ADMIN schema: CREATE TABLE demo_orders ( id NUMBER GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY, customer VARCHAR2(60), amount NUMBER(10,2), created DATE DEFAULT SYSDATE ); INSERT INTO demo_orders (customer, amount) SELECT 'Customer ' || LEVEL, ROUND(DBMS_RANDOM.VALUE(10, 1000), 2) FROM dual CONNECT BY LEVEL <= 1000; COMMIT; SELECT COUNT(*) rows_loaded, ROUND(AVG(amount), 2) avg_amount FROM demo_orders; Enter fullscreen mode Exit fullscreen mode That's a thousand rows to experiment against — enough to try indexes, execution plans, or a query you want to tune. Step 5 — Try something only the new releases can do The reason to sandbox on a current release is to play with what's new — and on 23ai the headline is AI Vector Search: storing embeddings (the numeric fingerprints of text, images, etc.) right next to your relational data and querying them by similarity instead of exact match. It's included on Always Free at no extra cost. You'd normally generate embeddings from a model, but you can see the whole mechanism with tiny hand-built vectors — no model required: CREATE TABLE docs ( id NUMBER PRIMARY KEY, content VARCHAR2(50), embedding VECTOR(3, FLOAT32) -- 3 dimensions, just to watch it work ); INSERT INTO docs VALUES (1, 'cat', TO_VECTOR('[0.9, 0.1, 0.0]')); INSERT INTO docs VALUES (2, 'kitten', TO_VECTOR('[0.8, 0.2, 0.0]')); INSERT INTO docs VALUES (3, 'airplane', TO_VECTOR('[0.0, 0.1, 0.9]')); COMMIT; -- which rows are most "similar" to cat? (smallest cosine distance) SELECT id, content, ROUND(VECTOR_DISTANCE(embedding, TO_VECTOR('[0.9, 0.1, 0.0]'), COSINE), 4) AS distance FROM docs ORDER BY distance FETCH FIRST 3 ROWS ONLY; Enter fullscreen mode Exit fullscreen mode kitten comes back as the nearest neighbor to cat and airplane as the farthest — the same operation that powers semantic search and RAG, just with three dimensions instead of a thousand. Swap in real embeddings from a model and add an HNSW or IVF vector index, and this is production AI search. On a free database you fully control. No OCI account? Run it locally. The AI Vector Search lab runs this exact VECTOR / VECTOR_DISTANCE demo on Oracle Database Free with Docker — vector search ships in the free image too — so you can watch kitten land nearest cat in about two minutes, no cloud signup: ./run.sh up && ./run.sh all. What this sandbox is good for — and what it isn't Verifying before you publish. Run the exact commands from a blog post, runbook, or Stack Exchange answer on a real instance instead of trusting memory. (It's how the SQL in these posts gets checked.) Original screenshots and reports. The OCI console, Database Actions, real query output — concrete assets that make your writing credible. You can even generate AWR-style performance reports to practice reading them. Learning the cloud-native side. Provisioning, scaling, automatic backups, and the managed-service model you'll meet on real Oracle-to-cloud migrations — Autonomous is one of the most common migration targets. What it won't do: because Always Free ADB is a fully managed service, you don't configure the infrastructure — so you won't stand up RAC or Data Guard on it (those are separate, self-managed exercises). But as a zero-cost, always-on, current-release Oracle database you fully control, it's very hard to beat. FAQ Is OCI Always Free actually free, or a trial? It is genuinely free forever, not a time-limited trial. The Always Free tier includes two Autonomous Databases (1 OCPU and 20 GB each), Arm Ampere A1 compute (up to 4 OCPUs and 24 GB RAM), and some storage and networking — all at no charge for as long as you use them. Oracle also offers a separate 30-day free trial with credits, but the Always Free resources persist beyond it. What version of Oracle Database does Always Free Autonomous run? Always Free Autonomous Database runs Oracle Database 23ai, the current release, and Oracle keeps the managed service patched and current for you. Run SELECT banner_full FROM v$version in Database Actions to see the exact build on the day you provision it. Can I use AI Vector Search on the free tier? Yes. AI Vector Search is a core Oracle Database 23ai feature and is included at no additional charge on Autonomous Database, including the Always Free tier. You can create VECTOR columns, run VECTOR_DISTANCE similarity queries, and build HNSW or IVF vector indexes without paying anything. Will my Always Free database get deleted? It can, if you abandon it. An Always Free Autonomous Database stops automatically after 7 days of inactivity (its data is preserved), and if it stays stopped for 90 cumulative days it may be reclaimed and permanently deleted. Connecting and running SQL resets the inactivity clock, so logging in every week or two keeps it indefinitely. Can I run RAC or Data Guard on Always Free Autonomous Database? No. Autonomous Database is a fully managed service, so you do not configure the underlying infrastructure — RAC and Data Guard are not something you set up on it. Oracle handles high availability and backups for you behind the scenes. To practice building RAC or Data Guard yourself, use your own Enterprise Edition environment or lab, not Always Free ADB. Do I need to install anything to connect? No. From the database detail page, open Database Actions and use the browser-based SQL worksheet — nothing to install. For desktop clients or applications, download the wallet (mTLS credentials) from the database page and use the connection strings it contains. Originally published at uptimearchitect.com. I write here in a personal capacity — questions or feedback are welcome via the contact page.