CloudBeaver vs pgAdmin

CloudBeaver and pgAdmin are two web-based database managers you can run on a server and open from any browser. pgAdmin is the long-standing administration tool built for PostgreSQL alone. CloudBeaver Community is the web edition of DBeaver, a general-purpose client that connects to PostgreSQL, MySQL, SQLite, SQL Server, ClickHouse and other JDBC databases. The real question is whether you administer PostgreSQL, or browse and query several kinds of database from one place. After that come two narrower checks: which features the free CloudBeaver edition keeps, and how much memory each one asks for.

One engine, or many

pgAdmin is written for PostgreSQL. Its README calls pgAdmin 4 "a rewrite of the popular pgAdmin3 management tool for the PostgreSQL database", and its PSQL tool connects to "PostgreSQL or EDB Advanced server". What you get for that focus is depth in PostgreSQL itself: an object browser, a Query Tool, role management, backup and restore, and the tools described below.

CloudBeaver Community ships with drivers already downloaded for PostgreSQL, MySQL, MariaDB, SQLite, SQL Server, Oracle, ClickHouse, DuckDB, Firebird, Db2, H2, Trino, Databend, LibSQL and Apache Kyuubi. Its docs note that not every database is available by default. Adding a driver to the Community Edition means changing the server build: the guide says "You must include a driver in the server build". Many more drivers, such as Cassandra, Couchbase and Athena, are listed only for the commercial editions.

If PostgreSQL is the only database on the box, pgAdmin is the natural fit. If the same person needs to look at a MySQL app database, a SQLite file and a ClickHouse instance, CloudBeaver covers them in one UI.

PostgreSQL tools

pgAdmin includes:

• The Query Tool, which shows execution plans as text, graphically or as a table, with Explain and Explain Analyze buttons.
• An ERD Tool that draws diagrams for an existing database, lets you design tables and relationships, and generates "ready to run SQL" from the design.
• Schema Diff, which compares two databases or schemas and generates synchronisation scripts.
• A PSQL Tool: psql in the browser, connected to the selected database.
• Backup and restore. The container includes pg_dump, pg_dumpall, pg_restore and psql, with binaries for several PostgreSQL major versions so they match the server you back up.

CloudBeaver Community has a SQL editor, a data editor, execution plans, and data import and export. Its 26.2 release added a References panel that shows related rows from linked tables. It also added an AI chat for writing and fixing queries, set up by an administrator with OpenAI or Copilot tokens. The release notes say "all AI features can be turned off" in server configuration. ER diagrams are not part of Community. CloudBeaver's Entity Diagrams page is tagged for its Enterprise, AWS and Team editions only.

Users and sign-in

pgAdmin in server mode (which the container runs) has its own user accounts. The first admin comes from PGADMIN_DEFAULT_EMAIL and PGADMIN_DEFAULT_PASSWORD, which the docs say are "required and must be set at launch time". AUTHENTICATION_SOURCES can add LDAP, Kerberos, OAuth2/OpenID Connect or webserver authentication (any front end that sets REMOTE_USER). Two-factor authentication supports email codes or an authenticator app. All of this is in the free, PostgreSQL-licensed product.

CloudBeaver Community offers three sign-in types: anonymous access, local users created by an administrator with roles, and reverse-proxy header authentication. SSO, OpenID (including Keycloak, Okta and Microsoft Entra ID), Google, JWT and NTLM are listed as Enterprise Edition methods. Audit logging is also tagged for the paid editions.

For single sign-on with a self-hosted identity provider, pgAdmin can do it directly. CloudBeaver Community needs a reverse proxy that handles the login and passes the user in a header.

Where each keeps its own data

pgAdmin stores its configuration database, users and saved server definitions under /var/lib/pgadmin. The container runs as UID 5050, so a bind-mounted host directory must be owned by 5050:5050. Server definitions can be preloaded from a servers.json file, which is read only on first launch. Any config.py option can be set with a PGADMIN_CONFIG_ prefixed environment variable.

CloudBeaver keeps everything in its workspace directory, /opt/cloudbeaver/workspace. Its internal database of users, access rules and query history is H2 in the Community Edition. PostgreSQL, MariaDB, Oracle and SQL Server are listed as options only for the paid editions. The first start normally opens a server-setup screen. The CB_SERVER_NAME, CB_SERVER_URL, CB_ADMIN_NAME and CB_ADMIN_PASSWORD variables skip it, which is what our install uses. Upstream says to back up the workspace before every update, and that "downgrades are not supported".

Memory and install

The published requirements are very different. CloudBeaver's Community guide for Google Cloud asks for "at least 1 CPU and 4GB RAM for optimal CloudBeaver performance". It is a Java server. pgAdmin publishes no RAM figure; our 512 MB is an estimate.

Measured idle on our GCP e2-standard-2 test box (2026-09-29), CloudBeaver used ~213 MB of RAM and pgAdmin ~227 MB. That is idle, not under load: queries, imports and more users need more. For CloudBeaver, upstream's 4 GB is the number to plan around.

Both installs are one docker run with a named volume, and both are tested on this site. We pin dpage/pgadmin4:9.18 and dbeaver/cloudbeaver:26.2.1. We rate both 1 out of 5 to deploy.

Neither should face the internet without TLS. Our installs serve plain HTTP (pgAdmin listens on port 80 unless PGADMIN_ENABLE_TLS is set; CloudBeaver on 8978), and each holds credentials for your databases.

Licences: pgAdmin uses the PostgreSQL licence; CloudBeaver Community is Apache-2.0. Both are OSI-approved.

Which to choose

Choose pgAdmin if your databases are PostgreSQL. It has the ERD tool, Schema Diff, the PSQL tool, and pg_dump/pg_restore for several server versions, plus LDAP, OAuth2/OIDC and two-factor login, all free.

Choose CloudBeaver Community if you need one browser UI for several engines (MySQL, SQLite, SQL Server, ClickHouse and PostgreSQL side by side) and can give it the memory upstream asks for. Put an authenticating reverse proxy in front if you want SSO. Plan on the paid edition if you need ER diagrams, OIDC login or audit logs.

Choosing between pgAdmin and CloudBeaver

Starting from pgAdmin, the usual reason to move is a second database engine that pgAdmin cannot open. Starting from CloudBeaver, the reason to move is usually PostgreSQL depth or free SSO. Nothing stops you running both, with pgAdmin for the PostgreSQL servers and CloudBeaver for the rest.