How it compares

Three ways to reach a database

Comparing feature checkboxes with other editors would go stale in a quarter and would not be why anyone switches. The difference that lasts is structural: where the tool runs relative to the data. There are three answers, each good at something, and each with a bill attached.

The shapes

What each approach is, and what it costs

The desktop client

DataGrip, DBeaver, TablePlus, pgAdmin in desktop mode

Good at The richest editors in the category, and they answer to nobody but you. Offline, fast, deeply featured, and yours to configure.

Costs you It runs on your laptop, so the database has to be reachable from your laptop: a public port, a bastion, a VPN or an SSH tunnel. Every new machine and every new colleague repeats that setup, and the credential ends up on each of those machines.

The hosted SQL editor

Vendor-run web consoles and SaaS query tools

Good at Nothing to install and nothing to run. A link, a login, and a query box.

Costs you Your data crosses a boundary you do not control, and the vendor needs a route into your database to make that work. For some data that is a procurement conversation; for some it is a no.

Deployed beside the data

LibreDB Studio

Good at One container inside the same network as the database. Nothing is exposed, nothing leaves the perimeter, and the browser is the only client anyone installs. Access is granted through your identity provider and revoked there too.

Costs you Something has to run it. That is a container or a Helm release you own and update — cheap next to a VPN rollout, but it is not nothing.

In practice

What the choice changes on an ordinary Tuesday

Two columns rather than a row per product: these follow from where the tool runs, so every desktop client shares the left column no matter which one you use.

The task	On your laptop	Beside the data
Reaching a database in a private network	Expose a port, or dig a tunnel per person, per machine.	Already inside it. Nothing is exposed to reach it.
Onboarding the fifth engineer	Install, licence, configure, distribute credentials.	Send a URL. Their SSO account decides what they can see.
Revoking access on their last day	Rotate every credential they were given a copy of.	Disable the account in your identity provider.
Answering "who ran that query?"	Whatever the database server happened to log.	An audit trail against a person, in the tool.
Reading production from a phone at 2am	Not a thing a desktop client does.	A browser is a browser.
Working across several engines at once	Depends on the tool; often several tools.	Seventeen engines behind one tab and one grid.

To be fair

This is not a claim that desktop clients are bad

They are the most capable editors in the category, and for a laptop-and-a-local-database day they are the right answer. The argument here is narrower: the databases moved into Kubernetes, into PaaS platforms and into customers’ private networks, and a tool that runs on your laptop now has to be tunnelled to each of them. LibreDB Studio moves the tool instead of the network.

The editor itself is MIT and free, so the way to settle this is to run it rather than read about it. Get started in about ten seconds → · Try it with no install →