Monghoul against DataGrip

Monghoul is a tool for MongoDB and nothing else. Autocomplete suggests your own field names and values and follows them through each stage of a pipeline, the aggregation builder runs a pipeline up to any stage, and a Database Map opens the document behind an id. In our test it took 105.6 MB on disk against 2,766.6 MB for DataGrip, and its window appeared in 0.07 s against 3.53 s. Its free tier allows commercial use.

DataGrip is one IDE for PostgreSQL, MySQL, Oracle, SQL Server, MongoDB and more. If you want one application for every engine you open, that is a strong reason to choose it, and Monghoul can sit beside it for the MongoDB part.

What was measured

The numbers, and how they were taken:

After a first launch at 8.72 seconds, which is the cost of a cold start, DataGrip settled between 3.38 and 3.57 s and held 1,185 to 1,210 MB. Monghoul ran 0.070 to 0.089 s and 396 to 418 MB.

The memory row flatters DataGrip, and we leave it that way. 1,194 MB is the welcome window, before a project is open and before it has indexed anything. That is the cheapest state DataGrip has, and it is the state the script can measure the same way for every client. Ours is a full application window with the sidebar loaded.

2,766.6 MB is not a typo. The measurement counted 2,210 files in the install directory, including a bundled Java runtime and a bundled Chromium. DataGrip ships two runtimes because it is a JetBrains IDE with an embedded browser that supports a dozen database engines.

Two startup numbers, because there are two windows. DataGrip shows a splash at 1.12 seconds and its welcome window at 3.53 seconds. The row above is the welcome window, because that is the first one you can do anything with.

Where Monghoul is ahead

Autocomplete from your data

Autocomplete covers the MongoDB 8.x operator surface, plus the active database’s real collection names, field paths, types and values. Typing db. suggests that database’s collections, and inside a pipeline the suggestions follow the fields each stage adds or removes. JetBrains document that only the default database is introspected until you add the others. Monghoul lists every database the connection can see, and it samples the collections your query uses, automatically by default. The full behaviour.

Aggregations as a first-class job

Stages are rows you add, drag, skip and run. Run to here executes the pipeline as far as that row and shows what it returns, and a stage your server is too old for is marked with the version it needs. Eight quick starts cover the common shapes. JetBrains’ MongoDB documentation does not mention a pipeline builder. The aggregation builder.

Ids that lead somewhere

In SQL a foreign key tells the IDE which row a value points to. MongoDB has no foreign keys, so the Database Map works it out: a scan reads a sample of each collection, finds the fields that hold another collection’s ids, and looks some of them up to check. After that, a click on an id in a result opens its document, and $lookup is offered the right collection and fields first. The map is on the free tier. How the Database Map works.

An explain plan built for MongoDB

A find or an aggregation gets its explain plan by default, drawn as a tree rather than shown as JSON and graded as one of four. An aggregation also gets a chart of how the document count changes through its stages. JetBrains’ MongoDB documentation does not mention an explain plan view. Explain.

A guard on production writes

Mark a connection, a database or a single collection as protected, and every write to it asks for confirmation first. The driver enforces it while the call runs, which also stops a write that came from an alias or a raw command. Where that check runs decides what it catches.

Credentials kept encrypted

JetBrains document that credentials on a URL-only connection are “stored in plain text as is, and is also available in IDE log file and data source information”. Ours are encrypted under a key held by the operating system’s credential store. Credential storage.

A free tier you can use at work

DataGrip has a free non-commercial licence and no Community edition, so commercial use is a subscription. Our free tier allows commercial use, with 8 saved connections, no time limit and no signup.

Where DataGrip wins

One window for every database you use. PostgreSQL, MySQL, Oracle, SQL Server and MongoDB in one IDE. Monghoul speaks MongoDB only, so if you also maintain a Postgres schema, it is a second application and DataGrip is not.

Older servers and a wide authentication list. JetBrains document MongoDB 3 through 8 with SCRAM, AWS IAM, x.509, Kerberos and LDAP. Monghoul requires server 4.4 or newer, so a MongoDB 3.6 instance that has not been migrated yet is a job for DataGrip.

The JetBrains editor, which many people already know. If you live in IntelliJ or WebStorm, the keymap, the refactoring and the version-control integration are already in your hands.

What each one costs to keep using

JetBrains say DataGrip cannot be split into free and paid tiers, so there is a free licence for non-commercial use and none for work. A subscription kept for 12 months in a row leaves a perpetual fallback licence for the version you had. The trial is 30 days.

Monghoul falls back to its free tier, which a company can use, with 8 saved connections and no expiry. Pro adds charts, cluster monitoring and AI access, and its trial is 14 days per device with no card.

Capability by capability

Quoted from JetBrains’ own MongoDB documentation. Where it does not mention a capability, this table says so rather than guessing: an undocumented feature is not the same as an absent one.

Which one to pick

Try both

A table cannot tell you whether DataGrip’s MongoDB support is deep enough for your work, or how either tool feels after a week. DataGrip has a 30 day trial and our free tier has no time limit and no signup, so opening your own cluster in both costs an afternoon.

Reproducing the numbers

git clone https://github.com/Kontsedal/monghoul-public
cd monghoul-public/benchmarks/client-footprint
.\measure.ps1 -Runs 7 -Only 'DataGrip|Monghoul'

It force-stops matching processes between runs, so close anything you care about first. The DataGrip path in the script carries its version number, so check it against your own install before running.