Learn

Architecture & pattern guides

Read the idea, build it on a Grid, prove it in a challenge. Short, practical guides — illustrated with RodGrid's own nodes — on system design, design patterns, flow modeling, and how to get the most out of the app.

All articles

Making the catalog yours: custom components

On Pro or higher, build a node the catalog doesn't have, reuse it across your Grids, and know what it keeps when it travels.

Your payments pass through a service called Tollbooth. It applies your fees, half your systems call it, and no catalog will ever list it. A custom component, which the app calls a custom node, puts it on a Grid under its own name and picture: you define it once, and it stays in your workspace.

NoteCustom components need Pro or higher, image upload included. On someone else's Grid, the owner's plan decides whether you can create one there: on a Premium owner's Grid, a guest on Free can, but it stays on that Grid and nothing reaches their own library. Leave Pro and the custom nodes already on your Grids stay, but you can't create or edit any.

Is it really missing?

Search the palette before you build. "Redis", "S3", "Kafka", even "ERP" and "Legacy System" are already in there, on the plans listed in The node palette & component families, and a custom twin of one costs you what the catalog entry knows and gains you little beyond a picture. Often only the name is yours: Tollbooth may simply be a REST API, in which case place REST API and rename it (Naming nodes: inline rename & the inspector).

Search before you build: a generic catalog node usually beats a custom one that looks more specific.

Building Tollbooth

In the palette's Custom group, press Create Custom Node; if it is locked, its line reads Requires Pro or Premium for your plan or Not included in this Grid's plan for the owner's. Fill in Name, Icon / Emoji, Description, Category and Accent Color, then press Create Node and drag Tollbooth from the group onto the canvas, because creating it doesn't place it.

A name is not a role

A catalog component carries a semantic role: what the traffic simulation reads to know that a cache answers a share of the reads itself, set by its hit rate. A custom component defines a title, a picture and a category, and no role. It lands in the Custom group whatever the category. The category labels the node and steers the connection checks, so a custom database drawn calling a backend gets a warning.

The category alone gives the simulator nothing. A custom node called Redis Cache in the database category simulates as a generic service with default capacity and no hit rate, which is the real cost of a custom twin and why one is no way around a locked catalog component. For Tollbooth that reading is fine, because Tollbooth does sit in the path as a service.

Its title is another matter, since the simulator still reads certain words there, and "cache" on its own is not one of them. Call a node Customer Hub and it reads as a source of traffic; call one CDN Proxy and it reads as a cache. If a custom node behaves oddly under load, check its title first, then the Description in its Node Inspector, where some of the same words count. What the panel then does with those readings belongs to Stress testing a design before it exists.

The AI features work mainly from what is written on the node: its title and its own Description, the one in the Node Inspector. The Description in the Custom Node dialog helps palette search and doesn't reach the placed node, so write what Tollbooth does in the inspector.

One definition, several copies

A custom component belongs to your workspace, not to the Grid you happened to draw it on, so the next Grid's Custom group offers it too. A workspace can hold 200 of these definitions and one Grid 50; placed copies don't count toward either. Placing one copies its definition into that Grid and its look into the node, which is why an exported or shared Grid keeps its title, icon and colour for someone who has never seen your library.

Those copies cut both ways. Say Tollbooth gets a new icon. The pencil beside it in the palette opens Edit Custom Node, and Save Changes updates your library and this Grid's palette entry. What it doesn't touch:

  • Nodes already placed. They keep the look they were placed with, on this Grid too. Only newly placed ones get the new icon.
  • Other Grids that hold it. Their palette entry stays as it was until you edit it there.

Remove from library, in the same dialog, takes it out of your library and this Grid's palette. Placed nodes stay, and so do copies on other Grids.

Logos travel less far than names

Upload image adds a logo that wins over the icon, and your account holds 50. The image is referenced, not copied: you can load your own while signed in, and so can people with access to the Grid you uploaded it from. Everyone else sees the icon or emoji: a guest on a different Grid, someone on another account who imports your JSON, a reader without an account. So choose the icon first, because choosing one afterwards clears the image. Deleting an image is permanent, and components using it fall back to a plain icon.

Stop at one

A Challenge's own palette has no Custom group (see How Challenge Mode works), so take the brief of the Serverless Web App on AWS Challenge to a new Diagram Grid and search the palette for each requirement: all of them are in there. Then add the one thing the brief lacks, an in-house service your functions call, as a custom node. The brief has no client, and CloudFront isn't one to the simulator, so add a Client App, connect it through to your functions, and see how Simulate reads your node. If you want a second custom node, search again first: it is usually a catalog component you haven't found.

Practice this