Migrating 20,000 diagrams in one weekend: how a global technology company moved from Gliffy to draw.io

draw.io got the chance to sit down with Asia, Software Engineering Enablement Lead at a global technology company operating in the communications and security space, to talk through their migration from Gliffy to draw.io.

We asked them why they made the switch, how their teams use draw.io day to day, and how it’s helped them successfully operate across a large, distributed organization.

The company had been using Gliffy for over ten years across roughly 15,000 Confluence users. When their Gliffy contract was coming to an end, their team ran its own comparison of the diagramming tools on the market, looking at draw.io, Lucidchart, and Gliffy itself.

Lucidchart was ruled out early, mainly because it sits outside the Atlassian ecosystem and is not a native Confluence app. From there, the decision came down to research: checking reviews, pricing, security, and functionality, and draw.io ticked all these boxes. That’s also where their team discovered that tools they were already using informally, PlantUML and Mermaid, were both supported natively within draw.io, which helped consolidate their toolset even further.

To stress-test draw.io before committing, the team ran trial licenses using some of their most complex technical diagrams; predominantly hardware and software network diagrams. Comparing the results side by side with their Gliffy originals, the outcome was reassuring: the diagrams held up well, and the structure and content were preserved.

The scale of the migration was substantial: over 20,000 diagrams spread across nearly 9,000 Confluence pages. Rather than tackle Confluence Data Center to Cloud migration and the Gliffy-to-draw.io switch at the same time, the team migrated diagramming tools first, before moving to the Cloud (this is what we also recommend).

“One weekend, turned it [the Gliffy Mass Importer] on — it just did it!”

The result exceeded expectations. Using draw.io’s one-click migration, the switch from Gliffy to draw.io took about a week from start to finish. Out of 20,000 diagrams, only a handful came back with errors; mostly diagrams living in personal spaces, for reasons unrelated to the migration itself.

Perhaps most tellingly: after the migration, there were zero support tickets, and zero complaints from users about not recognizing their own diagrams. One engineer on the team put it:

“People keep using draw.io without any training — it seems intuitive and natural.”

In the four weeks following the migration, they can see that over 100 diagrams have already been actively edited in draw.io, often colleagues making minor adjustments post-migration; a sign that users are able to get straight back into diagramming without any disruption.

Solving a migration puzzle with a little help from AI

The team also set out to migrate their PlantUML and Mermaid diagrams. They initially explored whether this could be done using a script, raising a support ticket to ask, but were told this wasn’t a feasible approach.

Rather than migrating these manually one by one, the team’s Confluence administrator spent a couple of days working with GitHub Copilot to build a script-based solution (and succeeded!). This speaks to both the resourcefulness of their internal team and a genuinely useful piece of feedback for any organization facing a similar migration challenge: with the right approach, scripting PlantUML and Mermaid diagram migrations is certainly achievable.

Given the technical nature of much of their diagramming work, including networking diagrams and diagrams representing hardware structures with many small, layered components, diagram fidelity was a key concern going into the migration.

The results were largely positive: diagrams stayed readable and faithful to the originals, with the underlying data preserved. Some connectors and legends required minor manual adjustments afterward, but nothing that stood in the way of teams picking up immediately from where they left off.

When asked what kind of teams and roles use draw.io, the Software Engineering Enablement Lead responded, “All kinds!” — While engineering is the largest group of draw.io users at the company, adoption spans far wider. As the team put it, draw.io is used by “everyone who wants to draw something.”

Interestingly, the team also uses Confluence’s native whiteboard feature alongside draw.io, but draws a clear line between the two: whiteboards work well for general, less technical use cases, while draw.io remains the tool of choice for technical diagramming.

Looking ahead, the team is keeping an eye on features that weren’t available in Data Center, like real-time collaboration and AI-assisted diagramming. While neither was a driver behind the original migration decision, both are welcome developments, though the team says it’s still early to comment on their impact in more detail.

Asked what they’d tell another large enterprise considering a similar move from Gliffy to draw.io, the team’s message was straightforward: do your own comparison, test it against your most complex real diagrams, and don’t be afraid of the scale of the migration itself; with draw.io’s one-click mass migration tool, it may be far more manageable than expected.

Want to dive deeper into the world of draw.io? Access our linktr.ee page to follow us on social media and learn how others use draw.io, as well as pick up some helpful tips and tricks.