Hi Folks,
I think that during a Managed-to-SaaS migration, one of the most time-consuming and challenging tasks is migrating the dashboards.
On the one hand, there can be a huge number of dashboards at clients. On the other hand, they may contain many features and data that are not directly "migrateable", such as Management Zone (MZ) filters, entity IDs, relationship filters, and so on.
In the "brand new" latest SaaS tenants, many Managed/Classic functionalities are no longer available. One simple example is Management Zones.
That is why many dashboards need to be redesigned during a Managed-to-SaaS migration.
I know that observability is not only about dashboards, but the observability maturity level of many organizations is still relatively low. As a result, their operations teams often rely heavily on dashboards.
And yes, I can already hear the voices saying that in the era of AI, you can do "anything".
So I decided to run a small test, mainly using Dynatrace tools:
- DT Assist
- dtctl + skills
- Dynatrace MCP
(+Visual Studio Code + GitHub Copilot)
I was lucky because we have a "not brand new" SaaS tenant where Classic functionalities are still available.
I created a simple TEST dashboard using four telemetry types:
- Host CPU
- Host memory
- Request count
- Response time
I used graph and table visualizations, including max and last values.
Then I tried the embedded dashboard migration functionality.
The graph visualizations were migrated correctly, but only one of the tables was migrated correctly: the request count table.
I then asked DT Assist to correct the CPU table's DQL query.
Once it was ready, I copied the corrected query back into the CPU dashboard tile.
At that point, I had an idea:
Why not use dtctl + Copilot to correct the whole dashboard?
So I jumped into Visual Studio Code and asked GitHub Copilot to correct the entire dashboard based on the already corrected CPU dashboard tile. At this point, the Memory and Response Time tiles were still incorrect.
With the help of the dtctl skills, I ended up with a new, corrected dashboard, while still keeping the original dashboard with the incorrect tables.
Another interesting point: although DT Assist supports English, I completed the entire process with DT Assist and GitHub Copilot in Visual Studio Code using my local language, without any issues.
I think there are many possible ways to migrate Managed dashboards to SaaS.
This is definitely not the only one...
Let's call this my "boomer-style" solution.
However, this approach only works in a "hybrid" SaaS environment where Classic functionalities are still available.
In a "brand new" latest SaaS environment with very limited Classic functionality, you need to find another approach because you cannot import a Classic dashboard into it, either directly via API or with the Upgrade Assistant.
I am very keen to see the new toolset that will support the Managed-to-SaaS migration process:
AI-driven upgrade toolkit for migration to latest Dynatrace
What is your thoughts about this topic?
I plan another post about the dashboard topic it will come soon… with the title „Grafana killer or not”…😉
I hope it helps.
Best regards,
János
Featured Posts
