New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[APM] Implement a dedicated tab for custom dashboards #163590
Comments
Pinging @elastic/apm-ui (Team:APM) |
We've started some explorations of how to improve the interaction and structure of the Dashboards with @sqren You can find them in the Figma link |
As discussed, the scope for 8.11 will be an MVP that adds a new tab and a page that can be used to link dashboards for each service. |
@boriskirov @akhileshpok are there any reasons why users need to "Link" dashboards instead of selecting the dashboard from a list? This is a very different UX pattern than the one we use elsewhere and it introduces one extra step for selecting a dashboard. |
@teresaalvarezsoler Imagine that there are 100s of dashboards in a cluster. We only want to list the relevant dashboards in APM. I'm not attached to the terminology of "linking" but we need a way to select a subset of the dashboards available, and list them in APM. Additionally, when a dashboard has been selected (linked) we specify which service(s) it should be available for. Some dashboards should show up for all services, some should only show up for java services, and some should only show up for a single service. The user specifies this in the "linking" step. Again, open to changing the terminology but some kind of step is needed. Lastly: If the visualisations in the dashboard uses LMK if you want to chat more about this and we can schedule a call. |
Thanks for the explanation @sqren, but I'm still not sure why we need the "Link" button, can you just show the list of available dashboards in the dropdown (highlighted below)? You can apply all the filters you mentioned and only show the relevant dashboards list there. You will get a cleaner UI and save users an extra step. |
@teresaalvarezsoler We will miss the persistence logic here if the dashboard is not linked. As a user, for this particular service, whenever i land on this page, i would want Dashboard A to be present and for another Service Dashboard B. With linking we can persist this view when the user lands here. Else we will have to always default to an empty screen unless the user has selected a dashboard from the dropdown or a default dashboard which may not make much sense in terms of the service. |
Selecting a dashboard could work as "linking" it behind the scenes, the latest selected dashboard will be displayed next time the user enters this view. The only caveat I can think of is the dropdown selector to choose the dashboard would always need to be visible but that's much better than unlinking/linking in my opinion. |
Are you suggesting to show all dashboards in the list? Again, this will provide a bad UX for people not familiar with the stack and just want to view the small handful of relevant dashboards for the selected service. |
How will can we "link" behind the scenes? What will determine which filters should apply? Again, the end user might not know these details so it is important that someone can set this up beforehand. |
I'm proposing to show only the relevant dashboards in the first dropdown |
How will we determine which dashboards are relevant for the service, if they haven't been linked? |
@teresaalvarezsoler I think i understand the confusion here. Step 1. - COE or equivalent will go through the list of Step 2. - Now the SRE's already have a Curated list of So most of the time, the SREs will only see 1 dropdown with the curated list of relevant dashboards (The step 2). The step 1 is something we don't expect to happen every time. Step 1 will be performed very few times by COE initially when the service is setup or in case new dashboards have been added. So this is not always a 2 step process. cc: @sqren |
Thanks for the explanation @achyutjhunjhunwala. It definitely makes more sense when there are two personas involved in the process. I think there are several assumptions in this design:
I'm not very knowledgable about your target users, but I guess you are fine with these assumptions. Maybe testing with users is a good idea (if you haven't done so) since it implies a very different pattern of how users find their dashboards elsewhere. |
I think that is a good list of assumption, and I think we are aware of most of them. I agree that it will not be a perfect fit for everyone (clusters where every dashboard is an APM-compatible dashboard for instance) but in the majority of cases I think it makes sense.
There doesn't necessarily need to be two personas involved. Even with just a single user it makes a lot of sense to reduce a long list of 100s of dashboards to just the 1 or 2 relevant ones.
I can't say for sure that all users will understand this up front. I think we are releasing this under a beta badge. We'll iterate on it as we get feedback. |
My 2c: wider dropdown menu of dashboards (especially if they have already a common prefix like the default dashboards from beats) and why don't have the option of selecting (or just filtering) dashboards by tags (either in the dropdown menu or globally in a Settings page)? |
A dedicated tab should be available for custom dashboards. This tab should also have the time picker and the search bar. This functionality will leverage the recent work on custom dashboards in #163504
Figma link
Tasks
The text was updated successfully, but these errors were encountered: