We are using Kibana/Elasticsearch /eck managed for our dev/production (non customer facing UIs) to analyse many things.
Sometime back kibana depricated beta version / non-guaranteed (non production ready) versions of Input Controls. Now Kibana advocates such that when theme is dark, it is very difficult to see data in the old input control values( background/foreground of input control values being in white color). Indirectly forcing to switch to new controls.
We abided by shifting our visualisations to new version. They work neat w.r.t. functional correctness. Thanks.
In the old controls:
We used to read data from a data set, let's call 'master data for meta information' into Input Controls.
Advantage of this approach of 'master data for meta information':
The data set size is small. Unique values / dependent values etc...selections make the load on Elasticsearch to certain extent optimum. We could reduce load on elasticsearch considerably.
Visualisation data: Each row/entry of visualisable data set (let's call business data) was enriched with the 'master data for meta information'.
This enrichment enabled us for filtering much easier.
Usecase: When/how many cold/warm 'create contact' user interaction, time taken, timestamp per tenant in DC USEast/ affinity if any /provider AWS / customer during last week
In the above example: tenant , dc, provider / customer --> are master data
Business data is cold/warm , create/delete/update, tenant , contact , end user id of the end user. As you can imagine, enriched data is combination of above two.
We published visualisation for 'warm' end user interactions by default (meaning filter will be interactionType=warm)
This worked beautifully with old input control (having two data sets - business data set for graphic visualisations and master-data set for input).
Now when we shifted to new input controls:
Since, interactiontype field is not available in 'meta data' data set, nothing appears in new input controls.
In old deprecated input control, the global filter was not applied input filters, there by values coming in.
So, we shifted to business data set (for new control to have unique meta data), we can see marked increase in resource utilization on Elasticsearch and wait time.
Question to you:
In such condition, do you suggest any trick?.. Or will you consider having this feature request of ignoring global filter?