Aggregate over multiple columns in Azure (Kusto Query Language)

Question

I have clickstream data in Azure monitor logs in this format:

Category   StepName     Count_    Median_Duration(secs)
   A         step1       1200        00:00
   A         step2       1000        24:00
   A         step3        800        19:00
   B         step1       4000        00:00
   B         step2       3800        37:00

I need to pivot the table to get this:

Category Step1_Count Step1_Duration Step2_Count Step2_Duration Step3_Count ...
   A       1200          00:00         1000        24:00          800      ...
   B       4000          00:00         3800        37:00           0       ...

Right now I am only able to aggregate over one column using evaluate pivot(StepName, sum(Count_)) or evaluate pivot(StepName, sum(Median_Duration)). Is it possible to get the above format without using joins?

Note: Similar formats to the output table are fine, just need the aggregate of the count and duration.

Answer 1

you could try something along the following lines:

datatable(Category:string, StepName:string, Count_:long, Median_Duration:timespan)
[
   "A", "step1", 1200, time(00:00:00),
   "A", "step2", 1000, time(00:00:24),
   "A", "step3",  800, time(00:00:19),
   "B", "step1", 4000, time(00:00:00),
   "B", "step2", 3800, time(00:00:37),
]
| summarize StepCount = sum(Count_), Duration = avg(Median_Duration) by Category, StepName
| project Category, p = pack(strcat(StepName, "_Count"), StepCount, strcat(StepName, "_Duration"), Duration)
| summarize b = make_bag(p) by Category
| evaluate bag_unpack(b)

or, if you're ok with a different output schema:

datatable(Category:string, StepName:string, Count_:long, Median_Duration:timespan)
[
   "A", "step1", 1200, time(00:00:00),
   "A", "step2", 1000, time(00:00:24),
   "A", "step3",  800, time(00:00:19),
   "B", "step1", 4000, time(00:00:00),
   "B", "step2", 3800, time(00:00:37),
]
| summarize StepCount = sum(Count_), Duration = avg(Median_Duration) by Category, StepName