I have the following DQL query that calculates the p50 response time per service:
timeseries
p50 = percentile(dt.service.request.response_time, 50, default:0),
interval:10m, nonempty:true,
by: { dt.request_attribute_Key },
filter: matchesValue(`dt.entity.service`, "x"),
from: now()-1d,
to: now()
| fieldsAdd ServiceName = entityName(dt.entity.service)
Now I’d like to add a dimension based on a request attribute (for example request_attribute.Key or request_attribute.Type) — so I can see the percentile (p50) per service and per request attribute value.
Is it possible to define that kind of dimension in the timeseries clause, or do I need to expand the attribute first?
If so, what would be the correct DQL syntax?
