Bi connector count(*)

Hi,

does anyone know why BI Connector count(*) is so slow?

Any idea on how to make it perform better?

Thanks
Luigi

to give a bit of context:

The collection has 400 millions document each of which is in the form:

{
    "_id": {
        "$oid": "61e420593c48c5ceffced5e7"
    },
    "publisherId": "404637380346545",
    "contentId": "1128711797939096",
    "granularity": "hour",
    "metricTimestamp": {
        "$date": {
            "$numberLong": "1642248000000"
        }
    },
    "sampling": "absolute",
    "contentCreatedTimestamp": {
        "$date": {
            "$numberLong": "1641052801000"
        }
    },
    "name": "post_video_view_time_by_region_id",
    "key": "Florida - United States",
    "value": {
        "$numberInt": "2220757"
    }
}

The operation takes 35 minutes.

Index is on publisherId, contentId and granularity