I don't think sending out 6000 - 10000 reports during peak hours is practical. We have a dedicated 8 core backgrounder box and we see delays of HOURS with reports in the hundreds.
Just curious, what kind of data source are these reports connected to? MS SQL? Oracle?
Yes we are using Oracle Database.
But i did some digging my self to figure how many cores do i need for 6000 reports, Curious to know how we can calculate the number of cores based on reports.
I was calculating it as "each report" = "each user" since when a request is hit to generate a report at the backgrounder, it is similar to a single user.
Please correct me if my assumption is right or wrong.
So what I did was
Tableau proposed solution : 100 concurrent users - 1 (8 cores) -.5 sec
In my case - 6000 users = 600 concurrent users - 6 (8 Cores) - .5 Sec
As per my requirement we need it in 5 Sec i.e. 6/10 =.6 which is close to 1 (8 Core), This assumption is only for backgrounder , and additionally i need another 8 core for VizQL.
Please do current me if any of my assumptions are close by. Still i am not sure if this is the way to go when calculating the sizing for reports.
I am doing the entire process in off-peak hours.