How to do it…

  1. We are using a live connection, which means that each interaction with a report will generate an SQL query and we should wait. To avoid this, we will pause Tableau Auto Updates, as shown in the following screenshot:

  1. Now, we can craft our report by dragging and dropping Dimensions and Measures to the canvas, as follows:

  1. You should add the following filters:

Moreover, we should convert D Year into discrete and sort by Revenue.

  1. We should unpause auto updates and run our query. It takes ~18 seconds for my cluster. If we missed the time, we can log in to the AWS Console and navigate to the Redshift cluster | Queries Tab and see all executed queries, their time, and plan. You might see that, in the case of Tableau, we don't have actual queries; instead, we have something like fetch 10000 in SQL_CUR7. Tableau is using cursors and we can run the following query to see queries for currently active cursors:
SELECT
   usr.usename AS username
  , min(cur.starttime) AS start_time
  , DATEDIFF(second, min(cur.starttime), getdate()) AS run_time
  , min(cur.row_count) AS row_count
  , min(cur.fetched_rows) AS fetched_rows
  , listagg(util_text.text)
    WITHIN GROUP (ORDER BY sequence) AS query
FROM STV_ACTIVE_CURSORS cur
  JOIN stl_utilitytext util_text
    ON cur.pid = util_text.pid AND cur.xid = util_text.xid
  JOIN pg_user usr
    ON usr.usesysid = cur.userid
GROUP BY usr.usename, util_text.xid;
  1. Let's run one more system query that will help us to identify the size of tables we are using in our Tableau data source, as follows:
select stv_tbl_perm.name as table, count(*) as mb from stv_blocklist, stv_tbl_perm
where stv_blocklist.tbl = stv_tbl_perm.id
and stv_blocklist.slice = stv_tbl_perm.slice and stv_tbl_perm.name in ('lineorder','part','dwdate','supplier')
group by stv_tbl_perm.name
order by 1 asc;

We can notice the size of the tables:

Finally, we can tune our cluster to improve the performance of the queries.