Well, I can say that for me this is a typical experience working with Hadoop based data sources and, for the most part, Tableau is not to blame.
Have you already done any investigations? What's on the viz? How much data are you trying to pull?
As stated, your question is way to broad for people here to answer.
When i searched about it they say that if we have more than 6 worksheet on a dashboard then we can have latency, but they haven't provided a way to over come this problem.
when we connect to excel sheet as data source we aren't facing any latency but when we connect to hadoop we are facing this problem.
We have joined multiple tables in hadoop cluster and we are trying to fetch 52 lakh records and dashboard contains about 20 worksheets with sheet swap functionality. The visualization that we are showing consist of just data tables, few graphs and images as shapes in tableau.