Every report has a main dataset defined during its creation, that is the one used in all the previous tutorials. But some times we need fileds that are not returned by the query of the main dataset or the fields needed could also be in a different datasource. To do this we can use the element "Dataset", this element permit to define many dataset inside the report, each one with its fields and datasource. Obviously every dataset is indipendent from the others, so its fields are separated from the ones of the main dataset, and also from the ones of the other datasets. With a report opened, to create a dataset right click on the report root node in the outline view and choose "Create Dataset":

Now a new wizard will be opened, in the first step you can select the name of the new dataset and if it use a connection to a datasource or it is simply empty. Since we want to read data from the databese select the first option and use as name "GroupData", as in the image below, then hit Next.

In the next step we have to select the data adapters of the new dataset, and since we use the same database of the main report even this time we will select "Sample DB - Database JBDC Connection". We must also provide a query to retrieve the fields and this is "select count(*), SHIPCITY from orders group by SHIPCITY". It returns the records grouped by the shipment city, in this way we have that every record read from the database with this query will contain two fields: the number of element for a city and the name of the city. After that we can click Next.

At this point will be shown the fields discovered using the provided query and we can choose which add to the report. Since we need them all hit the button to add all and then Next, as in the following image:

At this point we can define groups, but we don't need them so hit the Finish button. The dataset will be created and is visible from the outline view, if you explore its element you can see that it has its own fields, variables and groups.