Thanks Sterling,

So what is the purpose of registering a data source?

Is it basically so that a naive user who is building their own report (e.g. through WRS) can have a data source named "Sales" they can select from a dropdown list of "available" data sources, without knowing/caring whether that data source is a SQL server database, SAS dataset, or a view created from some combination of sources?

So as long as I'm writing my own SAS code, I shouldn't have to worry much about registering data? So as a SAS programmer, the fact that the admins have already registered a bunch of libnames pointing to various databases is just a convenience, not a necessity?

Thanks, --Q.

On Wed, 28 Mar 2012 08:35:13 -0700, Sterling Paramore <gnilrets@GMAIL.COM> wrote:

>To answer your main question: If it's all running through a stored >process, and the stored process has permissions to access the data >sources, then the data sources do not need to be registered. > >I've never built a stored process to output graphs in a WRS report, >although I've read that it's possible: >http://www.bi-notes.com/2011/11/sas-bi-web-report-studio-dashboard/ > > > >-Sterling > > >On Wed, Mar 28, 2012 at 8:00 AM, Quentin McMullen ><qmcmullen.sas@gmail.com> wrote: >> Hi All, >> >> My (new) group is setting up a SAS server, and will start building BI >> tools. And while I haven't actually gotten my hands on anything yet, >> I'm confused by some of what I'm hearing. >> >> Big picture for my role would be something like I would want to take a >> SAS macro that produces an accrual plot for a clinical trial, and make >> it into a stored procedure that could be accessed via the web (Web >> report studio, information delivery portal, whatever?) and let users >> pass in a few parameters and generate a plot. >> >> I'm confused as to the purpose/need for information maps, data >> integration studio, etc. >> >> Do I need to somehow "register" a dataset or libref with (with >> integration studio, or some metadata server) before I can use it in a >> stored process? (or even use it via EG running on the server?) >> >> Imagine the accrual plot macro reads data from an access database >> sitting on some file server on our network. Suppose it reads it via >> SAS/ACCESS to PC files, so in PC SAS all I need is a libref pointing >> to the access database. >> >> I'm sure I misunderstood, but in one meeting it sounded like in order >> to have a SAS job that runs on the server read from the database, I >> would need to either upload the actual database to the SAS server >> (madness), or would need to go through some process of registering the >> access database as a data source in the metadata server (which seems >> odd). >> >> So I'm confused. Seems to me if my SAS code points to the access >> database (or a SQL server database or a flat file or .....) I don't >> understand why I would need to register the data source with the SAS >> server at all. I can't imagine that by using stored process or EG, I >> lose the ability to simply create my own librefs within a session, or >> write my own code to connect to a SQL database and do pass-through >> queries. Do I? >> >> I'm sure this will become more clear once we have a sandbox set up, >> but since it's bothering me now, was hoping someone might be able to >> straighten me out. >> >> Thanks, >> --Quentin