如何最好地将数据从SQL推送到Power BI？

Question

We are getting increasing requirements from the business (product owners, marketing team) to provide up to date/quick answers on various data points in our system. An example might be that i want to know how many times this question was viewed in full, how many times it was served up in a search, etc. Note: these are 'app-level' stats, not Google Analytics events/tracking.

Currently, we store this in a SQL Server db, and the reporting involves a developer running a SQL query manually.

We'd like to move to a system like Power BI , where it makes sense of the raw data.

From my reading, it sounds like i should:

1. Setup a new SQL DB, let's call it "Reporting"
2. Setup SQL Server Replication between by current DB and the Reporting DB
3. Connect Power BI to my Reporting DB.

Now my questions:

1. Is this a good approach, or is replication overkill? I'm wary of allowing Power BI to connect directly to my live DB, since this is used by various apps/customers, and i want to avoid any potential locking/collisions - hence i'm thinking replication might be a good approach?
2. Should i just push the data straight from my app straight into Power BI (eg via REST call? Cons here is Power BI is now the primary 'owner' of my data
3. Is there something better/more suited than Power BI? (we don't have a huge budget right now - so something like Periscope is out of the question)

Some notes on our tech stack, if it helps:

• .NET stack (SPA frontend, .NET REST API backend)
• Everything running on Azure
• All 'app level' data stored in SQL Server

I realize this question is a little subjective/open-ended, but just looking for some guidance on the best approach.

Thanks in advance!

Answer 1

1. Replication is reasonable, but may be unnecessary. You can use Power BI against SQL in "Import" mode - it then caches the data in a dataset (effectively an SSAS Tabular cube, hosted in the Power BI web service). You can run scheduled refresh through a gateway (or directly if you use Azure SQL). This also lets you model your raw data to suit the requirements.

2. The REST API is still quite limited, and every data requirement/change will need a developer involved. The full product offers a lot more opportunities for user engagement.

3. If your budget is a limit, then Power BI is the obvious choice. You can go a long way with the Free license: https://powerbi.microsoft.com/en-us/pricing/