PowerBI + HP Vertica questions

[AntoineCh]

Hi all,

I'm currently working for a client trying to figure out if a switch from their current BI tool to PowerBI is relevant. Their analytics database runs on-premise with HP Vertica.

I understand that a build-in connector in PBI is not yet on the table for the dev team. On the other end, HP provides a step-by-step method to connect, which is what I will try next.

Is anyone knowledgeable on the matter ? Specifically :

1. How long does it take to create the connection? Is there a way to mass install across many computers?
2. Is it stable? Does it affect the available features of PowerBI in any way?
3. My understanding is that the tables are downloaded and the model recreated when using the desktop version; the calculations are thus done on one's computer. When uploaded online on the Pro version, are the calculations (both PowerQuery steps and PowerPivot measures) done on an Azure-located Vertipack or converted into SQL and fed through Vertica?
4. Is the performance of this connection good? Two aspects : A/ the client would be transitionning from SQL calculations on 30+ millions flat tables to a tabular model. I expect this is going to improve things quite a bit. Is there any reason why it wouldn't? B/ Performance of the proposed connection vs native PowerBI connections. Given the result of question 3, should I expect some trouble there?

Those would be my main concerns at this stage. I've had a first few very interesting talks with people a few days ago during a big data convention, but i'd like to know what's the word on this.

Best,

Antoine

[v-yuezhe-msft · Accepted solution]

Hi @AntoineCh,

Firstly, for the time that it takes to create DSN to HP Vertica, I would recommend you contact Vertica support. Once you create DSN, it would simple to connect to the DSN from Power BI Desktop. However, it depends on the amount of your data that how long it will take to completely load the data from DSN to Desktop. And do you want to mass install Power BI Desktop across many computers? If that is the case, you can consider to deploy Power BI Desktop via SCCM, for more details, please review this blog: https://blogs.msdn.microsoft.com/samlester/2016/09/30/installing-power-bi-desktop-through-sccm/ .
 
Secondly, it is stable, it will not affect the available features of Power BI Desktop after you connect to DSN.

Thirdly, after you import data from Vertica to Power BI Desktop, all the data are stored in Power BI Desktop file, and calculations are based on the imported data. Once you publish Power BI Desktop file to Power BI Service, Power BI manages the storage of data and metadata using Azure BLOB and Azure SQL Database.

In addition, if you have a large amount of data to import to Power BI Desktop, it may take you some time to load data to Desktop, from Power BI side, to improve performance, import data by entering SQL query to include the tables you want. And also contact Vertica support to improve Vertica ODBC performance.

Thanks,
Lydia Zhang

[AntoineCh]

Thanks Lydia, that's very helpful. I may have other questions later, but this covers it for now !

The volume should be manageable (couple Go at best), especially if pre-filtered on the appropriate time-frame.

Best,

Antoine