Is a cloud data source better than on-prem?

[chrisrdba]

Greetings. Apologies if this question comes through twice. I hit the Post button and it appeard succesful but I can't find the first one, so here goes....

Total PBI noob here so please bear with my ignorance. 

We use PBI in the could, as opposed to the on-prem version. My understanding is that we do not have Premium version, so we have shared capacity and share resources with other customers, which is leading to my question...

We currently do data transformations on-prem, then upload the data to an Azure SQL DB (NOT managed instance), which is the source of the PBI stuff. Now some people are question if we should just leave the data on-prem, and have PBI point there. 

So my question is if we changed where PBI points to in order to consume it's data on-prem would we be at risk of performance issues during the consumption? I know lots of folks point back to on-prem for data consumtion without issue, but what has me worried is doing this while not being on Premium. It seems to me that if we're drawing from a common resource pool we're at risk of occasional performance issues. If issues arise during our consumption time, would those issues be somewhat mitigated if data was already in the cloud, as opposed to needing to come back to on-prem to consume?

Now keep in mind this is all just a crazy theory from a noob, but I'd be the first guy people are going to ping if my crazy theory proves to be correct. 

Any insight on how to prove/ disprove this theory would be appreciated, thanks!

[Anand24]

Hi @chrisrdba ,

I'm assuming you mean DBs like SQL Server, Teradata, etc. when you say on-prem and Cloud DBs like AWS Redshift, Azure DB, Snowflake, etc. when you say Cloud. 

As an overview of connectivity, it's always better to have the data on cloud instead of on-prem. It's the modern connectivity method and is more productive than on-prem. I've worked a lot on Snowflake and Teradata and it's a no-brainer for me to opt Snowflake(Cloud DB) over TD(On-Prem) due to the ease of use and faster dataset refresh time. 

Another important point to note with on-prem is the tediosness one might face in setting data gateways for refreshing on-prem data in Power BI Services and the various dependencies on it like availability of gateway, are gateway clusters required, server/system availability, etc.

Give a Thumbs Up if this post helped you in any way and Mark This Post as Solution if it solved your query !!! Proud To Be a Super User !!! LinkedIn

[chrisrdba]

Thanks @Anand24 -- yes that's what I meant by on-prem and cloud. However there are some distinctions -- our cloud DB is a Azure SQL DB, which has some limitations compared to a real SQL DB on-prem. If we were to upgrade this to a managed instance those limitations would be gone, but that's not about to happen. 

It should also be noted that we have another environment that does point to on-prem for it's refreshes and they run without issue, but that's on a Premium license. That said the gateway thing wouldn't be an issue for us. It was done once, it could be done again I'm sure. 

All said, it would be tough for me to argue that we should keep the source data in the cloud, unless this goofy shared capacity/ resource license thing could bite us from a performance standpoint. I'd like to see it stay there, but need something concrete for a reason.