Costs – Part 10 | Migrating from AAS to PPU
One of the most important aspects when looking to migrate from AAS to PPU is what will costs be. This plays an important part in the decision.
In this blog post instead of doing an AAS and PPU comparison I am going to do this side by side, which I feel will make it easier to compare.
In my opinion it is only fair to compare the same sizing between AAS and PPU, and this can only be done by using the Memory allocation.
For this comparison I am going to compare the following:
- Memory – 100GB
Here are the previous 9 blog posts that I have completed in the series.
2 – Scalability – Part 2 Migrating AAS to PPU – Reporting/Analytics Made easy with FourMoo and Power BI
Comparison
When I started with this comparison it starts to get a little more complicated.
The reason is that for AAS it has a fixed monthly cost no matter how many users use AAS.
I compare this to PPU, where it has a fixed monthly cost PER USER.
Looking into this how can I compare APPLES WITH APPLES because the costs are different!
Here is the comparison as is below
What I did here is to get the monthly cost for an AAS S4 instance
To give a fair comparison, I do feel that it would be best to compare when there are at least 100 users.
As shown above the cost for PPU is significantly cheaper than AAS.
There are a whole host of other considerations to think about when looking to get an accurate comparison which would potentially be the following:
- How many users are going to be accessing the cube/dataset?
- Are the queries intensive or consume a lot of memory?
- What if you need more than 100 GB of Memory (which can happen, but very often the dataset can be optimized)?
-
Features available in PPU which are not currently available in AAS
Summary
In this blog post I have shown what the costs are between AAS and PPU and tried to compare them in a way that makes sense for your situation.
Once again it appears to me that based on the costs it would certainly be feasible to migrate from AAS to PPU.
Thanks for reading and comments or suggestions are most welcome.
As with the current series I once again do think in my opinion that it is extremely attractive to move to PPU.
Any questions or comments are always welcome in the section below.
Thanks for reading!
Great post! I am really glad you took the time to do this.
Another really nice thing that you get with PPU that you do not get when using AAS is deployment pipelines. I don’t know how you put a price on that, but I really enjoy using deployment pipelines when I can.
Hi Ginger,
Thanks for the comment!
And yes you are 100% correct as how would I put a costing on the Deployment pipelines, which make deployments significantly easier!
Thank you, great series of posts! In that same boat, looking to move off AzAS to PBI… And this non-supported option (we’ve got problems with it) for Analyze in Excel should also be possible (supported) when moving over…
https://docs.microsoft.com/en-us/power-bi/collaborate-share/service-analyze-in-excel
”
Analyze in Excel is especially useful for Power BI datasets and reports that connect to the following data sources:
Azure Analysis Services tabular data models and SQL Server Analysis Services (SSAS) tabular or multidimensional data models
Connection to datasets from live connection to Analysis Services (Azure and SQL Server) are not currently supported
“
Hi Nico, thanks for the comment.
I am not sure I follow exactly what you mean?
Do you mean to say that you cannot connect from Excel to your PPU dataset?
I certainly know that this is possible, you would need to change the connection string.
So the limitation is on Azure Analysis Services as per the link in my original post. You cannot use the Analyze in Excel feature inside of the Power BI Service, neither can you connect to an Azure Analysis Services instance from Excel directly, using the “Live Connection”-option. It’s not greyed out, but we’ve had a mixed-bag of experiences with this – only works for some users. I believe with PPU dataset there is no limitation, either using Analyze in Excel or connecting directly from Excel – so another + point for PPU dataset over AzAS. 🙂
Hi Nico,
Thanks for the clarification. And yes you cannot connect from the Power BI Service with Excel to an AAS dataset as it stands today.
For that PPU would be your only option.
Hopefully that will change in the future!
Enjoyed the post throughly, and got great information. But is there a way we can compare AAS VS PREMIUM PBI CAPACITY? we are planning to migrate S9v2 sized bunch of models.. and we have P3 and couple of P1s. Any advise how does it look? And yes we have very large user base.
Hi Jay,
Thanks for the question.
This is not a simple question to answer and would need a longer explanation with more time, but here is a quick overview.
With AAS S9v2 to you get 1280 QPU (Roughly 64 Virtual Cores) and 400GB RAM
https://azure.microsoft.com/en-us/pricing/details/analysis-services/
With Power BI Premium P3 you get 16 backend virtual cores and 100GB RAM
https://docs.microsoft.com/en-us/power-bi/admin/service-premium-what-is#capacity-nodes
Based on the above you will be having a lot less cores which will affect query and processing performance.
If you have a good model with good DAX measures this might not be a big change.
And as explained above it will go from 400GB of RAM to 100GB RAM
When I moved from AAS to PPU I did a lot of testing to ensure that I would get similar performance on my reports, initially from the DAX measures and then later from the user base. I would strongly suggest spinning up a P3 creating a like for like dataset and then do the testing to ensure that you are going to get the same performance?