Forum Discussion
simonphillips
7 years agoNew Member
SSAS Tabular in Azure or PowerBi service for storage?
Hi Folks -
We have a fairly complex model in SSAS Tabular hosted in Azure Analysis Services, about 3Gb in total and serveral thousand measures (about 150 base measures with different time comparison variants).
We have a release process (using Azure DevOps) for this. We're using PowerBi pro, but can't afford Premium. We're mainly using PowerBi reports in the service over the web, with some desktop use mainly for report creation.
I'd like to :
a) keep costs and complexity down
b) enable the use of Q&A, including synonyms
c) ensure we're future proofed with future PowerBi changes
It seems that we can either:
1) Stick with SSAS in Azure, hope that Q&A support (incl synonyms) improves for SSAS models (currently I can't get Q&A to show in dashboards at all, and essentially follow an "enterprise" route, albeit with SSAS being hosted rather than managed ourselves.
2) Migrate our model to the PowerBi service (using Tabular Editor?), Ditch SSAS entirely, and have our reports point to the service instead. This would apparently save a lot of money and mean that Q&A and other future enhancements work out of the box. But it would make our release process more complex (can't just deploy the model like SSAS), and possibly run against some other unseen limits/problems (eg # of measures, refresh interval).
Which route should we take? Right now we seem to be in the worst of both worlds - higher cost, higher complexity from managing SSAS, and some features like Q&A not working or complex (each report has its own dataset despite all being live connections to same SSAS model).
Thanks
Simon
We have a fairly complex model in SSAS Tabular hosted in Azure Analysis Services, about 3Gb in total and serveral thousand measures (about 150 base measures with different time comparison variants).
We have a release process (using Azure DevOps) for this. We're using PowerBi pro, but can't afford Premium. We're mainly using PowerBi reports in the service over the web, with some desktop use mainly for report creation.
I'd like to :
a) keep costs and complexity down
b) enable the use of Q&A, including synonyms
c) ensure we're future proofed with future PowerBi changes
It seems that we can either:
1) Stick with SSAS in Azure, hope that Q&A support (incl synonyms) improves for SSAS models (currently I can't get Q&A to show in dashboards at all, and essentially follow an "enterprise" route, albeit with SSAS being hosted rather than managed ourselves.
2) Migrate our model to the PowerBi service (using Tabular Editor?), Ditch SSAS entirely, and have our reports point to the service instead. This would apparently save a lot of money and mean that Q&A and other future enhancements work out of the box. But it would make our release process more complex (can't just deploy the model like SSAS), and possibly run against some other unseen limits/problems (eg # of measures, refresh interval).
Which route should we take? Right now we seem to be in the worst of both worlds - higher cost, higher complexity from managing SSAS, and some features like Q&A not working or complex (each report has its own dataset despite all being live connections to same SSAS model).
Thanks
Simon
2 Replies
- AnonymousNot applicable
simonphillips Check out some of the roadmap that they've made public here -> https://docs.microsoft.com/en-us/business-applications-release-notes/October18/intelligence-platform/planned-features - There may be opportunities to transition with the XMLA endpoint, but I would hold off a bit before transitioning.
- simonphillipsNew Member
Thanks, that's helpful.
I've been following the roadmap following its publication in the summer; the problem for me is that I can't find a clear declaration about where the balance is intended to be between SSAS (AAS) and the PowerBi Service's equivalent SSAS engine. It feels like we should probably be moving to the latter, but again there's no decent comparison out there that I can find.