Support for multidimensional models on Microsoft Azure Analysis Services

We are a small BI team that would love to move our AS to the cloud. At this point our data is small enough that we are using SQL Server standard. There needs to be a much lower production pricing tier for small companies like ours. These prices simply do not make sense for our size.

Allow changing of pricing tier for an Azure AS server. For example, scale down from an S4 (100GB) to an S1 (25GB).

Need to be able to do this in an automated way using an ARM request, so can automate scale/down up to match peak usage user demands.

Currently understanding what SSAS is doing for a particular specific query is difficult and not that transparent. Having a visual representation of the steps taken by the engine to solve a query (like SQL Server Relational engine has had since SQL 7) would improve manageability

Provide an easy way to set up scale out of Azure AS servers. Need to allow easy set up of load balancing and synchronization of data across servers. May need to copy data from a processing server to multiple front-end servers.

Support for ragged hierarchy display. Similar to hide member if that is in multi dim.

One of the nicest features of on premise SSAS Tabular is the ability to ugrapde a Power Pivot model from Excel directly into SSAS. From what I've seen so far, this isn't easy in Azure since there isn't direct access to the server and file structure and normally the SSAS service account needs permissions to access the Power Pivot file's location.

Allow backup and restore of Azure AS servers using a user specified Azure storage account.

The DMVs for SSAS Tabular (Azure and SQL 2016) are not documented anywhere. While the meaning of many of the fields is obvious, there are a few that are just IDs for which it would be nice to see all possible values and descriptions. It would make sense to add the definitions here: https://msdn.microsoft.com/en-us/library/hh230820.aspx

Support Custom ODBC Drivers for importing data
(ex: Apache Drill, Hive, PostGres etc)

Provide Geo-Redundancy features along the lines of the DR/scale out functionality available for Azure SQL databases

At present when I dissect a Power BI Desktop model it is compatibility mode of 1400, when I look at AAS model it is still only 1200. It would help if the two were in sync and cross compatible.

Instead of making Azure AS available on the internet, allow admins to move it so it's only available on a VNET.

Support for non-native applications. ie. Support for Java app using Analysis Service PaaS. REST or JDBC support

Pool of SSAS dbs for managing the collective performance of SSAS cubes and reducing costs. Much like Azure SQL db's elastic database pool feature.

It sounds like we need a new data source type in SSRS for Azure Analysis Services. While I think you can use an AAD service account username/password in the connection string, the current EffectiveUserName implementation will fail because it will say EffectiveUserName=DOMAIN\username rather than EffectiveUserName=[email protected]. I'm hoping that an Azure AS connection type in SSRS could use the UPN for EffectiveUserName or allow some sort of user level UPN override property or user UPN mapping.

This only being available in West Europe causes a bit of an issue. Data Factory is only available in North Europe hence I cannot have a modern data platform that uses DF to move data around and use Azure SSAS without moving data between data centres, are there plans to make this available in North Europe as well as the current set up does not really make a lot of sense

xEvents are tremendously helpful for monitoring and troubleshooting performance issues.

Ideally we'd get both "Event Stream" and "Event File" targets as there are good use cases for both.