In nearly 90% of these cases, the customer chose Tabular because of its effectiveness and simplicity. Often, when customers approached us asking for an engine, we developed both Multidimensional and Tabular proof of concepts. Other common issues are DISTINCT COUNT measures, many-to-many relationships, and leaf-level calculations: all of these are features where obtaining good performance is a challenge. Not to say that Multidimensional is slow, but creating an efficient Multidimensional database requires a major effort: it is hard to automate and it is complex to optimize when an application automatically generates the model. These companies usually considered Multidimensional, but chose not to adopt it for two main reasons: MDX complexity (from the developers’ point of view) and performance. Companies contacted us when they were looking for a solution to a specific problem-specifically, looking for an engine for their product or service with good performance and low maintenance cost. Until today, we have never been proactive in promoting Tabular for such scenarios. In other scenarios that we have helped to build, companies use a mix of both approaches. Others provide a multi-tenant service, creating databases on demand, and then offering to access them through standard clients (such as Excel, Reporting Services, and other third-party tools). The usage may vary: some companies created a single large database deployed manually and queried by their front end (this made it necessary to write their own DAX query builder). In other words, users do not even know that Tabular is powering their reports. Instead, they integrated Tabular features into their existing software as a back-end server for their analytical needs. I have seen many companies adopt Tabular as the analytical engine for their product or service without actually creating a “BI project” in a canonical way (gather requirements, design a prototype, get feedback, improve the model, and loop until it is done). My guess is that Microsoft was surprised by it as well. However, in this adoption process, something surprising happened. Thus, users have adopted Tabular mainly for new projects-especially in companies that had not used Multidimensional before. This is expected and very natural, as Multidimensional is complementary to Tabular and is not going to disappear. Existing customers of SSAS Multidimensional (Multidimensional hereinafter) have many reasons for not adopting Tabular: missing features, skills shortage, lack of tools, legacy (!) OLAP ecosystems. In these early years of SQL Server Analysis Services Tabular (SSAS Tabular, or Tabular hereinafter) adoption, SQLBI has helped several customers in their first implementation using the new xVelocity in-memory engine and the DAX language. You can download this article in PDF format at the end of this page. This article provides several reasons why Tabular could be the right choice for the analytical engine embedded in a service or an application. When the solution serves other software instead of a human user, the challenges are completely different. The point of view is unlike that of creating a “classic” Business Intelligence solution. Other scenarios use a mix of these two approaches. Others provide a multi-tenant service-creating databases on demand, and then accessing them through standard clients. Some have created a single large database deployed manually and queried by their front end. Many companies have adopted SQL Server Analysis Services Tabular as the analytical engine for their product or service.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |