I am in the process of creating a data warehouse using SQL Server for my company. I have created a POC with a simple SQL Relational database with a few(about 10) stored Procs for the ETL process. Now that I have business buy in to step it up to a production state, where I will be tripling the data, it's the right time to ensure the technology I use is best suited for our needs and is aligned with general best practices.
We will be processing about 500k records per day, with a maximum guess at about a million a day. The POC then aggregated all this data down into 5-minute chunks per client per day for the last 4 years. Total being a bit over 500k time slots, for each fact (POC had 4 fact tables) for each customer (Lets plan for 25). So for the POC, we are looking at approximately 50 million rows across all the facts by client and time. That leaves final views of the data at minor amounts, given that the last 6 months satisfies 85% of all questions, that means we should be sitting with result sets of approx 250k rows for 85% of the time if not less. Given that I expect to triple data, let's call it a maximum of 1 Million rows to satisfy 85% of queries, split across about 8 to 10 fact tables with about 4 or 5 dimension tables.
With that all said and done, as mentioned, I am in a very fortunate position where I can now choose the technology that is used for the backend of the warehouse. I have never really had the opportunity to really get entangled in SSAS and feel like its a technology I am lacking. Is SSAS still a popular choice for Data Warehouses or has its popularity declined over the past few years? Is it still used as much as it was 5 years ago?
SHORT VERSION: Is SSAS still something worth learning/investing time and effort into?
We are currently on SQL 2016 and this will be built on SQL Server 2017
Thank you for all your input.