[MUSIC] Hi, and welcome to Oracle University's oracle cloud infrastructure course on the autonomous database. My name is Kiana McDaniel and I'm a solution engineer here at Oracle. In this module we will talk about business models with the Oracle Autonomous Database. Let's get started. Data analysts typically don't work directly against tables in a database. It's rare for the semantics of a data model to be apparent from the physical data structure itself. Instead, data analysts work against a semantic model, this is a layer above the physical data structure that defines its business significance. But not all business models were created equal. Most popular self service analytic tools define the business model in the tool itself. A common problem with this is that, different analysts each with their own self service tool, can easily define different and contradictory business models, even on the very same data set. Oracle's approach is to push the business model into the database layer. There it only needs to be defined once, which in itself is a great productivity boost. Most importantly this promotes consistency, by sharing this common business model, all analysts get a consistent view of the business. A further advantage of defining business models in the database is performance. A key insight is that business analysts typically access data at the top level. For example, rather than an individual customers purchase of a particular movie last Tuesday evening, the data analysts might be more interested in consumption of a movie genre, by a demographic segment of customers in North America last April. By automatically recognizing the hierarchy and defining it in the database, autonomous database can automatically compute and store these top level aggregates, we call this materialization of aggregate caches. Because autonomous database knows about the hierarchy, and knows about the existence of the aggregate cache, it can transparently rewrite queries to access the precomputed aggregates, rather than having to compute them on the fliers. The result in exceptional performance, even with huge data sets, even with federated data sources, it all happens transparently behind the scenes. All the end user does is to browse and access the data, just as if it were stored locally in the autonomous database, and just as if it were all stored at the lowest grain. Here's a screenshot of the business model tool and autonomous database. We've made it simple to build sophisticated models on your data, by identifying dimensions, hierarchies and measures. With a nice clean way of saying how to aggregate, some average or whatever you choose. Here we say that we want to create model for table movie sales 2020Q2. The tool is now inspecting the schema, trying to identify potential dimension tables that can join to the fact table, and also identifying any hierarchies and measures in this data set. It's come up with a potential star schema design that looks quite promising. Now you see why the analyst wanted to save these data files for reuse and this quarters analysis. Let's expand each of these candidate dimensions. For each we see its columns, and for the purposes of this demo, we kept these tables simple. Hovering over the relationships, we see the candidate join conditions these all look good. Now let's look at the candidate hierarchies. The first one comes from table countries. This hierarchy would be better labelled as geography, so let's override the default name. The tool has correctly identified the hierarchy of country within continent, so this looks good. Notice that the tool has detected a hierarchy from devices to form factor. We'll see how this is used in just a minute. Next we look at this one with a rather odd name day num USA. I want this to be a day dimension. So again, I'll override this default name, I'll remove the day name USA level by selecting that level, and pressing the minus button above the hierarchy list. What I want to do here is to sort days by the day number of the week, rather than alphabetically. Similarly for the names hierarchy, I'll provide a more appropriate name, caption and description. I'll remove the level for month name, and then for the month level, I'll change this to sort by month name. Now we'll go on to measures and immediately we see that two good candidates have been identified, sales and purchases. For each of these we can specify the aggregation expression to be used, which could be sum, average, count, etc. In this case, the default of sum is appropriate for both measures. Okay, now let's create this business model. Here on the card for the newly created business model, we can click on the three vertical dots for more details. I personally like to show DDL, and take my time to scroll through all of it, and think how glad I am that I didn't have to type all of that in Maybe it's more interesting to do a quick and dirty analysis of his business model. Here we can build a simple pivot table for example, viewing across customer segments, the various form factors being used to watch movies. Under the measures tab I'll select purchases rather than sales, and here we see the data. Notice that it's been aggregated from specific devices to form factors, I'll stretches this table out a little so that we can see a bit more data. Briefly, we have some additional navigation options here, to take us directly to the catalogue or insights tool. Again, we have these options here from the hamburger menu, in this case though, we'll return to the database options page. To wrap up in this module, you saw how the Oracle autonomous database business model tool can automatically detect star schemers, enhance performance through hierarchies that are enabled by automatic aggregate caches, and how you can view and browse data for analysis.
