Presentation is loading. Please wait.

Presentation is loading. Please wait.

Building a Microsoft BI solution step-by-step

Similar presentations


Presentation on theme: "Building a Microsoft BI solution step-by-step"— Presentation transcript:

1 Building a Microsoft BI solution step-by-step
Farouq Abukhamireh Business Intelligence Architect

2 Agenda What is BI Why build BI solutions - ROI Methodology
Architecture Dimensional modelling Tools and Technologies Cloud vs. On Premises

3 What is Business Intelligence?
Data text /numbers/video or audio Sourced from business processes Data is not understandable by humans Business Intelligence Strategies. Technologies. Software. Architecture. Methodology. Information is a sequence of symbols that can be interpreted as a message. It provides knowledge or insight about a certain matter.

4 Why build BI solutions - ROI
Cost: Benefits: Experienced people. Software. Hardware. Time. Information about the status of the Predictive Analytics Effective inventory management Effective marketing Self-service BI (SSBI) Fast access to information Save on report/dashboard development cost

5 Architecture Presentation Layer: Semantic Layer: EDW: Data Source:
Dashboarding, Operational Reporting, Self Service Reporting and ad-hoc reporting and Analytics Semantic Layer Cloud -external On-Prem - internal ETL Mobile Staging Mirroring OLTP External Semantic Layer: Performance KPIs Hierarchies Measures (business logic) Consistent business friendly report building experience to empower ad-hoc reporting and analytics for business users OLAP In-Memory EDW: Staging Mirroring Data mart(s): Star Schema structure to simplify business reporting logic, improve performance, and provide data feed to the semantic layer. Data Marts (Star-Schema) At the core of the BI solution is the EDW, this is where historical data is stored, the most important part of the EDW is the Data Marts, these are Dimensional data bases that has a Star-Schema structure (Fact tables in the middle surrounded by Dimension tables), the EDW may also contain staging and mirroring data. The EDW get’s it’s data from the data sources, data sources can be operational data like sales data or bank transaction data. Data sources can also be external data like demographic or stock pricing data. Data sources can be in many formats, like flat files, tables or excel files. The Semantic layer comes on top of the EDW layer, it gets it’s data from the Data Marts. Symantec Layer offers many features: Performance : Split second responses to queries enabling analytics and slicing and dicing of data. Abstraction: Semantic layer works as a layer of abstraction to isolate business users from the details and complexity of the underlying data. KPIs: giving a context-sensitive performance scores allowing reporting and dashboarding tools to give decision makers visual feedback about business performance Hierarchies: Allowing users to drill up and down (example is product hierarchy Category-Subcategory-Product) Measures (business logic): context-sensitive aggregate values like totals, averages and counts, could be more complex using business logic. Consistent business friendly report building experience to empower ad-hoc reporting and analytics for business users Presentation Layer: This is the most visible part of the BI solution, it is what the business users, the decision makers see, we can do an excellent job building a BI solution, and the presentation layer should reflect that. Presentation layer is used to deliver Dashboards, Operational Reports, And an effective tool, should empower business users to build their own reports and dashboards with minimal training and this is what is know as Self Service and ad-hoc reporting and Analytics. Data Source: OLTP – Operational Data like sales data External data – Third party (example Stock prices, Demographic) Relational/Data files/Excel. Files

6 Methodology Project Planning Business Requirements Definition
Technical Architecture Dimensional Modeling BI Application Design Product Selection & installation Physical Design BI Application Development ETL Development Deployment Growth Maintenance Project Management Every software project starts with a project plan, project plans define: project goals and objectives. Defining tasks. Quantifying the resources needed. Determining budgets. Defining schedule and timelines for completion. Agile methodology is being adopted across the industry, Agile is a iterative and incremental and it is a good way to use when building a BI solution. Business Requirements Definition: Identify stakeholder reporting needs from the perspective of the questions they are trying to answer or decisions they are trying to make. (example: We in the mortgage loan division want to measure the performance of our mortgage sales reps based on a threshold, we also want to see change in performance over time, we operate in different locations and we would like to see the performance by location.) Prioritize reporting needs based on value and feasibility Progressively elaborate reporting requirements Communicate reporting requirements using examples Technical Architecture This step influences the overall development and implementation decisions. How do we extract the data from the source and load it into the EDW. What data should we keep in the EDW and for how long. How to present the information to business users (depending on number of users, their needs, security, type of users, are they external users or internal users)

7 Demo Power BI

8 Dimensional modelling

9 What is a Dimensional Model
Central Fact Table(s) and it’s associated dimensions. The dimensional model is also called a star schema because it looks like a star with the fact table in the middle and the dimension tables serving as the points on the star The Dimensional Model consists of a normalized fact table with denormalized dimension tables

10 Facts Each fact table contains the measurements associated with a specific business process like: Taking an order Handling a customer support request A record in a fact table is a measurement A measurement event can always produce a fact table record The Measurement Event have a numeric measurement that quantify the magnitude of the event. Quantity Sales amount Call duration Fact tables are highly normalized Fact tables are the largest tables in the data warehouse database.

11 Dimensions Dimensions describe the objects of the business:
Products Stores Employees Dimensions are the foundation of the Dimensional model. Dimensions are the nouns of the BI/DW system Dimensions are implemented as tables in the database Each dimension table contains a list of homogeneous entities. Dimension Table contains all instances of it’s entity. Dimension tables contain other entity attributes (properties) that identify, define, and classify the instance Facts are the verbs or actions of the business in which the nouns (dimensions) participate.

12 Dimensions (cont’d) Each dimension table links to all the business processes in which it participates. For example the product dimension will be linked to the sales, order, and customer services process. Think of Dimensions as things or objects Some descriptive attributes in a dimension relate to each other in a hierarchical or one to many fashion Product has Category/Sub Category/Name Store located in a Country/State/City Dimension data is normalized in Transactional Databases We De-normalize this data before loading it into dimension tables: Simple join path for faster queries Simplify the model from a user perspective

13 Bringing Facts and Dimensions together

14 Design concepts and techniques
Surrogate Keys Slowly Changing Dimensions Dates Degenerate Dimensions Snow-flacking Many-to-Many or multivalued Dimensions Hierarchies The Three Fact Table Types

15 Surrogate Keys Created in the data warehouse database.
Separate from the keys in the transaction source systems. Also known as meaningless keys, substitute keys, non-natural keys, or artificial keys. Usually an integer assigned to each row in the dimension. Becomes the primary key of the dimension table. Is used to join the dimension to the associated foreign key in the fact table.

16 Benefits of Surrogate Keys
Protect the DW/BI system from change in the source system. For example migration to a new software package which will likely create a new set of business keys. Allow the DW/BI system to integrate data from multiple source systems. Allow adding rows to dimension tables that do not exist in the source system. For example a sales transaction that comes with NULL in the sales rep field, a dimension table row can be added with the description “Sales Rep not yet assigned” Provide the means for tracking changes in the dimension attributes over time. Integer keys are an efficient key in the relational database and Analysis Services.

17 Slowly Changing Dimensions
Slowly changing dimensions (SCD) are dimensions that have changeable attribute values. Dimension attributes change over time. Example an employee who gets a promotion. ETL packages should be designed to handle this change. Some changes are significant to the business. The choice of which dimensions and attributes you need to track and how you track them is a business decision. SCD types: Type 1 Change: overwriting values Type 2 Change: tracking historical values.

18 Date Dimension Date is a fundamental business dimension across all organizations and industries. We need to create and maintain a robust date table in the DW/BI system database.

19 Snow-flacking The process of connecting lookup tables to fields in the dimension table. Snow-flacking involves re-normalization of dimension data. Snow-flacking is generally discouraged.

20 The Dimensional Modeling Process
Preparation (roles and participants, requirements, data architecture strategy) Data Profiling and Research Building Dimensional Models Developing the Detailed Dimensional Model Testing the Model Reviewing and Validating the Model

21 Tools and Technologies
SSAS Tabular SSAS MD Power BI Power BI Reporting Services

22 Magic Quadrant

23 SSAS Tabular High performance - In memory data structure
Provides business friendly report building experience. Provides a layer of abstraction. KPIs, measures, hierarchies. DAX language – easy to understand for developers. Object level security

24 SSAS MD MDX Pre-Aggregation Mature technology Can handle more data

25 Power BI Cloud based – Highly scalability and availability – no/minimum IT effort or involvement. Intuitive and easy to use also business friendly for Ad-Hoc reporting Interactive, powerful visualizations. Costs significantly less than other dashboarding tools.

26 Power BI Reporting Services
On Premises solutions. Pixel perfect reporting as well as interactive visual dashboards Modern reporting platform Free Desktop Authoring tools (report builder and Power BI desktop).

27 Methodology Agile Simple Tools Training Hire experts
Divide and conquered - Program Increment (PI) and Sprints. Iterative and incremental Simple Everything Should Be Made as Simple as Possible, But Not Simpler -Albert Einstein Tools Training Hire experts

28 Cloud vs. On Premises Pros Cons Fast Across the Board
Speed of deployment Speed of operation Speed of scalability Increase your Business Agility Fast deployment No need to setup expensive infrastructure Flexible Expenditure Automatic software Updates platform as a service Azure SQL Database Power BI Azure Analysis Services Cons Regulatory requirements healthcare financial services Connectivity and speed Visibility do you really know where your data is Accessibility Global company with remote sites in multiple countries Latency If access to data requires low and predictable latency times (Audio/Video) Lack Of Trust Security/Accessibility/Availability


Download ppt "Building a Microsoft BI solution step-by-step"

Similar presentations


Ads by Google