Overall Satisfaction with Oracle Autonomous Data Warehouse
A manufacturing company recently asked my company to do a Business Intelligence project to improve and standardize the analysis of its users. Before the project, users used spreadsheets and local databases based on Access to perform management analysis.
For the project, we proposed them an Oracle cloud full-stack architecture based on:
For the project, we proposed them an Oracle cloud full-stack architecture based on:
- Object storage
- Oracle Autonomous Data Warehouse
- Oracle Analytics cloud
- Export of data from the company ERP on flat files
- Creation of a staging area layer
- Creation of PL/SQL procedures to import files on staging structures, performing formal checks, cleaning and standardization processes
- Creation of entity-relation models for some Data Marts
- Creation of ETL flows to load data from the staging area on the Data Marts
- Creation of a series of institutional reports, based on Data Marts
- Profiling of users to access to the reporting layer and for free ad hoc analysis
- It's really fast to set up (like 2 minutes to create a new database)
- It's cheap, and its costs are based on dedicated resources (RAM and CPUs), and it can eventually be turned off
- Resources (RAM and CPUs) can be increased or decreased at run-time
- Patching and release upgrades are automatically performed by Oracle at scheduled times
- It's secure, without the need to implement a VPN: it provides a wallet that includes encryption methods for authentication
- It automatically extracts statistics (needed by Oracle database engine to improve performances) on its structures
- Backups are automatically performed and very easy to restore
- Disaster recovery is granted thanks to fault domains provided by Oracle
- It's really limited from a DBA point of view
- There is only 1 tablespace associated to all the users you create on the database
- The cost (license and monthly fees) are not always very clear
- The loading of data on the cloud is subjected to network speed, so huge amounts of data may take a lot of time to be loaded on the database
- No costs for hardware or server rooms
- Cost reduction, turning off or decreasing resources (RAM or CPUs) for the database when it's not needed
- No costs for maintenance of the database (i.e.: patching or resources monitoring)
- No costs for backups and disaster recovery