Oracle North America Sales Operations moves 2 TB forecasting and data science data marts, supporting 500 users, to Autonomous Database with no DBAs.
“Today two people, neither of whom are DBAs, are supporting a data mart that grew from 200 to 500 users, including both data scientists and business analysts. We were able to add support for 300 new users without adding additional support people. This is only possible because of Autonomous Database.”
For many years, North America Sales Operations maintained an on-premises data mart. Used for Oracle sales and forecasting analysis, the data mart was managed by an internal IT group. However, when IT was no longer able to manage that environment, Sales Operations needed a new cloud-based data mart that was much easier to manage and operate. Efficiency was also key as there was no budget for administration. This valuable data is also used by data scientists, so a new data mart could potentially provide new insights.
Why Oracle chose Autonomous Database
The team moved the data mart to Oracle Autonomous Database, adding Oracle Analytics Cloud as the self-service analytics tool. The lack of both DBA skills in the team and an external support team in IT meant that autonomous administration was essential to the project. Sales Operations also could improve efficiency by enabling autoscaling, right-sizing for peak loads.
This 2 TB internal data mart originally supported around 200 users. Because of autonomous administration, it has been easy to add support for other groups. When another department brought 300 new users and some new data, the existing support team was able to handle that without needing more resources. So now there are 500-plus users who analyze and forecast sales pipeline as well as consumption data.
A single instance of Autonomous Database for analytics and warehousing is used for the main data mart, supporting primarily the analytics users. A separate instance of Autonomous Database is for transaction processing and mixed workloads. This is used as a staging area for running transformations and to support the data science teams.
OCI Data Science helps data scientists build their machine learning models. Data is extracted from the Autonomous Database—used as a staging area—and results are placed back there to be loaded in to the main data mart.
A data mart like this naturally has a variable workload, with heavier use during the day than at night, and significantly heavier use at the end of the month. Autoscaling is switched on as an insurance policy. And it turns out that auto-scaling is indeed needed from time to time, to meet peak demand.
This team picked Autonomous Database because there were no DBA resources available to manage the environment. Today, the entire setup is supported by two people who perform no database administration. This enables them to focus on designing new analytics capabilities and features.