Any company worth its salt is turning to big data to help it stay on top of the game. One such company is Starbucks.
Starbucks has a very large database warehouse that grows around two to three terabytes a year. This enterprise data warehouse is accessed by thousands of its stores, providing store level reporting to each one. Data includes staffing, inventory and sales reports, among others. The company uses different tools including SQL Developer, Toad and others created by its QA and development team to extract, transform and load data from its enterprise data warehouse. In short, Starbucks is able to get data from outside sources, format this data to something that they could use and load the data into an end target, which is the data warehouse.
The company uses 4 CPU servers with 1.5 gigahertz of computing power and a 16-gigabyte memory. It is connected via a private interconnect network as well as a gigabit Ethernet public network. It uses Oracle Database 10.2.0.3 Enterprise Editions, as well as Recovery Manager for its backup.
Backing up Using Oracle Recovery Manager
The company uses Oracle Recovery Manager to back up its data. It set the enterprise data warehouse as the active or primary database and replicates data from there. New data is put up in read-write servers while older data (operationally, everything older than two months) would be placed in read-only data servers.
The company considered split mirror backups for its requirements, but decided on Oracle Recovery Manager because it was easy to implement, is more flexible and has a lower cost.
Oracle Recovery Manager gave Starbucks an efficient backup and recovery system, without sacrificing performance. The backups also took up as little space as possible because of its set compression and file multiplexing.
As a result, Starbucks was able to achieve their aim of:
- Minimizing the load on the array and servers when backing up the data.
- Getting faster backup times.
- Minimizing space consumed by duplicated files.
- Lowering storage and hardware costs.
- Scaling backup system as the enterprise data warehouse grows.
Oracle Recovery Manager also works well with tape backups, which Starbucks did on a weekly basis.
The company was also able to use multi-section backups, which is used to back up several pieces of a very large file in parallel. Oracle Recovery Manager will divide the work across different channels and each channel will be backing up one section of the file.
Further, the company also used stored scripts and command files that allow user input when run. This was perfect for doing read-only data backups as well as spreading the backup of older data over a month.
Then there is the undo optimization feature, which lets you bypass the typically large undo tablespaces with lots of useless information. This way, you are not backing up these useless data and wasting space. Plus Oracle Recovery Manager allowed them to copy from a live source database.
Four Cornerstone can help you gain the benefits that Starbucks got from Oracle Recovery Manager. Just call us today and we can help set up your backup systems for your data warehouse!
For more information and details regarding Starbucks’ use of Oracle products, you can read this 20-page case study: //www.oracle.com/technetwork/database/features/availability/s291487-2-chien-134699.pdf .