United Microelectronics Corporation Slashes Data Retrieval Time by 92% and Improves Overall System Performance by up to 10x
 
 

United Microelectronics Corporation Slashes Data Retrieval Time by 92% and Improves Overall System Performance by up to 10x

Established in 1980, United Microelectronics Corporation (UMC) was Taiwan’s first semiconductor company. It was also the first company in the country to provide specialized wafer services. Today, UMC has 10 wafer manufacturing plants, including two 12-inch wafer plants, with a facility that produces state-of-the-art, 28-nanometer customer products. The company plans to bring additional manufacturing facilities online.

 
 

 
 

Challenges

A word from United Microelectronics Corporation

  • “We chose Oracle Exadata Database Machine for its high performance and ability to support nonstop, onsite operations where data availability is critical to our wafer production. We’ve improved overall system performance by up to 10x, accelerated response time by 50%, slashed data retrieval time by 92%, and cut operational costs.” – Li En-Wei, Manager, Office of Information Engineering, Department of Fab 12A, Computer-Integrated Manufacturing, United Microelectronics Corporation

  • Expand data infrastructure to support rapidly-growing manufacturing plant operational needs in the next three to five years
  • Ensure 24/7 wafer plant operation and respond agilely to customer orders requiring production-line changes
  • Reduce failover switching time to avoid manufacturing disruption and delays
  • Analyze production-facility engineering problems faster

Solutions

  • Increased overall system performance by up to 10x to ensure agile response to customer orders by replacing a legacy database infrastructure that had reached full-load data capacity of 6 TB with Oracle Exadata Database Machine
  • Cut user’s lead time in analyzing engineering problems by reducing data retrieval time by 92%—from one hour, to three to five minutes
  • Reduced query response time for wafer plant, onsite, operating-system functions by about 50%—from more than 10 seconds to a few seconds—by using Oracle Exadata Smart Flash Cache, enabling staff to respond to frequent changes in production line
  • Cut storage costs by improving data compression ratio and performance by about 90%—from 125 GB to less than 13 GB—with Exadata Hybrid Columnar Compression
  • Ensured data availability and stability by using Oracle Exadata’s built-in failover mechanism to reduce production-line switching time, from 10 to 20 minutes, to 3 to 5 minutes, enabling continuous query operation without interruption
  • Achieved real-time production scheduling rather than updating once every hour, supporting a continuous manufacturing environment
  • Slashed migration time from another vendor’s environment to Oracle Exadata Database Machine, from up to six months, to only two months, minimizing operational cost and implementation risks

Why Oracle

UMC selected Oracle Exadata Database Machine to replace its traditional database framework because the company was already running Oracle Database. This accelerated and streamlined system migration while reducing operational risks and implementation costs.

“The total cost of ownership for both solutions was competitive, but it was the high system performance and availability that won the business for Oracle. We have great confidence that the data infrastructure can easily support our business growth for the next three to five years.” said Li En-Wei, manager, office of information engineering, department of fab 12A, computer-integrated manufacturing, United Microelectronics Corporation.

Implementation Process

UMC worked with Oracle Consulting to implement Oracle Exadata Database Machine X2-2 Quarter Rack. The data migration went smoothly, and it took only two months to complete, much shorter than the anticipated three to four months, thanks to Oracle Exadata’s high performance and stability.

The new system went live in May 2012. Today, four systems, including engineering data analysis and reporting, as well as plant automation, run on Oracle Exadata Database Machine.