SMART Data Warehouse
Intelligent. Well designed. Enabling.
A Universal Platform
Semiconductor factories often have data in separate places that are useful and valuable to combine. Depending on the specifics of the systems that store the data however, there are often practical difficulties that complicate such efforts. With the FPS Data Warehouse, these struggles disappear. Everything is smartly organized in one place.
Our Data Warehouse combines all important information from factory data sources:
Any other systems
Once the integration is done, it's done. All FPS Applications now have a common platform on which to run. When we install our products at your facility, we can install the same software regardless of the type of system you have. This unique architecture also allows the same data and common calculations to be used on every web page or report. So you can be confident that a number on one report will match the same number on another report.
RECORD THOUSANDS OF EVENTS A MINUTE ACROSS ALL OF YOUR FACILITIES, COMPUTE HUNDREDS OF THOUSANDS OF ROWS OF CYCLE TIME AND THROUGHPUT DATA, POWER FULL FACTORY SCHEDULERS AND NEXT MOVE, FEED CONSTANT QUERIES FROM REPORTING APPLICATIONS AND AD HOC ANALYSIS - 24x7x365.
Can it integrate with my systems? (Yes.)
At the start of a FPS project we develop the ETL to bring all relevant data into the DWH. We are experts at retrieving the data we need and leaving a light footprint on the sources. We use everything from database links and APF reports to shell scripts to read from any source - structured databases, spreadsheets, flat files, etc. We have never met a source we can't work with. However, ETL development will be faster for established MES software that we have integrated with before.
Automatic cycle time and throughput
The FPS DWH automatically calculates cycle time (CT) and throughput (THP) from lot event data, and provides sophisticated look up architecture with intelligent fill-in logic to use it effectively. The Data Warehouse always has a best guess for when a lot will finish on a tool, or arrive at an operation 6 steps away. Why is this important? CT and THP are the data that FPS Schedulers use to look into the future and optimize schedules, and that Next Move relies on to call out upcoming actions.
We have seen many automatically calculated THP and CT data sets, but few tools that help assess the quality of the set. We also provide all the tools needed to understand the statistical quality and accuracy of the data. Our comprehensive throughput analysis can quickly identify tool sets where event data is not reliable enough to generate good data, and where external or manually overriding data should be used.
A decade of refinement and industry wisdom
The same Data Warehouse product serves all of our clients. With more than 20 distinct and diverse facilities currently using the DWH, this has put constant pressure on our developers to solve client requests always in a general, rather than a site-specific way. As a result, our Data Warehouse is a repository of industry knowledge and best practices. We compute and store a huge array of useful and value added quantities that many clients do not think to ask about but are delighted when they discover them in our system. Besides key features like Enhanced Tool Performance states, Cycle Time, and Throughput, there are smaller statistics like transit times, lot ships, line balance quantities, test wafer data, and more. When a new client gets the FPS DWH, they are getting solutions to problems and requests from more than a decade of diverse client feedback.
why doesn't my mes include all of this?
Although a MES provides the majority of data in our DWH, there is a good technical reason for maintaining a separate database for computing and storing value-added extrinsic quantities. A MES bears a heavy transactional load recording all Fab events in real time. Any computationally-intensive work is therefore better done on a different system. The FPS DWH retrieves all information logged in the MES, but does so in batches on the order of every minute, rather than right at the millisecond each event happens. For all practical purposes the data in the FPS DWH is still real time, but with this approach there is plenty of computational power left over to do the heavy lifting.