Research in Real-Time

技术

应用基础设施与中间件 - 数据库管理和存储

适用行业

教育

适用功能

产品研发

服务

数据科学服务

挑战

The National Snow and Ice Data Center (NSIDC) at the University of Colorado is conducting research on the impact of global warming. The project, titled 'Data Rods: Enabling Time-Series Analysis of Massive Multi-Modality Cryospheric Data', is focused on the Greenland ice sheet. The challenge was to process billions of time-series information in real-time to enable a time-centric change analysis of data. The database for the project contains over 10 billion persistent objects. Indexed queries to the database can span millions of data rods simultaneously across time and space and must achieve response times of only a few seconds. Using a relational database was completely impractical due to the large size of the data sets and required response time.

关于客户

The National Snow and Ice Data Center (NSIDC) at the University of Colorado is a research institution focused on the study of the cryosphere - the portions of the Earth's surface where water is in solid form. The NSIDC is currently working on a project titled 'Data Rods: Enabling Time-Series Analysis of Massive Multi-Modality Cryospheric Data'. The project's initial target area is the Greenland ice sheet. The NSIDC is developing a system to answer questions related to the impact of global warming, such as how dust and other pollutants alter the reflective properties of snow, how global warming is impacting the Western U.S. water supply, and how permafrost melt will affect local hydrology and soil stability, as well as increase emissions of methane and other trace gases.

解决方案

The NSIDC chose to use the Versant Object Database to solve their data management challenges. This pure-object database was ideal for the project as it combines the flexibility to store arbitrarily complex data objects with true database management systems functionality: persistence, distribution, integrity, concurrency, and recovery. The Versant database easily provides the incredibly high data ingestion rates the NSIDC time-series analysis demands by eliminating the mapping layer and abundance of index structures otherwise required with a relational database. The database model representation is the same as the in-memory object model without any intermediate data definition language or runtime translation layer. Furthermore, it transparently manages object distribution across any number of physical database servers as if they are one logical database, thus providing object migration and distributed transactions.

运营影响

The Versant Object Database achieved orders of magnitude improvement in information retrieval as object models became increasingly complex.
The database transparently manages object distribution across any number of physical database servers as if they are one logical database, thus providing object migration and distributed transactions.
The database model representation is the same as the in-memory object model without any intermediate data definition language or runtime translation layer, providing high data ingestion rates.

数量效益

The database for the NSIDC project contains over 10 billion persistent objects.
Indexed queries to the database can span millions of data rods simultaneously across time and space and achieve response times of only a few seconds.