Incremental updates using Data Warehouse versus Data Marts
Sonali Ashish Chakraborty, Jyotika Doshi · 2018
Organization performs analytical processing on warehouse data and generates results using OLAP queries. Query result retrieval time is high as it traverses through huge data. With periodic warehouse refresh, frequent OLAP queries fired next time may require incremental update. For processing incremental results, all records in data warehouse are scanned. To overcome this issue, executed OLAP queries are stored in a relational database, MQDB, along with its result and metadata information. Further, a copy of incremental records is stored in data marts. When a query is fired next time, incremental results are generated by accessing data marts. Final result is based on combining existing results from MQDB and incremental results from data marts. Hence, need for repeated data warehouse access reduces, resulting in faster query result retrieval. This paper evaluates query execution time of materialized queries with incremental updates when done using data marts and using data warehouse.