Deepola: Online aggregation for deeply nested queries

摘要: With the advent of data-driven operating model, deriving useful insights from big data analysis has become very important. But the ever-increasing volume of data has made obtaining insights at "rates resonant with the pace of human thought" [4] more challenging. Online Aggregation (OLA) [3] is a technique that tries to counter this by incrementally improving the query result estimates and allowing the user to observe the query progress as well as control its execution on the fly. OLA provides the user with an approximate estimate of the query result as soon as it has processed a small portion (hereafter referred to as a partition) of the data. The user based on their latency-error trade-off requirements can choose to stop the execution of the query.