Figure 2
From: DRUMS: Disk Repository with Update Management and Select option for high throughput sequencing data
High level overview of select process. When a request is sent to DRUMS, four steps are done to read the requested records. 1) The bucket of interest is determined. 2) The correct chunk of the first requested record is identified, using a sparse index. 3) The position of the requested key-value pair is determined. 4) A sequential read is performed until the requested range is completely processed.