Data Layout For Spatio-temporal Data In Time Domain Astronomy

Astronomy is one of the first disciplines which stepped into the era of big data.With the increasing of the telescope’s observation data,astronomical data storage has become a new issue gradually.Different from the traditional data,Astronomy data possesses the characteristics of both space and time,and has a large volume.As the branch of astronomy,Time domain astronomy is more affected by this data feature.Time domain astronomy puts emphasis on how astronomical objects change with time.The data are usually stored according to the observation time sequence whereas accessed in spatial order.Traditional storage layout always consumes a lot of energy in data search stage because it usually needs to visit all the disks to get the target data.However,energy is the most limitation factor in Antarctic.So a layout specifically designed for the Antarctic astronomical observations is necessary.In this paper,based on the survey of the traditional data layout system and energy storage technology,a new data layout of the Antarctic astronomical observations with spatio-temporal data has been designed.This layout is different from the traditional storage layout which stores data according to the time order.Instead,it combines the disk array and the observation area and uses a cache disk to transform the original time-ordered observation data into the space-ordered ones.Then it stores data in the corresponding disk.On other hand,the layout uses a tree structure on the hot area and the redundant layout on the cold area to reduce the energy consumption.In order to evaluation the performance of the storage layout,this paper uses the full average data,simulation data and real observation data to conduct an experiment on the Antarctic astronomical observation telescope AST3.The results show that compared with the traditional layout of the data,the new data layout’s number of disk visit times reduced from 10.16 times to 1.85 times,and compared with the original energy consumption,it reduced about 67.57% in the query phase.