基于压缩-字对齐位图的天文海量数据实时索引

澄江一米新真空大型天文望远镜（NVST）当前每天最大能产生2 TB，约十多万条的观测数据。由于这些数据量巨大并具有非结构化特性，使用离线构建索引会带来巨大时间开销，传统的关系型数据库难以满足快速索引和检索需求。针对这些问题，结合数据采集流程，提出了使用基于压缩的字对齐位图索引算法来在线实时构建索引。这种方式不仅克服了离线构建索引方式时，文件访问、FITS头读取和解析FITS头等操作带来的大量额外时间消耗问题，而且有助于解决海量太阳观测数据的高效检索难题。通过实验证明了在线实时构建索引方式能够极大地降低时间开销，也表明了该方式在天文海量数据索引和检索应用中的有效性和可行性。

Massive astronomical real-time data indexing based on compressed word-aligned bitmap

At present, New Vacuum Solar Telescope（NVST） is generating data more than 2 TB per day, and due to the characteristic of massive non-structure data, it is beyond traditional database systems to search efficiently for a subset of these extremely large with low latency and response time and time overtime is increasing by the way of off-line index building. Aiming at these problems, combined with the strategy of on-line real data indexing in observation data capture system, using an approach of the compressed word-aligned bitmap index for observation data indexing and querying is proposed. This approach cannot only save the time cost of accessing FITS file, reading FITS file header and parsing header keyword in off-line indexing mode, but also can solve the problem of massive solar data retrieval. Experiments show that time overhead can be reduced by using real-time data indexing, and the effectiveness and feasibility of the method are proved in the experiments.