高能暂现源 · 每日文献综述

The increasing temporal resolution and structural diversity of modern solar instruments place growing demands on database systems used in observational astronomy. At the Center for Radio Astronomy and Astrophysics Mackenzie (CRAAM), this challenge is amplified by the need to consolidate heterogeneous data streams from multiple telescopes within a single virtual machine. With only 32GB of RAM available (16GB allocated to the database), a central design question emerged: when restricted to a single physical host, can a virtualized sharded cluster offer practical scalability advantages over a standalone deployment? To investigate this, we conducted an empirical evaluation of MongoDB using 10ms observations from the POEMAS radiotelescope, tested at volumes of 15M, 150M, and 500M documents. Results show that, although sharding introduces coordination overhead for selective queries, it provides substantial gains for global aggregations, achieving speedups above 600<mml:math><mml:mo>×</mml:mo></mml:math> while maintaining compression ratios near 85%. The analysis identifies an operational threshold of roughly 150 million documents per collection to sustain stable performance under the available resources. Based on these findings, the same single-node configuration used in the benchmarks was employed to process the full historical POEMAS dataset, totaling 3.3 billion records and producing approximately 50GB of consolidated FITS products. These products and their associated metadata are made available to the community through a cloud-hosted portal with reduced operational cost. This work documents practical scalability boundaries for astronomical time-series in resource-constrained environments and supports the deployment currently operating at CRAAM.