TSG: A New Approach to Preserving Same-Timestamp Data in Time-Series Databases

Soonbeom Kwon, Cheongu Kim, Young-Kyoon Suh

Abstract: With the growing adoption of Internet of Things (IoT)sensors and applications, the volume of time-series data is expanding rapidly, underscoring the increasing importance of time-series databases (TSDBs) for efficient data management across edge and cloud environments. TSDBs are required to deliver high throughput and low latency to meet performance demands. However, due to strict data management rules, conventional TSDBs often overwrite redundant time-series data with identical timestamps. To overcome this limitation, we propose Tag-based Sequential Grouping (TSG), a novel approach that preserves time-series data with identical timestamps in TSDBs. TSG leverages tag identifiers as a secondary indexing mechanism to store data without overwriting, ensuring complete retention. We evaluate TSG on state-of-the-art TSDBs using real-world datasets. Our experimental results demonstrate that TSG successfully retains 100% of time-series data with identical timestamps across various TSDBs without data loss. Moreover, TSG significantly enhances read latency, outperforming existing methods by up to 322 times on a 6-hour interval range query.

External IDs:dblp:conf/noms/KwonKS25