Abstract
We use a metadata service at the Belle experiment which provides a mechanism to locate files using descriptive information. However, at the Belle II experiment, we will have 50 60 times more data than that of the Belle experiment.
Therefore, it is expected that the existing metadata service has problems with performance, scalability, and durability, in particular, if it is extended to an event-level for searching metadata. To deal with this issue, we have designed a new metadata schema for Belle II which significantly reduces disk space for metadata, and proposed a new metadata service system which provides good performance and scalability based on Arda Metadata catalog for Grid Application.
The control of the event-level metadata provides an efficient scheme of processing such as events with many tracks.