RDBArchive
This archive system is an upgrade of the ChannelArchiver, where the maintenance of the custom binary data and index files became overwhelming, at least for use at the SNS.
The new system in short:
- Data and configuration is stored in RDB (Oracle, MySQL)
- Sample engine is written in Java ('headless' Eclipse/CSS)
- CSS
can plot data from both old ChannelArchiver and RDB
Compared to the ChannelArchiver, pure read/write performance is reduced. Tests with Oracle allowed writing up to 8000 samples per second, while the ChannelArchiver storage library usually execeeds 60000/sec. Tests with MySQL were better, but the exact limitations due to MySQL table were not investigated.
The main reason for the new development were maintainability and accessibility. The ChannelArchiver setup at the SNS, based on file size limitations and worries about data loss, had grown into a setup with about 80 sub-archives. Access to data was cumbersome.
With an RDB setup, data access for customized reports is trivial, and the state of the system (which engine stored how many samples in the last hour?) is easy to determine.
- https://ics-web.sns.ornl.gov/css/devel.html#archiver
- Manual of the new RDBArchiveEngine