 News |
FAQ
|
| Search
|
Home
|
| Getting Started | Documentation | Glish | Learn More | Programming | Contact Us |
|
| Version 1.9 Build 803 |
|
The fundamental assumption underlying the MAIN table is that it contains interferometer or single-dish data, represented in cross-power or autocorrelation form. The composite key is baseline-based and auxiliary antenna-based data, that are consequently not fully functionally dependent on the baseline-based key, are stored in some of the associated sub-tables. As the latter data are often time-variable, the design admits multi-key lookups in these tables, as is already assumed in v1.0 (eg. SYSCAL table). In keeping with realistic databases, the design of the MAIN table is not fully normalized. The term is used here in the conventional database sense (see for example, Date 1986). Adding indices to MAIN to avoid sub-table lookups involves a trade-off between the anticipated increase in the physical file size (which is storage manager dependent), and the expected size of the sub-table and frequency with which the auxiliary data need to be accessed in time-critical applications. It also depends on whether the index is commonly used in data selection. Note also that auxiliary data may be sampled on different intervals, which is an important consideration in deciding which antenna-based information should be placed in the same sub-table. These factors have been balanced subject to the following guidelines: i) storage manager independence has been retained (i.e. the worst case is assumed regarding extra MAIN indices; ii) single-key lookups in sub-tables are strongly favoured; and iii) it is assumed that small sub-tables will be held in cache during MS access.
The changes proposed for the MAIN table are presented below: