Edit this page | Blame

Pre-Cache Datasets

Tags

  • assigned:
  • priority: medium
  • type: enhancement
  • status: open
  • keywords: cache, optimisation

Description

To improve the performance of the system when running computations (correlations, mappings, etc), we need to pre-cache the datasets in text files.

The triggers for pre-caching could be:

  • creation of a new dataset
  • changes in data for (a) trait(s) in the dataset
  • changes in sample list for a dataset

I propose an external job be triggered whenever any of the triggers above happen. The job could, among other things:

  • Delete existing cache file
  • Create new cache file with new data

Maybe, if possible, we could have the pre-cache service generate the cache files based on dates for latest changes without deleting older cache files -- we could look into whether this is possible.

Related Issues

(made with skribilo)