Skip to content

Latest commit

 

History

History
15 lines (11 loc) · 1.47 KB

README.md

File metadata and controls

15 lines (11 loc) · 1.47 KB

0. Content

The configs directory contains the experiment configuration files.

  • setting.txt will be read by the scheduler and function servers. It contains the following important information:
    • proc_step: the processing step. Options are profile, aggregate and algorithm, which means profiling, performance model building, and run a specific algorithm, respectively.
    • algorithm: the algorithm to run. Options are elastic (Ditto), new_nimble (NIMBLE). This field is only used when proc_step is algorithm.
    • mode: the objective that the user concern about. Options are jct and cost. This field is only used when proc_step is algorithm and algorithm is elastic.
    • distribution: the distribution of the number of function slots on each function server. It is a list of integers separated by space. The length of the list should be equal to the number of function servers.
    • external: the type of external storage. Options are S3 and Redis.

The netconfig.txt file describes the network configuration. It will be read by the scheduler and function servers. The format is as mentioned in the root README.md.

The dsq*.txt files are the DAG description files for the foru queries.

Currently, the configuration files are only used for the 10GB dataset. For the 1TB dataset, the configuration files are hard-coded in the configs-1t directory. Users need to copy the files to the configs directory before running the 1TB experiments.