Open ArturAkh opened 2 years ago
To put this into perspective, the above described issue refers to the initialization of the files at simulation start. Once a bigger simulation is started with many caching jobs changing the status of the caches, this initialization might not have an influence at all.
Additionally, some configurable option for the initial input-file staging at simulation start needs to be designed.
According to the following code:
https://github.com/HerrHorizontal/DistCacheSim/blob/test/sgbatch/sgbatch/src/SimpleSimulator.cpp#L398-L406
The files are duplicated on each cache at the beginning, leading to the same set of files and same hitrate on each cache at the beginning.
An alternative strategy can be implemented, e.g. by putting files randomly on only one cache, until the configured (average) hitrate is reached.
Could be done via an additional CLI option.