Tell me how to organize, for example, buggyg (or, possibly, boosting) on ​​several machines? I would like to understand how they do it in two scenarios:

  • All data is on the same machine, but in computing power is in the amount of k machines
  • All data lies on a cluster of k machines, for example, in HBase.

Here is the question of whether someone runs on several machines and how, if so?

  • one
    Well, this is already some kind of Map-Reduce needed in the direction of Hadoop, Spark, etc. If not so big, then Dask can work on a cluster of machines, but I didn’t really understand this option. - CrazyElf

0