Hello,
I recently studied the code of sparrow. I have one short question. From the Fig 6 of the SOSP paper, it seems that the nodeMonitor and the scheduler can be launched on different machines. But from the code, these two parts are integrated into one component called Sparrow Deamon. Both of them will be launched on the same machine once a sparrow deamon is launched. I wanna to do some experiments with sparrow on my own cluster with 15 machines. Three of them will be running the FrontEnds/Scehdulers. And 12 of them will be running the NodeMonitors/Bandends. What could be the right way to do that? Do i need to comment out the NodeMonitor code when i launch the Deamon on the 3 machines, and then comment out the Scheduler code when launch the Deamon on the 12 machines?
Thank you and Merry Christmas.
Dingming