Simulation model for task pool size calculation.
This model allows to perform experiments of task distribution process in bft environment. Model has 3 parameters:
- number of nodes
-n - number of tasks
-v - precision (number of experiments)
-e
Model produces size of node's task pool for chosen precision.
Detailed information about task distribution problem can be found on medium article.
You can run model inside a docker container or run app manually with
go run. Don't forget to turn on gomodules in that case!
$ make up
Sending build context to Docker daemon 114.7kB
. . .
/ #
/ # bft-task-distrib-model -n 10 -v 100 -e 1000
# BFT TASK DISTRIBUTION MODEL v.1.0
# N = 10 V = 100
# EXP = 1000
# AVAILABLE WORKERS = 8
# STARTING AT: 2019-01-14T09:49:00Z
3:67 (1.000) 1(1)
6:50 (0.980) 2(2)
9:45 (1.000) 1(1)
# ENDED AT: 2019-01-14T09:49:00Z
# DURATION: 40.430027ms
Output string 3:67 (1.000) 1(1) means, that for 3 nodes, pool
size must be 67 tasks.
Relation between max pool size and obtained pool size is
1.000 - so they equal.
Last column is a number of tries to reduce pool size.
There is also gnuplot file with example from medium article:
$ gnuplot
Terminal type is now 'qt'
gnuplot> load 'results.gp'
This project is licensed under the GPLv3 License - see the LICENSE.md file for details