----------------------------------------------------------------------------- -- | -- Module : Data.SBV.Examples.Optimization.VM -- Copyright : (c) Levent Erkok -- License : BSD3 -- Maintainer : erkokl@gmail.com -- Stability : experimental -- -- Solves a VM allocation problem using optimization features ----------------------------------------------------------------------------- module Data.SBV.Examples.Optimization.VM where import Data.SBV -- | The allocation problem. Inspired by: <http://rise4fun.com/z3opt/tutorialcontent/guide#h25> -- -- - We have three virtual machines (VMs) which require 100, 50 and 15 GB hard disk respectively. -- -- - There are three servers with capabilities 100, 75 and 200 GB in that order. -- -- - Find out a way to place VMs into servers in order to -- -- - Minimize the number of servers used -- -- - Minimize the operation cost (the servers have fixed daily costs 10, 5 and 20 USD respectively.) -- -- We have: -- -- >>> optimize Lexicographic allocate -- Optimal model: -- x11 = False :: Bool -- x12 = False :: Bool -- x13 = True :: Bool -- x21 = False :: Bool -- x22 = False :: Bool -- x23 = True :: Bool -- x31 = False :: Bool -- x32 = False :: Bool -- x33 = True :: Bool -- noOfServers = 1 :: Integer -- cost = 20 :: Integer -- -- That is, we should put all the jobs on the third server, for a total cost of 20. allocate :: Goal allocate = do -- xij means VM i is running on server j x1@[x11, x12, x13] <- sBools ["x11", "x12", "x13"] x2@[x21, x22, x23] <- sBools ["x21", "x22", "x23"] x3@[x31, x32, x33] <- sBools ["x31", "x32", "x33"] -- Each job runs on exactly one server constrain $ pbStronglyMutexed x1 constrain $ pbStronglyMutexed x2 constrain $ pbStronglyMutexed x3 let need :: [SBool] -> SInteger need rs = sum $ zipWith (\r c -> ite r c 0) rs [100, 50, 15] -- The capacity on each server is respected let capacity1 = need [x11, x21, x31] capacity2 = need [x12, x22, x32] capacity3 = need [x13, x23, x33] constrain $ capacity1 .<= 100 constrain $ capacity2 .<= 75 constrain $ capacity3 .<= 200 -- compute #of servers running: let y1 = bOr [x11, x21, x31] y2 = bOr [x12, x22, x32] y3 = bOr [x13, x23, x33] b2n b = ite b 1 0 let noOfServers = sum $ map b2n [y1, y2, y3] -- minimize # of servers minimize "noOfServers" (noOfServers :: SInteger) -- cost on each server let cost1 = ite y1 10 0 cost2 = ite y2 5 0 cost3 = ite y3 20 0 -- minimize the total cost minimize "cost" (cost1 + cost2 + cost3 :: SInteger)