Increasing　research　efforts　have　been　devoted　to　the　reliability　analysis　of　multi-state　systems　(MSSs)　with　different　performance　sharing　mechanisms.　This　paper　investigates　the　reliability　optimization　problem　of　an　MSS　with　two　performance　sharing　groups.　Such　systems　consist　of　multiple　components　and　two　common　buses.　The　components　connected　by　the　same　common　bus　make　up　a　group　and　the　performance　can　be　shared　in　the　group　to　meet　the　demand.　Specially,　components　can　be　in　one　group　or　both.　Taking　advantage　of　this　property,　the　components　with　larger　performance　and　demand　are　preferentially　placed　in　the　overlapping　part　of　two　groups　since　more　performance　can　be　shared.　The　objective　is　to　efficiently　derive　the　optimal　configuration　policy　that　maximizes　the　system　reliability.　Universal　generating　function　is　utilized　to　develop　the　reliability　evaluation　algorithm.　Then,　stochastic　orders　are　utilized　to　determine　the　components　to　be　placed　in　the　overlapping　part,　which　reduces　the　solution　space.　Genetic　algorithm　is　applied　to　find　the　optimization　results　in　the　reduced　solution　space.　Numerical　examples　validate　the　effectiveness　of　the　proposed　reliability　evaluation　algorithm.　The　improved　optimization　method　also　outperforms　the　method　without　solution　space　reduction　in　improving　the　system　reliability.