Edge　computing　is　an　emerging　paradigm　that　settles　some　servers　on　the　near-user　side　and　allows　some　real-time　requests　from　users　to　be　directly　returned　to　the　user　after　being　processed　by　these　servers　settled　on　the　near-user　side.　In　this　paper,　we　focus　on　saving　the　energy　of　the　system　to　provide　an　efficient　scheduling　strategy　in　edge　computing.　Our　objective　is　to　reduce　the　power　consumption　for　the　providers　of　the　edge　nodes　while　meeting　the　resources　and　delay　constraints.　We　propose　a　two-stage　scheduling　strategy　which　includes　the　scheduling　and　resource　provisioning.　In　the　scheduling　stage,　we　first　propose　an　efficient　scheme　based　on　the　branch　and　bound　method.　In　order　to　reduce　complexity,　we　propose　a　heuristic　algorithm　that　guarantees　users＇　deadlines.　In　the　resource　provisioning　stage,　we　first　approach　the　problem　by　virtualizing　the　edge　nodes　into　master　and　slave　nodes　based　on　the　sleep　power　consumption　mode.　After　that,　we　propose　a　scheduling　strategy　through　balancing　the　resources　of　virtual　nodes　that　reduce　the　power　consumption　and　guarantees　the　user＇s　delay　as　well.　We　use　iFogSim　to　simulate　our　strategy.　The　simulation　results　show　that　our　strategy　can　effectively　reduce　the　power　consumption　of　the　edge　system.　In　the　test　of　idle　tasks,　the　highest　energy　consumption　was　27.9%　lower　than　the　original　algorithm.