diff --git a/cmd/integration/integration.go b/cmd/integration/integration.go
index d6d96d04a04..5ada6f59a14 100644
--- a/cmd/integration/integration.go
+++ b/cmd/integration/integration.go
@@ -221,7 +221,7 @@ func startComponents(firstManifestURL, secondManifestURL, apiVersion string) (st
 			api.ResourceName(api.ResourceMemory): resource.MustParse("10G"),
 		}}
 
-	nodeController := nodeControllerPkg.NewNodeController(nil, "", machineList, nodeResources, cl, fakeKubeletClient{}, 10, 5*time.Minute, util.NewFakeRateLimiter(), 40*time.Second, 60*time.Second, 5*time.Second)
+	nodeController := nodeControllerPkg.NewNodeController(nil, "", machineList, nodeResources, cl, fakeKubeletClient{}, 10, 5*time.Minute, util.NewFakeRateLimiter(), 40*time.Second, 60*time.Second, 5*time.Second, "")
 	nodeController.Run(5*time.Second, true)
 	cadvisorInterface := new(cadvisor.Fake)
 
diff --git a/cmd/kube-controller-manager/app/controllermanager.go b/cmd/kube-controller-manager/app/controllermanager.go
index 8d6f33d7d9a..aa063b6a9f1 100644
--- a/cmd/kube-controller-manager/app/controllermanager.go
+++ b/cmd/kube-controller-manager/app/controllermanager.go
@@ -70,6 +70,7 @@ type CMServer struct {
 	NodeMemory   resource.Quantity
 
 	KubeletConfig   client.KubeletConfig
+	ClusterName     string
 	EnableProfiling bool
 }
 
@@ -91,6 +92,7 @@ func NewCMServer() *CMServer {
 			EnableHttps: true,
 			HTTPTimeout: time.Duration(5) * time.Second,
 		},
+		ClusterName: "kubernetes",
 	}
 	return &s
 }
@@ -132,7 +134,8 @@ func (s *CMServer) AddFlags(fs *pflag.FlagSet) {
 	fs.Int64Var(&s.NodeMilliCPU, "node_milli_cpu", s.NodeMilliCPU, "The amount of MilliCPU provisioned on each node")
 	fs.Var(resource.NewQuantityFlagValue(&s.NodeMemory), "node_memory", "The amount of memory (in bytes) provisioned on each node")
 	client.BindKubeletClientConfigFlags(fs, &s.KubeletConfig)
-	fs.BoolVar(&s.EnableProfiling, "profiling", true, "Enable profiling via web interface host:port/debug/pprof/")
+	fs.StringVar(&s.ClusterName, "cluster_name", s.ClusterName, "The instance prefix for the cluster")
+	fs.BoolVar(&s.EnableProfiling, "profiling", false, "Enable profiling via web interface host:port/debug/pprof/")
 }
 
 func (s *CMServer) verifyMinionFlags() {
@@ -198,7 +201,7 @@ func (s *CMServer) Run(_ []string) error {
 
 	nodeController := nodeControllerPkg.NewNodeController(cloud, s.MinionRegexp, s.MachineList, nodeResources,
 		kubeClient, kubeletClient, s.RegisterRetryCount, s.PodEvictionTimeout, util.NewTokenBucketRateLimiter(s.DeletingPodsQps, s.DeletingPodsBurst),
-		s.NodeMonitorGracePeriod, s.NodeStartupGracePeriod, s.NodeMonitorPeriod)
+		s.NodeMonitorGracePeriod, s.NodeStartupGracePeriod, s.NodeMonitorPeriod, s.ClusterName)
 	nodeController.Run(s.NodeSyncPeriod, s.SyncNodeList)
 
 	resourceQuotaManager := resourcequota.NewResourceQuotaManager(kubeClient)
diff --git a/cmd/kubernetes/kubernetes.go b/cmd/kubernetes/kubernetes.go
index f32e253db93..a610c840e4f 100644
--- a/cmd/kubernetes/kubernetes.go
+++ b/cmd/kubernetes/kubernetes.go
@@ -131,7 +131,7 @@ func runControllerManager(machineList []string, cl *client.Client, nodeMilliCPU,
 	kubeClient := &client.HTTPKubeletClient{Client: http.DefaultClient, Port: ports.KubeletPort}
 
 	nodeController := nodeControllerPkg.NewNodeController(
-		nil, "", machineList, nodeResources, cl, kubeClient, 10, 5*time.Minute, util.NewTokenBucketRateLimiter(*deletingPodsQps, *deletingPodsBurst), 40*time.Second, 60*time.Second, 5*time.Second)
+		nil, "", machineList, nodeResources, cl, kubeClient, 10, 5*time.Minute, util.NewTokenBucketRateLimiter(*deletingPodsQps, *deletingPodsBurst), 40*time.Second, 60*time.Second, 5*time.Second, "")
 	nodeController.Run(10*time.Second, true)
 
 	endpoints := service.NewEndpointController(cl)
diff --git a/pkg/cloudprovider/cloud.go b/pkg/cloudprovider/cloud.go
index c50d0627ebb..3f76e80147f 100644
--- a/pkg/cloudprovider/cloud.go
+++ b/pkg/cloudprovider/cloud.go
@@ -42,6 +42,10 @@ type Clusters interface {
 	Master(clusterName string) (string, error)
 }
 
+func GetLoadBalancerName(clusterName, serviceNamespace, serviceName string) string {
+	return clusterName + "-" + serviceNamespace + "-" + serviceName
+}
+
 // TCPLoadBalancer is an abstract, pluggable interface for TCP load balancers.
 type TCPLoadBalancer interface {
 	// TCPLoadBalancerExists returns whether the specified load balancer exists.
diff --git a/pkg/cloudprovider/controller/nodecontroller.go b/pkg/cloudprovider/controller/nodecontroller.go
index 43f13a69960..8b6965d27fb 100644
--- a/pkg/cloudprovider/controller/nodecontroller.go
+++ b/pkg/cloudprovider/controller/nodecontroller.go
@@ -89,6 +89,9 @@ type NodeController struct {
 	// check node status posted from kubelet. This value should be lower than nodeMonitorGracePeriod.
 	// TODO: Change node status monitor to watch based.
 	nodeMonitorPeriod time.Duration
+	clusterName       string
+	// Should external services be reconciled during syncing cloud nodes, even though the nodes were not changed.
+	reconcileServices bool
 	// Method for easy mocking in unittest.
 	lookupIP func(host string) ([]net.IP, error)
 	now      func() util.Time
@@ -107,7 +110,8 @@ func NewNodeController(
 	deletingPodsRateLimiter util.RateLimiter,
 	nodeMonitorGracePeriod time.Duration,
 	nodeStartupGracePeriod time.Duration,
-	nodeMonitorPeriod time.Duration) *NodeController {
+	nodeMonitorPeriod time.Duration,
+	clusterName string) *NodeController {
 	eventBroadcaster := record.NewBroadcaster()
 	recorder := eventBroadcaster.NewRecorder(api.EventSource{Component: "controllermanager"})
 	if kubeClient != nil {
@@ -133,6 +137,7 @@ func NewNodeController(
 		nodeStartupGracePeriod:  nodeStartupGracePeriod,
 		lookupIP:                net.LookupIP,
 		now:                     util.Now,
+		clusterName:             clusterName,
 	}
 }
 
@@ -220,6 +225,62 @@ func (nc *NodeController) RegisterNodes(nodes *api.NodeList, retryCount int, ret
 	}
 }
 
+// reconcileExternalServices updates balancers for external services, so that they will match the nodes given.
+// Returns true if something went wrong and we should call reconcile again.
+func (nc *NodeController) reconcileExternalServices(nodes *api.NodeList) (shouldRetry bool) {
+	balancer, ok := nc.cloud.TCPLoadBalancer()
+	if !ok {
+		glog.Error("The cloud provider does not support external TCP load balancers.")
+		return false
+	}
+
+	zones, ok := nc.cloud.Zones()
+	if !ok {
+		glog.Error("The cloud provider does not support zone enumeration.")
+		return false
+	}
+	zone, err := zones.GetZone()
+	if err != nil {
+		glog.Errorf("Error while getting zone: %v", err)
+		return false
+	}
+
+	hosts := []string{}
+	for _, node := range nodes.Items {
+		hosts = append(hosts, node.Name)
+	}
+
+	services, err := nc.kubeClient.Services(api.NamespaceAll).List(labels.Everything())
+	if err != nil {
+		glog.Errorf("Error while listing services: %v", err)
+		return true
+	}
+	shouldRetry = false
+	for _, service := range services.Items {
+		if service.Spec.CreateExternalLoadBalancer {
+			nonTCPPort := false
+			for i := range service.Spec.Ports {
+				if service.Spec.Ports[i].Protocol != api.ProtocolTCP {
+					nonTCPPort = true
+					break
+				}
+			}
+			if nonTCPPort {
+				// TODO: Support UDP here.
+				glog.Errorf("External load balancers for non TCP services are not currently supported: %v.", service)
+				continue
+			}
+			name := cloudprovider.GetLoadBalancerName(nc.clusterName, service.Namespace, service.Name)
+			err := balancer.UpdateTCPLoadBalancer(name, zone.Region, hosts)
+			if err != nil {
+				glog.Errorf("External error while updating TCP load balancer: %v.", err)
+				shouldRetry = true
+			}
+		}
+	}
+	return shouldRetry
+}
+
 // SyncCloudNodes synchronizes the list of instances from cloudprovider to master server.
 func (nc *NodeController) SyncCloudNodes() error {
 	matches, err := nc.GetCloudNodesWithSpec()
@@ -238,6 +299,7 @@ func (nc *NodeController) SyncCloudNodes() error {
 
 	// Create nodes which have been created in cloud, but not in kubernetes cluster
 	// Skip nodes if we hit an error while trying to get their addresses.
+	nodesChanged := false
 	for _, node := range matches.Items {
 		if _, ok := nodeMap[node.Name]; !ok {
 			glog.V(3).Infof("Querying addresses for new node: %s", node.Name)
@@ -255,6 +317,7 @@ func (nc *NodeController) SyncCloudNodes() error {
 			if err != nil {
 				glog.Errorf("Create node %s error: %v", node.Name, err)
 			}
+			nodesChanged = true
 		}
 		delete(nodeMap, node.Name)
 	}
@@ -267,6 +330,15 @@ func (nc *NodeController) SyncCloudNodes() error {
 			glog.Errorf("Delete node %s error: %v", nodeID, err)
 		}
 		nc.deletePods(nodeID)
+		nodesChanged = true
+	}
+
+	// Make external services aware of nodes currently present in the cluster.
+	if nodesChanged || nc.reconcileServices {
+		nc.reconcileServices = nc.reconcileExternalServices(matches)
+		if nc.reconcileServices {
+			glog.Error("Reconcilation of external services failed and will be retried during the next sync.")
+		}
 	}
 
 	return nil
diff --git a/pkg/cloudprovider/controller/nodecontroller_test.go b/pkg/cloudprovider/controller/nodecontroller_test.go
index 6d6e1f1b2f9..2ea248b0836 100644
--- a/pkg/cloudprovider/controller/nodecontroller_test.go
+++ b/pkg/cloudprovider/controller/nodecontroller_test.go
@@ -258,7 +258,7 @@ func TestRegisterNodes(t *testing.T) {
 			nodes.Items = append(nodes.Items, *newNode(machine))
 		}
 		nodeController := NewNodeController(nil, "", item.machines, &api.NodeResources{}, item.fakeNodeHandler, nil, 10, time.Minute,
-			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		err := nodeController.RegisterNodes(&nodes, item.retryCount, time.Millisecond)
 		if !item.expectedFail && err != nil {
 			t.Errorf("unexpected error: %v", err)
@@ -344,7 +344,7 @@ func TestCreateGetStaticNodesWithSpec(t *testing.T) {
 	}
 	for _, item := range table {
 		nodeController := NewNodeController(nil, "", item.machines, &resources, nil, nil, 10, time.Minute,
-			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		nodes, err := nodeController.GetStaticNodesWithSpec()
 		if err != nil {
 			t.Errorf("unexpected error: %v", err)
@@ -406,7 +406,7 @@ func TestCreateGetCloudNodesWithSpec(t *testing.T) {
 
 	for _, item := range table {
 		nodeController := NewNodeController(item.fakeCloud, ".*", nil, &api.NodeResources{}, nil, nil, 10, time.Minute,
-			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		nodes, err := nodeController.GetCloudNodesWithSpec()
 		if err != nil {
 			t.Errorf("unexpected error: %v", err)
@@ -516,7 +516,7 @@ func TestSyncCloudNodes(t *testing.T) {
 			item.fakeNodeHandler.Fake = testclient.NewSimpleFake()
 		}
 		nodeController := NewNodeController(item.fakeCloud, item.matchRE, nil, &api.NodeResources{}, item.fakeNodeHandler, nil, 10, time.Minute,
-			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		if err := nodeController.SyncCloudNodes(); err != nil {
 			t.Errorf("unexpected error: %v", err)
 		}
@@ -577,7 +577,7 @@ func TestSyncCloudNodesEvictPods(t *testing.T) {
 			matchRE:              ".*",
 			expectedRequestCount: 2, // List + Delete
 			expectedDeleted:      []string{"node1"},
-			expectedActions:      []testclient.FakeAction{{Action: "list-pods"}, {Action: "delete-pod", Value: "pod0"}},
+			expectedActions:      []testclient.FakeAction{{Action: "list-pods"}, {Action: "delete-pod", Value: "pod0"}, {Action: "list-services"}},
 		},
 		{
 			// Delete node1, but pod0 is running on node0.
@@ -591,7 +591,7 @@ func TestSyncCloudNodesEvictPods(t *testing.T) {
 			matchRE:              ".*",
 			expectedRequestCount: 2, // List + Delete
 			expectedDeleted:      []string{"node1"},
-			expectedActions:      []testclient.FakeAction{{Action: "list-pods"}},
+			expectedActions:      []testclient.FakeAction{{Action: "list-pods"}, {Action: "list-services"}},
 		},
 	}
 
@@ -600,7 +600,7 @@ func TestSyncCloudNodesEvictPods(t *testing.T) {
 			item.fakeNodeHandler.Fake = testclient.NewSimpleFake()
 		}
 		nodeController := NewNodeController(item.fakeCloud, item.matchRE, nil, &api.NodeResources{}, item.fakeNodeHandler, nil, 10, time.Minute,
-			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		if err := nodeController.SyncCloudNodes(); err != nil {
 			t.Errorf("unexpected error: %v", err)
 		}
@@ -617,6 +617,71 @@ func TestSyncCloudNodesEvictPods(t *testing.T) {
 	}
 }
 
+func TestSyncCloudNodesReconcilesExternalService(t *testing.T) {
+	table := []struct {
+		fakeNodeHandler       *FakeNodeHandler
+		fakeCloud             *fake_cloud.FakeCloud
+		matchRE               string
+		expectedClientActions []testclient.FakeAction
+		expectedUpdateCalls   []fake_cloud.FakeUpdateBalancerCall
+	}{
+		{
+			// Set of nodes does not change: do nothing.
+			fakeNodeHandler: &FakeNodeHandler{
+				Existing: []*api.Node{newNode("node0"), newNode("node1")},
+				Fake:     testclient.NewSimpleFake(&api.ServiceList{Items: []api.Service{*newService("service0", true), *newService("service1", false)}})},
+			fakeCloud: &fake_cloud.FakeCloud{
+				Machines: []string{"node0", "node1"},
+			},
+			matchRE:               ".*",
+			expectedClientActions: nil,
+			expectedUpdateCalls:   nil,
+		},
+		{
+			// Delete "node1", target pool for "service0" should shrink.
+			fakeNodeHandler: &FakeNodeHandler{
+				Existing: []*api.Node{newNode("node0"), newNode("node1")},
+				Fake:     testclient.NewSimpleFake(&api.ServiceList{Items: []api.Service{*newService("service0", true), *newService("service1", false)}})},
+			fakeCloud: &fake_cloud.FakeCloud{
+				Machines: []string{"node0"},
+			},
+			matchRE:               ".*",
+			expectedClientActions: []testclient.FakeAction{{Action: "list-pods"}, {Action: "list-services"}},
+			expectedUpdateCalls: []fake_cloud.FakeUpdateBalancerCall{
+				{Name: "kubernetes-namespace-service0", Hosts: []string{"node0"}},
+			},
+		},
+		{
+			// Add "node1", target pool for "service0" should grow.
+			fakeNodeHandler: &FakeNodeHandler{
+				Existing: []*api.Node{newNode("node0")},
+				Fake:     testclient.NewSimpleFake(&api.ServiceList{Items: []api.Service{*newService("service0", true), *newService("service1", false)}})},
+			fakeCloud: &fake_cloud.FakeCloud{
+				Machines: []string{"node0", "node1"},
+			},
+			matchRE:               ".*",
+			expectedClientActions: []testclient.FakeAction{{Action: "list-services"}},
+			expectedUpdateCalls: []fake_cloud.FakeUpdateBalancerCall{
+				{Name: "kubernetes-namespace-service0", Hosts: []string{"node0", "node1"}},
+			},
+		},
+	}
+
+	for _, item := range table {
+		nodeController := NewNodeController(item.fakeCloud, item.matchRE, nil, &api.NodeResources{}, item.fakeNodeHandler, nil,
+			10, time.Minute, util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "kubernetes")
+		if err := nodeController.SyncCloudNodes(); err != nil {
+			t.Errorf("unexpected error: %v", err)
+		}
+		if !reflect.DeepEqual(item.expectedClientActions, item.fakeNodeHandler.Actions) {
+			t.Errorf("expected client actions mismatch, expected %+v, got %+v", item.expectedClientActions, item.fakeNodeHandler.Actions)
+		}
+		if !reflect.DeepEqual(item.expectedUpdateCalls, item.fakeCloud.UpdateCalls) {
+			t.Errorf("expected update calls mismatch, expected %+v, got %+v", item.expectedUpdateCalls, item.fakeCloud.UpdateCalls)
+		}
+	}
+}
+
 func TestPopulateNodeAddresses(t *testing.T) {
 	table := []struct {
 		nodes             *api.NodeList
@@ -640,7 +705,7 @@ func TestPopulateNodeAddresses(t *testing.T) {
 
 	for _, item := range table {
 		nodeController := NewNodeController(item.fakeCloud, ".*", nil, nil, nil, nil, 10, time.Minute,
-			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			util.NewFakeRateLimiter(), testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		result, err := nodeController.PopulateAddresses(item.nodes)
 		// In case of IP querying error, we should continue.
 		if err != nil {
@@ -840,7 +905,7 @@ func TestMonitorNodeStatusEvictPods(t *testing.T) {
 	for _, item := range table {
 		nodeController := NewNodeController(nil, "", []string{"node0"}, nil, item.fakeNodeHandler, nil, 10,
 			evictionTimeout, util.NewFakeRateLimiter(), testNodeMonitorGracePeriod,
-			testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		nodeController.now = func() util.Time { return fakeNow }
 		if err := nodeController.MonitorNodeStatus(); err != nil {
 			t.Errorf("unexpected error: %v", err)
@@ -1042,7 +1107,7 @@ func TestMonitorNodeStatusUpdateStatus(t *testing.T) {
 
 	for _, item := range table {
 		nodeController := NewNodeController(nil, "", []string{"node0"}, nil, item.fakeNodeHandler, nil, 10, 5*time.Minute, util.NewFakeRateLimiter(),
-			testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod)
+			testNodeMonitorGracePeriod, testNodeStartupGracePeriod, testNodeMonitorPeriod, "")
 		nodeController.now = func() util.Time { return fakeNow }
 		if err := nodeController.MonitorNodeStatus(); err != nil {
 			t.Errorf("unexpected error: %v", err)
@@ -1083,6 +1148,10 @@ func newPod(name, host string) *api.Pod {
 	return &api.Pod{ObjectMeta: api.ObjectMeta{Name: name}, Spec: api.PodSpec{Host: host}}
 }
 
+func newService(name string, external bool) *api.Service {
+	return &api.Service{ObjectMeta: api.ObjectMeta{Name: name, Namespace: "namespace"}, Spec: api.ServiceSpec{CreateExternalLoadBalancer: external}}
+}
+
 func sortedNodeNames(nodes []*api.Node) []string {
 	nodeNames := []string{}
 	for _, node := range nodes {
diff --git a/pkg/cloudprovider/fake/fake.go b/pkg/cloudprovider/fake/fake.go
index 624c2d61d86..664fd175be9 100644
--- a/pkg/cloudprovider/fake/fake.go
+++ b/pkg/cloudprovider/fake/fake.go
@@ -33,6 +33,12 @@ type FakeBalancer struct {
 	Hosts      []string
 }
 
+type FakeUpdateBalancerCall struct {
+	Name   string
+	Region string
+	Hosts  []string
+}
+
 // FakeCloud is a test-double implementation of Interface, TCPLoadBalancer and Instances. It is useful for testing.
 type FakeCloud struct {
 	Exists        bool
@@ -46,7 +52,7 @@ type FakeCloud struct {
 	MasterName    string
 	ExternalIP    net.IP
 	Balancers     []FakeBalancer
-
+	UpdateCalls   []FakeUpdateBalancerCall
 	cloudprovider.Zone
 }
 
@@ -105,6 +111,7 @@ func (f *FakeCloud) CreateTCPLoadBalancer(name, region string, externalIP net.IP
 // It adds an entry "update" into the internal method call record.
 func (f *FakeCloud) UpdateTCPLoadBalancer(name, region string, hosts []string) error {
 	f.addCall("update")
+	f.UpdateCalls = append(f.UpdateCalls, FakeUpdateBalancerCall{name, region, hosts})
 	return f.Err
 }
 
diff --git a/pkg/cloudprovider/gce/gce.go b/pkg/cloudprovider/gce/gce.go
index 745c8df8998..188aa34ba52 100644
--- a/pkg/cloudprovider/gce/gce.go
+++ b/pkg/cloudprovider/gce/gce.go
@@ -32,6 +32,7 @@ import (
 	"github.com/GoogleCloudPlatform/kubernetes/pkg/api"
 	"github.com/GoogleCloudPlatform/kubernetes/pkg/api/resource"
 	"github.com/GoogleCloudPlatform/kubernetes/pkg/cloudprovider"
+	"github.com/GoogleCloudPlatform/kubernetes/pkg/util"
 
 	"code.google.com/p/gcfg"
 	compute "code.google.com/p/google-api-go-client/compute/v1"
@@ -95,7 +96,12 @@ func getProjectAndZone() (string, string, error) {
 	if len(parts) != 4 {
 		return "", "", fmt.Errorf("unexpected response: %s", result)
 	}
-	return parts[1], parts[3], nil
+	zone := parts[3]
+	projectID, err := metadata.ProjectID()
+	if err != nil {
+		return "", "", err
+	}
+	return projectID, zone, nil
 }
 
 func getInstanceID() (string, error) {
@@ -292,15 +298,41 @@ func (gce *GCECloud) CreateTCPLoadBalancer(name, region string, externalIP net.I
 
 // UpdateTCPLoadBalancer is an implementation of TCPLoadBalancer.UpdateTCPLoadBalancer.
 func (gce *GCECloud) UpdateTCPLoadBalancer(name, region string, hosts []string) error {
-	var refs []*compute.InstanceReference
-	for _, host := range hosts {
-		refs = append(refs, &compute.InstanceReference{host})
+	pool, err := gce.service.TargetPools.Get(gce.projectID, region, name).Do()
+	if err != nil {
+		return err
 	}
-	req := &compute.TargetPoolsAddInstanceRequest{
-		Instances: refs,
+	existing := util.NewStringSet(pool.Instances...)
+
+	var toAdd []*compute.InstanceReference
+	var toRemove []*compute.InstanceReference
+	for _, host := range hosts {
+		link := makeHostLink(gce.projectID, gce.zone, host)
+		if !existing.Has(link) {
+			toAdd = append(toAdd, &compute.InstanceReference{link})
+		}
+		existing.Delete(link)
+	}
+	for link := range existing {
+		toRemove = append(toRemove, &compute.InstanceReference{link})
 	}
 
-	op, err := gce.service.TargetPools.AddInstance(gce.projectID, region, name, req).Do()
+	add := &compute.TargetPoolsAddInstanceRequest{
+		Instances: toAdd,
+	}
+	op, err := gce.service.TargetPools.AddInstance(gce.projectID, region, name, add).Do()
+	if err != nil {
+		return err
+	}
+	err = gce.waitForRegionOp(op, region)
+	if err != nil {
+		return err
+	}
+
+	rm := &compute.TargetPoolsRemoveInstanceRequest{
+		Instances: toRemove,
+	}
+	op, err = gce.service.TargetPools.RemoveInstance(gce.projectID, region, name, rm).Do()
 	if err != nil {
 		return err
 	}
diff --git a/pkg/registry/service/rest.go b/pkg/registry/service/rest.go
index 38b58c5dce0..f84e53ef6ad 100644
--- a/pkg/registry/service/rest.go
+++ b/pkg/registry/service/rest.go
@@ -283,10 +283,6 @@ func (rs *REST) ResourceLocation(ctx api.Context, id string) (*url.URL, http.Rou
 	return nil, nil, fmt.Errorf("no endpoints available for %q", id)
 }
 
-func (rs *REST) getLoadbalancerName(ctx api.Context, service *api.Service) string {
-	return rs.clusterName + "-" + api.NamespaceValue(ctx) + "-" + service.Name
-}
-
 func (rs *REST) createExternalLoadBalancer(ctx api.Context, service *api.Service) error {
 	if rs.cloud == nil {
 		return fmt.Errorf("requested an external service, but no cloud provider supplied.")
@@ -313,7 +309,7 @@ func (rs *REST) createExternalLoadBalancer(ctx api.Context, service *api.Service
 	if err != nil {
 		return err
 	}
-	name := rs.getLoadbalancerName(ctx, service)
+	name := cloudprovider.GetLoadBalancerName(rs.clusterName, api.NamespaceValue(ctx), service.Name)
 	var affinityType api.AffinityType = service.Spec.SessionAffinity
 	if len(service.Spec.PublicIPs) > 0 {
 		for _, publicIP := range service.Spec.PublicIPs {
@@ -386,7 +382,8 @@ func (rs *REST) deleteExternalLoadBalancer(ctx api.Context, service *api.Service
 	if err != nil {
 		return err
 	}
-	if err := balancer.DeleteTCPLoadBalancer(rs.getLoadbalancerName(ctx, service), zone.Region); err != nil {
+	name := cloudprovider.GetLoadBalancerName(rs.clusterName, api.NamespaceValue(ctx), service.Name)
+	if err := balancer.DeleteTCPLoadBalancer(name, zone.Region); err != nil {
 		return err
 	}
 	return nil