diff --git a/test/performance/.env.example b/test/performance/.env.example
new file mode 100644
index 000000000000..1bd197098cd2
--- /dev/null
+++ b/test/performance/.env.example
@@ -0,0 +1,19 @@
+# Clusterloader home directory (checkout https://github.com/kubernetes/perf-tests)
+export CL2_HOME_DIR="/Users/johny/perf-tests/clusterloader2"
+
+# Run the performance test with Kueue (this requires Kueue to be pre-deployed to the cluster)
+# or without Kueue
+export USE_KUEUE=false
+
+# Test iterations:
+# number-of-small-jobs number-of-large-jobs job-replica-running-time test-timeout cluster-queue-CPU-quota cluster-queue-memory-quota
+export EXPERIMENTS=(
+    "10 2 0 2s 3m 100 100Gi"
+    "20 2 0 2s 5m 100 100Gi"
+)
+
+# Kubeconfig file location
+export KUBECONFIG="$HOME/.kube/config"
+
+# Kubernetes kind
+export PROVIDER="gke"
diff --git a/test/performance/.gitignore b/test/performance/.gitignore
new file mode 100644
index 000000000000..c1499e75fc48
--- /dev/null
+++ b/test/performance/.gitignore
@@ -0,0 +1,4 @@
+*report*/
+prerequisites/cluster-queue.yaml
+tmp_manifests/
+.env
diff --git a/test/performance/README.md b/test/performance/README.md
new file mode 100644
index 000000000000..5c9ad9876b70
--- /dev/null
+++ b/test/performance/README.md
@@ -0,0 +1,67 @@
+# Kueue Performance Testing
+
+## Measurements
+
+### Job startup latency
+
+How fast do jobs transition from `created` to `started` state?
+Time spent between the transition from `job.CreationTimestamp.Time` to `job.Status.StartTime.Time` state.
+
+High Job startup latency in Kueue is expected when the total quota is not enough to schedule all jobs immediately, because the jobs need to queue.
+
+### Job startup throughput
+
+The best workload admission rate per second within 1 minute intervals. 
+The rate is measured every 5 seconds (see more details in [PromQL examples](https://prometheus.io/docs/prometheus/latest/querying/examples/#subquery)):
+
+`max_over_time(sum(rate(kueue_admitted_workloads_total{cluster_queue="{{$clusterQueue}}"}[1m]))[{{$testTimeout}}:5s])`
+
+This measurement is not accurate if the cluster quota is big enough to schedule all workloads of the test immediately, because Kueue immediately admits all the workloads and the `kueue_admitted_workloads_total` never increases. In this case, the PromQL query returns 0.
+## How to run the test?
+
+### Prerequisites
+
+1. Deploy [Kueue](https://github.com/kubernetes-sigs/kueue/blob/main/docs/setup/install.md)
+2. Make sure you have `kubectl`, [jq](https://stedolan.github.io/jq/download/), [golang version](https://github.com/mikefarah/yq) of `yq` and `go` 
+3. Checkout `Clusterloader2` framework: https://github.com/kubernetes/perf-tests and build `clusterloader` binary:
+
+    * change to `clusterloader2` directory
+    * run `go build -o clusterloader './cmd/'` 
+
+### Run the test
+
+1. Copy an environment file example to `.env` file: 
+
+    * `cp .env.example .env`
+    
+2. Edit the environment variables
+
+| Variable          | Description |
+| -----------       | ----------- |
+| CL2_HOME_DIR      | Clusterloader home directory (checkout https://github.com/kubernetes/perf-tests)       |
+| USE_KUEUE         | Run the performance test with Kueue (this requires Kueue to be pre-deployed to the cluster) or without Kueue    |
+| EXPERIMENTS       | Configuration of iterations iterations (see configuration example in the file) |
+| KUBECONFIG        | Kubeconfig file location |
+| PROVIDER          | Kubernetes kind (tested on `gke` only)
+
+3. Run the `run-test.sh` file 
+
+### Test results
+
+Every test execution creates a `report_<timestamp>` directory inside `TEST_CONFIG_DIR` with `summary.csv` file, where the following metrics are available:
+
+* P50 Job Create to start latency (ms)
+* P90 Job Create to start latency (ms)
+* P50 Job Start to complete latency (ms)
+* P90 Job Start to complete latency (ms)
+* Max Job Throughput (max jobs/s)
+* Total Jobs
+* Total Pods
+* Duration (s)
+
+Additionally, the following metrics are added to the results only for reference. Kueue doesn't influence them directly.
+
+* Avg Pod Waiting time (s)
+* P90 Pod Waiting time (s)
+* Avg Pod Completion time (s)
+* P90 Pod Completion time (s)
diff --git a/test/performance/config.yaml b/test/performance/config.yaml
new file mode 100644
index 000000000000..1e4d361c6342
--- /dev/null
+++ b/test/performance/config.yaml
@@ -0,0 +1,181 @@
+{{$MODE := DefaultParam .MODE "Indexed"}}
+{{$LOAD_TEST_THROUGHPUT := DefaultParam .CL2_LOAD_TEST_THROUGHPUT 10}}
+
+{{$smallJobs := DefaultParam .CL2_SMALL_JOBS 10}}
+{{$mediumJobs := DefaultParam .CL2_MEDIUM_JOBS 2}}
+{{$largeJobs := DefaultParam .CL2_LARGE_JOBS 0}}
+
+{{$namespaces := DefaultParam .CL2_NAMESPACES 1}}
+
+{{$smallJobsPerNamespace := DivideInt $smallJobs $namespaces}}
+{{$mediumJobsPerNamespace := DivideInt $mediumJobs $namespaces}}
+{{$largeJobsPerNamespace := DivideInt $largeJobs $namespaces}}
+
+{{$smallJobSize := 5}}
+{{$mediumJobSize := 20}}
+{{$largeJobSize := 100}}
+
+{{$jobRunningTime := DefaultParam .CL2_JOB_RUNNING_TIME "30s"}}
+
+{{$clusterQueue := "default-cluster-queue"}}
+{{$localQueue := "local-queue"}}
+
+{{$testTimeout := DefaultParam .CL2_TEST_TIMEOUT "5m"}}
+
+{{$namespacePrefix := "queue-test"}}
+
+{{$useKueue := DefaultParam .CL2_USE_KUEUE false}}
+
+name: batch
+
+namespace:
+  number: {{$namespaces}}
+  prefix: {{$namespacePrefix}}
+
+tuningSets:
+- name: UniformQPS
+  qpsLoad:
+    qps: {{$LOAD_TEST_THROUGHPUT}}
+
+steps:
+- name: Start measurements
+  measurements:
+  - Identifier: Timer
+    Method: Timer
+    Params:
+      action: start
+      label: job_performance  
+  - Identifier: WaitForFinishedJobs
+    Method: WaitForFinishedJobs
+    Params:
+      action: start
+      labelSelector: group = test-job
+  - Identifier: JobLifecycleLatency
+    Method: JobLifecycleLatency
+    Params:
+      action: start
+      labelSelector: group = test-job
+  - Identifier: GenericPrometheusQuery
+    Method: GenericPrometheusQuery
+    Params:
+      action: start
+      metricName: Job (Kueue) API performance
+      metricVersion: v1
+      unit: s
+      queries:
+        - name: total_jobs_scheduled
+          query: count(kube_job_info{namespace=~"{{$namespacePrefix}}.*"})
+        - name: total_pods_scheduled
+          query: count(kube_pod_info{namespace=~"{{$namespacePrefix}}.*"})
+        - name: avg_pod_running_time
+          query: (avg(kube_pod_completion_time{namespace=~"{{$namespacePrefix}}.*"} - kube_pod_start_time{namespace=~"{{$namespacePrefix}}.*"}))
+        - name: perc_90_pod_completion_time
+          query: quantile(0.90, kube_pod_completion_time{namespace=~"{{$namespacePrefix}}.*"} - kube_pod_start_time{namespace=~"{{$namespacePrefix}}.*"})
+        - name: avg_pod_waiting_time
+          query: (avg(kube_pod_start_time{namespace=~"{{$namespacePrefix}}.*"} - kube_pod_created{namespace=~"{{$namespacePrefix}}.*"})) 
+        - name: perc_90_pod_waiting_time
+          query: quantile(0.90, kube_pod_start_time{namespace=~"{{$namespacePrefix}}.*"} - kube_pod_created{namespace=~"{{$namespacePrefix}}.*"})                
+        - name: max_job_throughput
+          query: max_over_time(sum(rate(kueue_admitted_workloads_total{cluster_queue="{{$clusterQueue}}"}[1m]))[{{$testTimeout}}:5s])                               
+- name: Sleep
+  measurements:
+  - Identifier: sleep
+    Method: Sleep
+    Params:
+      duration: 10s  
+{{if $useKueue}}
+- name: Create local queue
+  phases: 
+  - namespaceRange:
+      min: 1
+      max: {{$namespaces}}
+    replicasPerNamespace: 1
+    tuningSet: UniformQPS
+    objectBundle:
+    - basename: {{$localQueue}}
+      objectTemplatePath: "local-queue.yaml"
+      templateFillMap:
+        ClusterQueue: {{$clusterQueue}}     
+{{end}}
+- name: Create {{$MODE}} jobs
+  phases:
+  - namespaceRange:
+      min: 1
+      max: {{$namespaces}}
+    replicasPerNamespace: {{$smallJobsPerNamespace}}
+    tuningSet: UniformQPS
+    objectBundle:
+    - basename: small
+      objectTemplatePath: "job.yaml"
+      templateFillMap:
+        UseKueue: {{$useKueue}}
+        Replicas: {{$smallJobSize}}
+        Mode: {{$MODE}}
+        Sleep: {{$jobRunningTime}}
+        LocalQueue: "{{$localQueue}}-0"        
+  - namespaceRange:
+      min: 1
+      max: {{$namespaces}}
+    replicasPerNamespace: {{$mediumJobsPerNamespace}}
+    tuningSet: UniformQPS
+    objectBundle:
+    - basename: medium
+      objectTemplatePath: "job.yaml"
+      templateFillMap:
+        UseKueue: {{$useKueue}} 
+        Replicas: {{$mediumJobSize}}
+        Mode: {{$MODE}}
+        Sleep: {{$jobRunningTime}}
+        LocalQueue: "{{$localQueue}}-0"
+  - namespaceRange:
+      min: 1
+      max: {{$namespaces}}
+    replicasPerNamespace: {{$largeJobsPerNamespace}}
+    tuningSet: UniformQPS
+    objectBundle:
+    - basename: large
+      objectTemplatePath: "job.yaml"
+      templateFillMap:
+        UseKueue: {{$useKueue}}
+        Replicas: {{$largeJobSize}}
+        Mode: {{$MODE}}
+        Sleep: {{$jobRunningTime}}
+        LocalQueue: "{{$localQueue}}-0"                
+- name: Wait for {{$MODE}} jobs to finish
+  measurements:
+  - Identifier: JobLifecycleLatency
+    Method: JobLifecycleLatency
+    Params:
+      action: gather
+      timeout: {{$testTimeout}}       
+  - Identifier: WaitForFinishedJobs
+    Method: WaitForFinishedJobs
+    Params:
+      action: gather
+      timeout: {{$testTimeout}}
+- name: Stop Timer  
+  measurements: 
+  - Identifier: Timer
+    Method: Timer 
+    Params:
+      action: stop
+      label: job_performance  
+- name: Gather Timer  
+  measurements: 
+  - Identifier: Timer
+    Method: Timer 
+    Params:
+      action: gather            
+- name: Sleep
+  measurements:
+  - Identifier: sleep
+    Method: Sleep
+    Params:
+      duration: 30s            
+- name: Gather Prometheus measurements
+  measurements:
+  - Identifier: GenericPrometheusQuery
+    Method: GenericPrometheusQuery
+    Params:
+      action: gather
+      enableViolations: true      
diff --git a/test/performance/job.yaml b/test/performance/job.yaml
new file mode 100644
index 000000000000..ce82f3be0b41
--- /dev/null
+++ b/test/performance/job.yaml
@@ -0,0 +1,30 @@
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: {{.Name}}
+  labels:
+    group: test-job
+  {{if .UseKueue}}
+  annotations:
+    kueue.x-k8s.io/queue-name: {{.LocalQueue}}
+  {{end}}
+spec:
+  suspend: {{.UseKueue}}
+  parallelism: {{.Replicas}}
+  completions: {{.Replicas}}
+  completionMode: {{.Mode}}
+  template:
+    metadata:
+      labels:
+        group: test-pod
+    spec:
+      containers:
+      - name: {{.Name}}
+        image: gcr.io/k8s-staging-perf-tests/sleep:v0.0.3
+        args:
+          - {{.Sleep}}
+        resources:
+          requests:
+            cpu: "200m"
+            memory: "100Mi"          
+      restartPolicy: Never
diff --git a/test/performance/local-queue.yaml b/test/performance/local-queue.yaml
new file mode 100644
index 000000000000..88422fcb181d
--- /dev/null
+++ b/test/performance/local-queue.yaml
@@ -0,0 +1,6 @@
+apiVersion: kueue.x-k8s.io/v1alpha2
+kind: LocalQueue
+metadata:
+  name: {{.Name}}
+spec:
+  clusterQueue: {{.ClusterQueue}}
diff --git a/test/performance/prerequisites/cluster-queue.template b/test/performance/prerequisites/cluster-queue.template
new file mode 100644
index 000000000000..515b263ac6d0
--- /dev/null
+++ b/test/performance/prerequisites/cluster-queue.template
@@ -0,0 +1,17 @@
+apiVersion: kueue.x-k8s.io/v1alpha2
+kind: ClusterQueue
+metadata:
+  name: default-cluster-queue
+spec:
+  namespaceSelector: {}
+  resources:
+    - name: "cpu"
+      flavors:
+        - name: default
+          quota:
+            min: 100
+    - name: "memory"
+      flavors:
+        - name: default
+          quota:
+            min: 50Gi
diff --git a/test/performance/prerequisites/resource-flavor.yaml b/test/performance/prerequisites/resource-flavor.yaml
new file mode 100644
index 000000000000..8937e21cb036
--- /dev/null
+++ b/test/performance/prerequisites/resource-flavor.yaml
@@ -0,0 +1,4 @@
+apiVersion: kueue.x-k8s.io/v1alpha2
+kind: ResourceFlavor
+metadata:
+  name: default
diff --git a/test/performance/run-test.sh b/test/performance/run-test.sh
new file mode 100755
index 000000000000..b1f9628b2878
--- /dev/null
+++ b/test/performance/run-test.sh
@@ -0,0 +1,115 @@
+#!/bin/bash
+
+# shellcheck disable=SC1091
+if test -f .env; then
+  source .env
+fi
+
+CL2_HOME_DIR=${CL2_HOME_DIR:=/Users/johny/perf-tests/clusterloader2} 
+CL2_BINARY_NAME=${CL2_BINARY_NAME:=clusterloader}
+
+USE_KUEUE=${USE_KUEUE:=true}
+
+DEFAULT_EXPERIMENTS=(
+    "10 2 0 2s 3m 100 100Gi"
+)
+
+EXPERIMENTS=("${EXPERIMENTS[@]:=${DEFAULT_EXPERIMENTS[@]}}")
+
+KUBECONFIG=${KUBECONFIG:="$HOME/.kube/config"}
+export KUBECONFIG
+
+PROVIDER=${PROVIDER:=gke}
+
+export EXEC_DEPLOYMENT_YAML="$CL2_HOME_DIR/pkg/execservice/manifest/exec_deployment.yaml"
+
+cp -r "$CL2_HOME_DIR/pkg/prometheus/manifests/" tmp_manifests
+trap 'rm -r tmp_manifests' EXIT
+export PROMETHEUS_MANIFEST_PATH=$(pwd)/tmp_manifests
+
+now=$(date +%Y-%m-%d-%H.%M.%S)
+
+if [[ "$USE_KUEUE" == true ]]; then
+    export CL2_USE_KUEUE=true
+    # Kustomize places all Kubernetes object manifests (role, rolebinding and servicemonitor) in the same file, 
+    # however, Clusterloader expects, that there is one manifest per file, otherwise it does not create all the 
+    # objects from the file. 
+    # The yq expression below splits produced manifest into 3 files and then moves 
+    # to temporary $PROMETHEUS_MANIFEST_PATH
+    kubectl kustomize ../../config/prometheus | yq -s '.kind' -o yaml
+    mv Role.yml "$PROMETHEUS_MANIFEST_PATH/prometheus-kueue-role.yaml"
+    mv RoleBinding.yml "$PROMETHEUS_MANIFEST_PATH/prometheus-kueue-role-binding.yaml"
+    mv ServiceMonitor.yml "$PROMETHEUS_MANIFEST_PATH/prometheus-kueue-service-monitor.yaml"
+    kubectl apply -f prerequisites/resource-flavor.yaml
+    report_dir_name="kueue_report_$now"
+else 
+    report_dir_name="report_$now"
+fi
+mkdir -p "$report_dir_name"
+
+{
+    echo -ne "Test Arguments,"
+    echo -ne "P50 Job Create to start latency (ms),"
+    echo -ne "P90 Job Create to start latency (ms),"
+    echo -ne "P50 Job Start to complete latency (ms),"
+    echo -ne "P90 Job Start to complete latency (ms),"
+    echo -ne "Max Job Throughput (max jobs/s),"
+    echo -ne "Total Jobs,"
+    echo -ne "Total Pods,"
+    echo -ne "Duration (s),"
+    echo -ne "Avg Pod Waiting time (s),"
+    echo -ne "P90 Pod Waiting time (s),"
+    echo -ne "Avg Pod Completion time (s),"
+    echo "P90 Pod Completion time (s)"
+} >>"$report_dir_name/summary.csv"
+
+for item in "${EXPERIMENTS[@]}"; do
+    IFS=" " read -ra conditions <<<"$item"
+    export CL2_SMALL_JOBS="${conditions[0]}"
+    export CL2_MEDIUM_JOBS="${conditions[1]}"
+    export CL2_LARGE_JOBS="${conditions[2]}"
+    export CL2_JOB_RUNNING_TIME="${conditions[3]}"
+    export CL2_TEST_TIMEOUT="${conditions[4]}"
+    cores="${conditions[5]}"
+    memory="${conditions[6]}"
+    experiment_dir="$report_dir_name/$CL2_SMALL_JOBS-$CL2_MEDIUM_JOBS-$CL2_LARGE_JOBS-$CL2_JOB_RUNNING_TIME-$CL2_TEST_TIMEOUT-$cores-$memory"
+    mkdir -p "$experiment_dir"
+    echo "======================================================================================"
+    echo "Running an experiment with [$CL2_SMALL_JOBS, $CL2_MEDIUM_JOBS, $CL2_LARGE_JOBS, $CL2_JOB_RUNNING_TIME, $CL2_TEST_TIMEOUT, $cores, $memory]"
+    if [[ "$USE_KUEUE" == true ]]; then
+        cp prerequisites/cluster-queue.template prerequisites/cluster-queue.yaml        
+        yq -i e ".spec.resources[0].flavors[0].quota.min=$cores" prerequisites/cluster-queue.yaml
+        yq -i e ".spec.resources[1].flavors[0].quota.min=\"$memory\"" prerequisites/cluster-queue.yaml
+        kubectl apply -f prerequisites/cluster-queue.yaml
+    fi
+    "$CL2_HOME_DIR/$CL2_BINARY_NAME" \
+        --testconfig=config.yaml \
+        --enable-prometheus-server=true \
+        --provider="$PROVIDER" \
+        --v=2 --prometheus-scrape-metrics-server=true \
+        --prometheus-scrape-kube-state-metrics=true \
+        --report-dir="$experiment_dir"
+    echo "Experiment finished. Extracting results from the report..."
+    {
+        echo -ne "$CL2_SMALL_JOBS $CL2_MEDIUM_JOBS $CL2_LARGE_JOBS $CL2_JOB_RUNNING_TIME $CL2_TEST_TIMEOUT $cores $memory,"
+        echo -ne "$(jq '.dataItems[] | select(.labels.Metric=="create_to_start").data.Perc50' "$experiment_dir"/JobLifecycleLatency*.json),"
+        echo -ne "$(jq '.dataItems[] | select(.labels.Metric=="create_to_start").data.Perc90' "$experiment_dir"/JobLifecycleLatency*.json),"
+        echo -ne "$(jq '.dataItems[] | select(.labels.Metric=="start_to_complete").data.Perc50' "$experiment_dir"/JobLifecycleLatency*.json),"
+        echo -ne "$(jq '.dataItems[] | select(.labels.Metric=="start_to_complete").data.Perc90' "$experiment_dir"/JobLifecycleLatency*.json),"
+        echo -ne "$(jq '.dataItems[0].data.max_job_throughput' "$experiment_dir"/GenericPrometheusQuery*.json),"
+        echo -ne "$(jq '.dataItems[0].data.total_jobs_scheduled' "$experiment_dir"/GenericPrometheusQuery*.json),"
+        echo -ne "$(jq '.dataItems[0].data.total_pods_scheduled' "$experiment_dir"/GenericPrometheusQuery*.json),"
+        echo -ne "$(jq '.dataItems[0].data.job_performance' "$experiment_dir"/Timer*.json)",
+        echo -ne "$(jq '.dataItems[0].data.avg_pod_waiting_time' "$experiment_dir"/GenericPrometheusQuery*.json),"
+        echo -ne "$(jq '.dataItems[0].data.perc_90_pod_waiting_time' "$experiment_dir"/GenericPrometheusQuery*.json),"
+        echo -ne "$(jq '.dataItems[0].data.avg_pod_running_time' "$experiment_dir"/GenericPrometheusQuery*.json),"
+        jq '.dataItems[0].data.perc_90_pod_completion_time' "$experiment_dir"/GenericPrometheusQuery*.json
+    } >>"$report_dir_name/summary.csv"
+    if [[ "$USE_KUEUE" == true ]]; then
+        kubectl delete -f prerequisites/cluster-queue.yaml
+    fi
+done
+
+if [[ "$USE_KUEUE" == true ]]; then
+    kubectl delete -f prerequisites/resource-flavor.yaml
+fi