feat: ycsb support request distribution (#60)

Author: fengluodb

api/v1alpha1/ycsb_types.go

@@ -72,6 +72,30 @@ type YcsbSpec struct {
 	// +optional
 	ScanProportion int `json:"scanProportion,omitempty"`
 
+	// the distribution of requests accross the keyspace.
+	// uniform: each key has an equal probability of being accessed.
+	// sequential: keys are accessed in sequential order.
+	// zipfian: some keys are accessed more frequently than others.
+	// latest: the most recently inserted keys are accessed more frequently.
+	// hotspot: a small number of keys are accessed more frequently.
+	// exponential: keys are accessed in an exponential distribution.
+	// +kubebuilder:validation:Enum={uniform,sequential,zipfian,latest,hotspot,exponential}
+	// +kubebuilder:default=uniform
+	// +optional
+	RequestDistribution string `json:"requestDistribution,omitempty"`
+
+	// the distribution of scan lengths
+	// +kubebuilder:validation:Enum={uniform,zipfian}
+	// +kubebuilder:default=uniform
+	// +optional
+	ScanLengthDistribution string `json:"scanLengthDistribution,omitempty"`
+
+	// the distribution of field lengths
+	// +kubebuilder:validation:Enum={constant,uniform,zipfian,histogram}
+	// +kubebuilder:default=constant
+	// +optional
+	FieldLengthDistribution string `json:"fieldLengthDistribution,omitempty"`
+
 	// the number of threads
 	// +kubebuilder:validation:MinItems=1
 	// +kubebuilder:default={1}

config/crd/bases/benchmark.apecloud.io_ycsbs.yaml

@@ -42,6 +42,14 @@ spec:
                 items:
                   type: string
                 type: array
+              fieldLengthDistribution:
+                default: constant
+                enum:
+                - constant
+                - uniform
+                - zipfian
+                - histogram
+                type: string
               insertProportion:
                 default: 0
                 maximum: 100
@@ -65,6 +73,16 @@ spec:
                 default: 10000
                 minimum: 1
                 type: integer
+              requestDistribution:
+                default: uniform
+                enum:
+                - uniform
+                - sequential
+                - zipfian
+                - latest
+                - hotspot
+                - exponential
+                type: string
               resourceLimits:
                 properties:
                   cpu:
@@ -79,6 +97,12 @@ spec:
                   memory:
                     type: string
                 type: object
+              scanLengthDistribution:
+                default: uniform
+                enum:
+                - uniform
+                - zipfian
+                type: string
               scanProportion:
                 default: 0
                 maximum: 100

deploy/helm/crds/benchmark.apecloud.io_ycsbs.yaml

@@ -42,6 +42,14 @@ spec:
                 items:
                   type: string
                 type: array
+              fieldLengthDistribution:
+                default: constant
+                enum:
+                - constant
+                - uniform
+                - zipfian
+                - histogram
+                type: string
               insertProportion:
                 default: 0
                 maximum: 100
@@ -65,6 +73,16 @@ spec:
                 default: 10000
                 minimum: 1
                 type: integer
+              requestDistribution:
+                default: uniform
+                enum:
+                - uniform
+                - sequential
+                - zipfian
+                - latest
+                - hotspot
+                - exponential
+                type: string
               resourceLimits:
                 properties:
                   cpu:
@@ -79,6 +97,12 @@ spec:
                   memory:
                     type: string
                 type: object
+              scanLengthDistribution:
+                default: uniform
+                enum:
+                - uniform
+                - zipfian
+                type: string
               scanProportion:
                 default: 0
                 maximum: 100

internal/controller/ycsb_job.go

@@ -82,6 +82,9 @@ func NewYcsbPrepareJobs(cr *v1alpha1.Ycsb) []*batchv1.Job {
 	cmd = fmt.Sprintf("%s -p recordcount=%d", cmd, cr.Spec.RecordCount)
 	cmd = fmt.Sprintf("%s -p operationcount=%d", cmd, cr.Spec.OperationCount)
 	cmd = fmt.Sprintf("%s -p threadcount=%d", cmd, cr.Spec.Threads[0])
+	cmd = fmt.Sprintf("%s -p requestdistribution=%s", cmd, cr.Spec.RequestDistribution)
+	cmd = fmt.Sprintf("%s -p scanlengthdistribution=%s", cmd, cr.Spec.ScanLengthDistribution)
+	cmd = fmt.Sprintf("%s -p fieldlengthdistribution=%s", cmd, cr.Spec.FieldLengthDistribution)
 	cmd = fmt.Sprintf("%s %s", cmd, strings.Join(cr.Spec.ExtraArgs, " "))
 	cmd = fmt.Sprintf("%s 2>&1 | tee /var/log/ycsb.log", cmd)
 
@@ -115,6 +118,9 @@ func NewYcsbRunJobs(cr *v1alpha1.Ycsb) []*batchv1.Job {
 	cmd = fmt.Sprintf("%s %s", cmd, NewYcsbWorkloadParams(cr))
 	cmd = fmt.Sprintf("%s -p recordcount=%d", cmd, cr.Spec.RecordCount)
 	cmd = fmt.Sprintf("%s -p operationcount=%d", cmd, cr.Spec.OperationCount)
+	cmd = fmt.Sprintf("%s -p requestdistribution=%s", cmd, cr.Spec.RequestDistribution)
+	cmd = fmt.Sprintf("%s -p scanlengthdistribution=%s", cmd, cr.Spec.ScanLengthDistribution)
+	cmd = fmt.Sprintf("%s -p fieldlengthdistribution=%s", cmd, cr.Spec.FieldLengthDistribution)
 	cmd = fmt.Sprintf("%s %s", cmd, strings.Join(cr.Spec.ExtraArgs, " "))
 
 	totalProportion := cr.Spec.ReadProportion + cr.Spec.UpdateProportion + cr.Spec.InsertProportion + cr.Spec.ReadModifyWriteProportion + cr.Spec.ScanProportion