Dataset and Model

Motivation

Currently, the Edge AI features depend on the object dataset and model.

This proposal provides the definitions of dataset and model as the first class of k8s resources.

Goals

  • Metadata of dataset and model objects.

  • Used by the Edge AI features

Non-goals

  • The truly format of the AI dataset, such as imagenet, coco or tf-record etc.

  • The truly format of the AI model, such as ckpt, saved_model of tensorflow etc.

  • The truly operations of the AI dataset, such as shuffle, crop etc.

  • The truly operations of the AI model, such as train, inference etc.

Proposal

We propose using Kubernetes Custom Resource Definitions (CRDs) to describe the dataset/model specification/status and a controller to synchronize these updates between edge and cloud.

Use Cases

  • Users can create the dataset resource, by providing the dataset url, format and the nodeName which owns the dataset.

  • Users can create the model resource by providing the model url and format.

  • Users can show the information of dataset/model.

  • Users can delete the dataset/model.

Design Details

CRD API Group and Version

The Dataset and Model CRDs will be namespace-scoped. The tables below summarize the group, kind and API version details for the CRDs.

  • Dataset

Field

Description

Group

sedna.io

APIVersion

v1alpha1

Kind

Dataset

  • Model

Field

Description

Group

sedna.io

APIVersion

v1alpha1

Kind

Model

CRDs

Dataset CRD

crd source

apiVersion: apiextensions.k8s.io/v1
kind: CustomResourceDefinition
metadata:
  name: datasets.sedna.io
spec:
  group: sedna.io
  names:
    kind: Dataset
    plural: datasets
  scope: Namespaced
  versions:
    - name: v1alpha1
      subresources:
        # status enables the status subresource.
        status: {}
      served: true
      storage: true
      schema:
        openAPIV3Schema:
          type: object
          properties:
            spec:
              type: object
              required:
                - url
                - format
              properties:
                url:
                  type: string
                format:
                  type: string
                nodeName:
                  type: string
            status:
              type: object
              properties:
                numberOfSamples:
                  type: integer
                updateTime:
                  type: string
                  format: datatime


      additionalPrinterColumns:
        - name: NumberOfSamples
          type: integer
          description: The number of samples in the dataset
          jsonPath: ".status.numberOfSamples"
        - name: Node
          type: string
          description: The node name of the dataset
          jsonPath: ".spec.nodeName"
        - name: spec
          type: string
          description: The spec of the dataset
          jsonPath: ".spec"

  1. format of dataset

We use this field to report the number of samples for the dataset and do dataset splitting.

Current we support these below formats:

  • txt: one nonempty line is one sample

Model CRD

crd source

apiVersion: apiextensions.k8s.io/v1
kind: CustomResourceDefinition
metadata:
  name: models.sedna.io
spec:
  group: sedna.io
  names:
    kind: Model
    plural: models
  scope: Namespaced
  versions:
    - name: v1alpha1
      subresources:
        # status enables the status subresource.
        status: {}
      served: true
      storage: true
      schema:
        openAPIV3Schema:
          type: object
          properties:
            spec:
              type: object
              required:
                - url
                - format
              properties:
                url:
                  type: string
                format:
                  type: string
            status:
              type: object
              properties:
                updateTime:
                  type: string
                  format: datetime
                metrics:
                  type: array
                  items:
                    type: object
                    properties:
                      key:
                        type: string
                      value:
                        type: string


      additionalPrinterColumns:
        - name: updateAGE
          type: date
          description: The update age
          jsonPath: ".status.updateTime"
        - name: metrics
          type: string
          description: The metrics
          jsonPath: ".status.metrics"

CRD type definition

  • Dataset

go source

package v1alpha1

import (
    metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
)

// +genclient
// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object

// Dataset describes the data that a dataset resource should have
type Dataset struct {
    metav1.TypeMeta `json:",inline"`

    metav1.ObjectMeta `json:"metadata,omitempty"`

    Spec   DatasetSpec   `json:"spec"`
    Status DatasetStatus `json:"status"`
}

// DatasetSpec is a description of a dataset
type DatasetSpec struct {
    URL  string `json:"url"`
    Format   string `json:"format"`
    NodeName string `json:"nodeName"`
}

// DatasetStatus represents information about the status of a dataset
// including the time a dataset updated, and number of samples in a dataset
type DatasetStatus struct {
    UpdateTime      *metav1.Time `json:"updateTime,omitempty" protobuf:"bytes,1,opt,name=updateTime"`
    NumberOfSamples int          `json:"numberOfSamples"`
}

// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object

// DatasetList is a list of Datasets
type DatasetList struct {
    metav1.TypeMeta `json:",inline"`
    metav1.ListMeta `json:"metadata"`

    Items []Dataset `json:"items"`
}

  • Model

go source

package v1alpha1

import (
    metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
)

// +genclient
// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object

// Model describes the data that a model resource should have
type Model struct {
    metav1.TypeMeta `json:",inline"`

    metav1.ObjectMeta `json:"metadata,omitempty"`

    Spec   ModelSpec   `json:"spec"`
    Status ModelStatus `json:"status"`
}

// ModelSpec is a description of a model
type ModelSpec struct {
    URL string `json:"url"`
    Format   string `json:"format"`
}

// ModelStatus represents information about the status of a model
// including the time a model updated, and metrics in a model
type ModelStatus struct {
    UpdateTime *metav1.Time `json:"updateTime,omitempty" protobuf:"bytes,1,opt,name=updateTime"`
    Metrics    []Metric     `json:"metrics,omitempty" protobuf:"bytes,2,rep,name=metrics"`
}

// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object

//  ModelList is a list of Models
type ModelList struct {
    metav1.TypeMeta `json:",inline"`
    metav1.ListMeta `json:"metadata"`

    Items []Model `json:"items"`
}

Crd samples

  • Dataset

apiVersion: sedna.io/v1alpha1
kind: Dataset
metadata:
  name: "dataset-examp"
spec:
  url: "/code/data"
  format: "txt"
  nodeName: "edge0"

  • Model

apiVersion: sedna.io/v1alpha1
kind: Model
metadata:
  name: model-examp
spec:
  url: "/model/frozen.pb"
  format: pb

Controller Design

In the current design there is downstream/upstream controller for dataset, no downstream/upstream controller for model.

The dataset controller synchronizes the dataset between the cloud and edge.

  • downstream: synchronize the dataset info from the cloud to the edge node.

  • upstream: synchronize the dataset status from the edge to the cloud node, such as the information how many samples the dataset has.

Here is the flow of the dataset creation:

For the model:

  1. Model’s info will be synced when sync the federated-task etc which uses the model.

  2. Model’s status will be updated when the corresponding training/inference work has completed.