← Analytics / analytics / graphql-api / features

Datasets (tables)

Cloudflare Analytics offers a range of datasets, including both general and product-specific ones. Datasets use a consistent naming scheme that explicitly identifies the type of data they return:

• Domain - Each dataset is named after the field it describes and is associated with a set of nodes. Product-specific data nodes incorporate the name of the relevant product, for instance loadBalancingRequests* nodes.

• Adaptive Sampling - Nodes that represent data acquired using adaptive sampling incorporate the Adaptive suffix. For more details, refer to sampling.

• Aggregated data - Nodes that represent aggregated data include the Groups suffix. For example, the loadBalancingRequestsAdaptiveGroups node represents aggregated data for Load Balancing requests. Aggregated data is returned in an array of ...Group objects. Please note: we have a node that currently excluded from that naming convention - workersInvocationsAdaptive (beta).

• Raw data - Raw data nodes, such as loadBalancingRequestsAdaptive, are not aggregated and so do not incorporate the Groups suffix. Raw data is returned in arrays containing objects of the relevant data type. For example, a query to loadBalancingRequestsAdaptive returns a variety of LoadBalancingRequest objects.

To find out more information about datasets, availability, beta, and deprecation statuses, please refer to GraphQL discovery features.

This example illustrates the structure for Groups:

type WhateverGroup {
    count # No subfields, it is just the group size. Not available for roll-up tables.
    sum {
        # fields that support summing (numbers, maps of numbers)
    }
    avg {
        # fields that support averaging (numbers)
    }
    uniq {
        # fields that support uniqueing (numbers, strings, enums, IPs, dates, etc.)
    }
}

Unique values are not available as a dimension but can be queried as demonstrated in this example:

{
  # Get number of bytes and unique IPs in each minute.
  httpRequests1mGroups {
    sum {
      bytes
    }
    uniq {
      uniques # unique IPs
    }
    dimensions {
      datetimeMinute
    }
  }

  # Count the number of events in each hour.
  firewallEventsAdaptiveGroups {
    count
    dimensions {
      datetimeHour
    }
  }
}

Every exposed table has a GraphQL type definition. Type definitions observe the following rules:

• Regular fields represent themselves.
• Every field, including nested fields, has a type and represents a list of that type.
• The enum type represents an enumerated field.

Here is an example type definition for ContentTypeMapElem:

type ContentTypeMapElem {
    edgeResponseContentType: UInt32!
    requests: UInt64!
    bytes: UInt64!
}

# An array of httpRequestsGroup is the result of httpRequests1hGroups or
# httpRequests1mGroups query.
type httpRequestsGroup {
    date: Date!
    timeslot: DateTime!
    requests: UInt64!
    contentTypeMap: [ContentTypeMapElem!]!
    # ... other fields
}

enum TrustedClientCategory {
    UNKNOWN
    REAL_BROWSER
    HONEST_BOT
}

# An array of Request is the result of httpRequests query.
type Request {
    trustedClientCategory: TrustedClientCategory!
    # ... other fields
}