NetworkUsage Class

Definition

public ref class NetworkUsage sealed

/// [Windows.Foundation.Metadata.ContractVersion(Windows.Foundation.UniversalApiContract, 65536)]
/// [Windows.Foundation.Metadata.MarshalingBehavior(Windows.Foundation.Metadata.MarshalingType.Agile)]
class NetworkUsage final

[Windows.Foundation.Metadata.ContractVersion(typeof(Windows.Foundation.UniversalApiContract), 65536)]
[Windows.Foundation.Metadata.MarshalingBehavior(Windows.Foundation.Metadata.MarshalingType.Agile)]
public sealed class NetworkUsage

Public NotInheritable Class NetworkUsage

Inheritance

Object Platform::Object IInspectable NetworkUsage

Attributes

ContractVersionAttribute MarshalingBehaviorAttribute

Windows requirements

Remarks

Each NetworkUsage entry represents aggregated usage over a contiguous interval defined by the granularity supplied to ConnectionProfile.GetNetworkUsageAsync.

Usage considerations

• Sequence: Entries are chronological; the last entry can be a partial interval if the span does not align with the granularity boundary.
• Sparsity: Zero-traffic intervals can be omitted depending on provider data; do not assume a dense, gap-free time series.
• Estimation latency: Byte counts can reflect accounting delay; avoid sub-minute quota or throttling decisions.
• Empty list: Interpret as "no recorded usage in window" (not an error). Re-query later or widen the interval.
• Aggregation: Sum BytesSent + BytesReceived (use 64-bit integers). Discard large raw lists after summarizing to reduce memory.
• Partial trailing interval: If the end time falls inside an open bucket its values are provisional and may grow; re-query later for a finalized total.
• Reconciliation with attributed usage: Summed AttributedNetworkUsage can be less than or equal to aggregate NetworkUsage; the delta can represent unattributed system traffic or accounting alignment.
• Multiple profiles: Query each relevant ConnectionProfile separately; this type never merges across profiles or interfaces. Aggregate in app code for a device-wide view.
• Sliding window collection: Track the last fully closed bucket (aligned to granularity) and only commit closed buckets to avoid double counting when polling.
• Overflow safety: Use checked 64-bit accumulation over long historical ranges; never downcast to 32-bit.