Optimize usage for Coldline and Archive storage data

This page describes how to use Storage Intelligence advisor to optimize usage by mitigating a high rate of Class A and B operations on data in Coldline storage and Archive storage.

Coldline storage and Archive storage are intended for infrequently accessed data. These storage classes offer lower storage rates than Standard storage and Nearline storage, but they incur higher fees for data retrieval and Class A and B operations.

Storage Intelligence advisor displays the finding Spike in Class A/B operations on Coldline or Archive data when request volume increases for Coldline storage and Archive storage. A finding is an anomaly detected from your metrics that needs your attention. For an overview of Storage Intelligence advisor, see About Storage Intelligence advisor. For a full list of findings, see Findings reference catalog.

Thresholds that trigger a finding

Storage Intelligence creates a baseline for your project, folder, or organization based on historical trends. Storage Intelligence advisor generates a finding when Class A or Class B operations in Coldline storage or Archive storage increase above the baseline.

Findings reflect the previous day's activity. For example, a spike occurring on Tuesday appears in Storage Intelligence advisor on Wednesday.

Respond to a finding

  1. View the finding details.

  2. Use the following table to select the recommended strategy for your scenario:

    Scenario Recommended action Details
    Ongoing or persistent spikes in operations Enable Autoclass If you experience ongoing or persistent spikes in operations, enable Autoclass to automatically move frequently accessed objects to Standard storage. This action helps reduce retrieval fees and stabilize costs.
    Accidental or unintended data scanning Use managed folders to narrow permissions If overly permissive access, such as a misconfigured script, causes unexpected operations across buckets, use managed folders to apply granular access controls. Grant only the necessary permissions for your workloads.
    Spikes in operations on a small subset of objects Investigate with Storage Insights datasets If spikes occur on a small subset of objects, export metadata to BigQuery to identify specific objects and operations that are contributing to the spike. For example, you can identify the most frequently accessed objects within a specific prefix.
    One-time or historical events causing spikes Monitor Storage Intelligence advisor If a one-time event such as a migration or audit causes the spike, no action is necessary. Continue to monitor Storage Intelligence advisor to detect future spikes.

What's next