In Cortex XDR/XSIAM, querying cold storage datasets consumes Compute Units (CU) based on the timeframe, dataset size, and query complexity. To prevent unnecessary CU consumption and speed up repetitive administrative tasks or consecutive incident lookups, Palo Alto Networks utilizes a caching architecture:
Rewarmed Data Cache: When an XQL query pulls historical data from cold storage, the retrieved logs are "rewarmed" and copied into a temporary hot storage cache.
Duration and Extensions: This rewarmed data remains instantly available in the cache for 24 hours at no additional CU cost. If you or another analyst run a subsequent query covering the same time range within that window, the 24-hour expiration timer resets. This rolling extension can be repeated up to a maximum lifetime of 7 days, after which the cache expires completely, and a brand-new retrieval from cold storage is required.