Search speed matters because investigations are a chain of questions, and each slow answer stretches the whole incident. An analyst asks where a login came from, then what else that IP touched, then whether the same user did anything unusual last month. If each search takes twenty minutes, or older data needs a restore request first, investigations stall and analysts stop asking the questions that would have found the attacker.
The goal is for a search across a year of data to feel about as fast as a search across a day.
Rehydration is the process of restoring archived data to a searchable tier before you can query it. Many SIEMs and archives work this way for older data. It is slow, often takes a request or a ticket and can carry its own cost. In practice, teams avoid it, which means older data might as well not exist. Designs that keep all data directly searchable remove that barrier. See hot, warm and cold security data.
Is a data lake slower to search than a SIEM?
Not necessarily. A well-designed security data lake with columnar storage, partitioning and indexing can search large historical ranges quickly, often faster than a SIEM that must rehydrate archived data.
How should I test search performance?
Run your real investigation queries, such as an IP across all sources for 12 months, against realistic data volumes before you buy or migrate.
The Anomali Unified Security Data Lake keeps full-fidelity telemetry searchable for long periods without cold-storage retrieval delays, with data normalized and enriched at ingest.
Read Full fidelity at cold storage cost or request a demo.