Cassandra非PK/无二级索引列查询与书籍描述矛盾的疑问

Great question—this is a common point of confusion, and it boils down to Cassandra's feature evolution and potentially imprecise wording in your book. Let's break it down clearly:

Why You're Seeing a Warning Instead of a Failure #

Cassandra's core design prioritizes performance and predictable query patterns. Queries that target non-primary-key columns without a secondary index require cross-partition or full-cluster scans, which can cripple performance at scale.

Instead of outright blocking these queries entirely, Cassandra uses a protective guardrail:

• By default, it throws that warning message to alert you to the severe performance risks.
• It requires you to explicitly add ALLOW FILTERING to the query to confirm you understand the tradeoffs and still want to proceed.

This isn't a "failure"—it's a balanced approach that keeps users safe from accidental performance hits while allowing flexibility for edge cases (like small-scale testing or ad-hoc checks on tiny datasets).

Why the Book Says the Operation Should Fail #

There are two likely explanations for this discrepancy:

1. Outdated Cassandra Version: If the book was written for a very early Cassandra release (pre-1.2, around 2012), those versions didn't have the ALLOW FILTERING option. Back then, Cassandra would indeed reject such queries outright with no way to override the restriction.
2. Imprecise Wording: The author might have simplified the behavior, framing "default blocked (requires explicit opt-in)" as "operation should fail." It's a common shorthand, but it misses the nuance of modern Cassandra's guardrail system.

Key Takeaway #

The ALLOW FILTERING feature was added later to give users intentional control over risky queries, replacing the hard block from older versions. Your experience matches modern Cassandra's behavior, while the book's statement is either outdated or a simplified explanation.

内容的提问来源于stack exchange，提问作者Alex Barysevich