OCC DATASET V1

Purpose:

Stable benchmark corpus definition.

Corpora:

mini

Purpose:

correctness

HDFS 1GB

Purpose:

real workload

future 5GB

Purpose:

scaling

Patterns:

1B patterns

2B patterns

4B patterns

8B patterns

16B patterns

blk_* patterns

high-frequency patterns

low-frequency patterns

absent patterns

Rules:

Same corpus.

Same patterns.

Same machine spec.

Cold/warm separated.

Benchmark changes require protocol version bump.