The original ticket (CASSANDRA-8894) was committed to 3.0 alpha1 two months ago. We need to get proper performance tests before GA.
See benedict's comment for more details.
Our default buffer size for (uncompressed) buffered reads should be smaller, and based on the expected record size
Compressed writer and reader should support non-compressed data.
Channel.size() is costly, mutually exclusive, and on the critical path