Ivan Balashov
Reputation: 1963
HDFS random read and disk IO
When we read a single byte from HDFS file, does it mean a full HDFS block gets read from the disk somewhere on the cluster (assuming no caching involved)?
Upvotes: 0
Views: 530
Answers (1)
Ivan Balashov
Reputation: 1963
The answer is no.
This is affected (may be not exclusively) by dfs.datanode.readahead.bytes parameter, which defaults to 4MB in hadoop v2.2.0
Upvotes: 1
Related Questions
- How does HDFS write to a disk on the data node
- What is meant by "HDFS lacks random read and write access"?
- Random Read/Writes in HDFS
- Hadoop HDFS: Read/Write parallelism?
- Why HDFS is write once and read multiple times?
- I/O time when reading from HDFS in Hadoop
- what's the HDFS writing consistency
- Random access performance in HBase & block size in HDFS
- HDFS buffered write/read operations
- HDFS guaranteed read/write of data from/to file