Note: See the corresponding cookbook about HDFS. This page has lecture notes.


HDFS was modeled after the “Google File System” paper (2003): PDF. It’s main purpose is to store large amounts of data in a way that’s reliable and efficient (bandwidth, not latency). To achieve these goals, big files are broken into “blocks” and copied across a cluster of machines.

Most Hadoop-related technologies (YARN, MapReduce, HBase, etc.) expect data to be kept in HDFS.

The NameNode is responsible for keeping track of which machine has which file blocks, and for reading/writing data to the right places. Assuming you have port 9000 tunneled on your delenn connection, you can open a browser to http://localhost:9000/hadoop/namenode:50070/ to browse our HDFS store and check out some stats about disk usage, etc. (If the page does not load, reload the page while holding the shift or control key.)


These quotes are copied from the original paper (PDF). Emphasis added.

The system is built from many inexpensive commodity components that often fail. It must constantly monitor itself and detect, tolerate, and recover promptly from component failures on a routine basis.

The system stores a modest number of large files. We expect a few million files, each typically 100 MB or larger in size. Multi-GB files are the common case and should be managed efficiently. Small files must be supported, but we need not optimize for them.

The workloads primarily consist of two kinds of reads: large streaming reads and small random reads. In large streaming reads, individual operations typically read hundreds of KBs, more commonly 1 MB or more. Successive operations from the same client often read through a contiguous region of a file. A small random read typically reads a few KBs at some arbitrary offset. Performance-conscious applications often batch and sort their small reads to advance steadily through the file rather than go back and forth.

The workloads also have many large, sequential writes that append data to files. Typical operation sizes are similar to those for reads. Once written, files are seldom modified again. Small writes at arbitrary positions in a file are supported but do not have to be efficient.

The system must efficiently implement well-defined semantics for multiple clients that concurrently append to the same file. Our files are often used as producer-consumer queues or for many-way merging. Hundreds of producers, running one per machine, will concurrently append to a file. Atomicity with minimal synchronization overhead is essential. The file may be read later, or a consumer may be reading through the file simultaneously.

High sustained bandwidth is more important than low latency. Most of our target applications place a premium on processing data in bulk at a high rate, while few have stringent response time requirements for an individual read or write.

Small files

These notes are adapted from our Hadoop book, p. 148.

There are three major issues with processing many small files in Hadoop:

The default HDFS block size is 128 MB. If you find the need to store many files less than 128 MB, then you should consider compaction techniques. These techniques may be found in the HDFS cookbook.

CINF 401 material by Joshua Eckroth is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Source code for this website available at GitHub.