History[08][04][09][10][11]

Hypertable was released by the company Zvents at 2008, and Doug Judd promoted to the Hypertable. Baidu, the search engine in China, became project sponsor in January 2009. The development ended in March 2016.

Checkpoints[12][13]

Hypertable backup by outputting table data in random order. Hypertable will always be back into a consistent and operational state at the checkpoint.

Compression[14][15][16]

Over time, the RangeServers will write in-memory CellCaches to on-disk files, called CellStores. CellStores contain a series of sorted blocks of compressed sorted key/value pairs. These blocks are the minimum unit of data transfer from disk. User can choose the compressor used to compress CellStore blocks in each access group. The compressor option includes zlib, lzo, quicklz, snappy, bmz, none. The default compressor is quicklz.

Concurrency Control[17]

Hypertable uses Multi-Version Concurrency Control (MVCC). It uses auto-assign timestamps as revision numbers.

Data Model[06]

Hypertable uses a set of related columns.

Foreign Keys

Isolation Levels[15]

Hypertable provides snapshot isolation for queries with 8-byte timestamps.

Joins[06]

Hypertable does not support Joins.

Query Interface[18]

Hypertable provides the Hypertable Query Language (HQL) to create, modify, and query tables. HQL can also be used to invoke administrative commands. HQL can be interpreted by hypertable command line interface (ht shell), Thrift API methods, and Hypertable::HqlInterpreter C++ class.

Storage Architecture[15]

Hypertable is able to run on top of any filesystem. A File System (FS) broker process all filesystem requests. FS brokers currently support HDFS, MapR, Ceph, KFS, and local.

Storage Model

Storage Organization[15]

Stored Procedures

System Architecture[15]

The architecture of the Hypertable is shown below.

Hyperspace - Hyperspace is a lock manager. It contains a filesystem to store some metadata. Equivalent to Google's Chubby service.

Master - The Master handles all the meta operations. It also detects range server failures. It can re-assign ranges if necessary.

Range Server - Range servers handle the reading and writing of data.

FS Broker - File System (FS) broker process all filesystem requests. FS brokers currently support HDFS, MapR, Ceph, KFS, and local.

ThriftBroker - ThriftBroker provides the application interface. It is implemented with Apache Thrift. It currently supports Java, PHP, Ruby, Python, Perl, and C++.

Views