Today we continue discussing Big Data.
We compare the following NoSQL databases :
Cassandra: This has an Apache license and is written in Java. It claims to store huge datasets with a query language that is similar to SQL except that it has no joins and no aggregate functions. The queries are done by the keys or key ranges and there is tunable parameter for determining the tradeoffs between distribution and replication. Data can have expiration set on insert. Writes can be much faster than reads. It has very good and reliable cross center data replication. It enables Map/Reduce with Apache Hadoop
MongoDB: This is written entirely in C++. It claims to retain query and indexing properties from SQL It has an AGPL license. It enables Master or Slave replication and auto failover with replica sets. Sharding is built-in. It uses a custom binary protocol called BSON and queries are written as Javascript expressions. These are executed server side. It runs arbitrary javascript functions on server-side.It has a better update in place than CouchDB. It uses memory -mapped files for data storage. Text Search is integrated. This is a convenient transition from SQL to NoSQL where we want to move away from schema and still retain the indexing and querying
ElasticSearch: this has an advanced search capability. It stores json documents and versions them. It can maintain hierarchy between documents. It enables stat-groups that help with debugging. It is querying is rich and powerful and moreover it is script able. It sorts by score and enables fuzzy or approximate searches.
CouchDB is a classic document and big table store. It is written in Erlang and claims to have DB consistency. It has Apache license. It uses HTTP/REST protocol. It supports bidirectional replication in both continuous and adhoc mode, with conflict resolution and hence master to master replication. Because it supports multiversion concurrency document, it supports older versions of documents. It has embedded map/reduce and shares updates in real time with _changes.
Courtesy:kkovacs.eu
We compare the following NoSQL databases :
Cassandra: This has an Apache license and is written in Java. It claims to store huge datasets with a query language that is similar to SQL except that it has no joins and no aggregate functions. The queries are done by the keys or key ranges and there is tunable parameter for determining the tradeoffs between distribution and replication. Data can have expiration set on insert. Writes can be much faster than reads. It has very good and reliable cross center data replication. It enables Map/Reduce with Apache Hadoop
MongoDB: This is written entirely in C++. It claims to retain query and indexing properties from SQL It has an AGPL license. It enables Master or Slave replication and auto failover with replica sets. Sharding is built-in. It uses a custom binary protocol called BSON and queries are written as Javascript expressions. These are executed server side. It runs arbitrary javascript functions on server-side.It has a better update in place than CouchDB. It uses memory -mapped files for data storage. Text Search is integrated. This is a convenient transition from SQL to NoSQL where we want to move away from schema and still retain the indexing and querying
ElasticSearch: this has an advanced search capability. It stores json documents and versions them. It can maintain hierarchy between documents. It enables stat-groups that help with debugging. It is querying is rich and powerful and moreover it is script able. It sorts by score and enables fuzzy or approximate searches.
CouchDB is a classic document and big table store. It is written in Erlang and claims to have DB consistency. It has Apache license. It uses HTTP/REST protocol. It supports bidirectional replication in both continuous and adhoc mode, with conflict resolution and hence master to master replication. Because it supports multiversion concurrency document, it supports older versions of documents. It has embedded map/reduce and shares updates in real time with _changes.
Courtesy:kkovacs.eu
No comments:
Post a Comment