Drizzle was forked from the (now defunct) MySQL 6.0 tree in 2008. Since then there have been a lot of changes. Drizzle is in some ways similar to MySQL, and in other ways, unrecognizable.
This section of documentation aims to explore some of the notable differences between MySQL and Drizzle, and has been modified from its original state on the Drizzle Wiki.
- There is no embedded server. The Drizzle Server is not loadable as a shared library.
- Drizzle is optimized for massively concurrent environments. If we have the choice of improving performance for 1024 simultaneous connections to the detriment of performance with only 64 connections, we will take that choice.
- It is designed for modern POSIX systems
- Microsoft Windows is not a supported platform (neither is HP-UX or IRIX).
- Drizzle doesn’t use timezones. Everything is UTC.
- No scripts/mysql_install_db or similar. Drizzle aims for a “just works” installation, without administrative overhead.
- No system database that needs upgrading between versions.
- Drizzle can listen on the Drizzle port (4427) and/or MySQL port (3306) and speak the respective protocols.
Drizzle is designed around the concept of being a microkernel. There should be a small core of the server with most functionality being provided through small, efficient and hard to misuse plugin interfaces. The goal is a small, light-weight kernel that is easy to maintain, understand and extend.
Drizzle is written in C++ and makes use of the Standard Template Library (STL) and Boost. Only where performance or correctness proves to be inadequate will we consider rolling our own; our preference is to fix the upstream library instead.
Pluggable network protocols allow Drizzle to speak one (or more) of several protocols. Currently we support the MySQL protocol (compatible with existing MySQL client libraries) and the Drizzle protocol, which is still under development.
The Drizzle protocol embodies several important differences from MySQL:
- Client sends first packet (rather than the server)
- Built in sharding
- Multi statement support (without requiring a semicolon to separate them)
- Room for expansion to include NoSQL-type commands inline with SQL commands.
There is also a console plugin – instead of providing access over a network socket, this plugin allows access from the current tty.
The existing plugin APIs that Drizzle inherited from MySQL have been reworked.
- User Defined Functions (UDFs) now follow the same API as a given server instead of a different C API. This means that UDFs are on the exact same level as built-in functions
- Some parts of the storage Engine API have been extensively reworked, especially around transactions and DDL
- Logging is now pluggable
- Authentication is pluggable
- Replication is pluggable
- INFORMATION_SCHEMA plugins have been replaced by the function_engine, which is a lot more space and time efficient
- Network protocols are pluggable
- Scheduler is pluggable (multi_thread, pool_of_threads, etc)
- Plugin points for manipulating rows before/after operations: these can be used for replication and the PBMS Blob Streaming plugin
Drizzle does not currently have any plugins that implement stored procedures. We viewed the implementation in MySQL to be non-optimal. They bloat the parser and only support one language (SQL2003 stored procedures), which was not well known.
Fundamentally, stored procedures usually are not the correct architectural decision for applications that need to scale. Pushing more computation down into the database (which is the trickiest layer to scale) isn’t a good idea.
We do recognize the value of using stored procedures to reduce the time row locks are held, but think we can achieve the same advantage by improved batching of commands over the wire. This removes adding and administering stored procedures from the list of things that can go wrong in administering the database.
Drizzle does not currently have any plugin that provides SQL triggers. We have some hooks for callbacks inside the server so that plugins can hook into points that triggers could.
SQL Views are not currently supported in Drizzle. We believe they should be implemented via a query rewrite plugin.
See the Query Rewrite Blueprint on launchpad.
The INFORMATION_SCHEMA provides access to database metadata.
The INFORMATION_SCHEMA in Drizzle is strictly ANSI compliant. If you write a query to any of the tables in the INFORMATION_SCHEMA in Drizzle, you can directly run these on any other ANSI compliant system.
For information that does not fit into the standard, there is also the DATA_DICTIONARY schema. Use of tables in DATA_DICTIONARY is non-portable.
This allows developers to easily know if the query is portable or not.
Authentication lies in Drizzle plugins. Currently there are PAM and HTTP AUTH plugins for authentication. Through the PAM plugin, you can use any PAM module (such as LDAP).
For more information, see our AUTHENTICATION doc.
We’ve stopped the confusion: -p means port and -P means password.
No gotcha of using the unix socket when localhost is specified and then connecting you to the wrong database server.
There is no Drizzle admin command.
- MERGE storage engine has been removed
- FEDERATED storage engine has been removed (all current development is focused on FederatedX, so having FEDERATED made no sense).
- CSV engine is now for temporary tables only. See the filesystem_engine for the future of reading files as database tables.
- MyISAM is for temporary tables only.
- ARCHIVE is fully supported
- PBXT is merged
There are no FRM files in Drizzle. Engines now own their own metadata. Some still choose to store these in files on disk. These are now in a documented file format (using the google protobuf library).
Several SHOW commands have been removed, replaced with INFORMATION_SCHEMA or DATA_DICTIONARY views. All SHOW commands are aliases to INFORMATION_SCHEMA queries. Our INFORMATION_SCHEMA implementation does not have the drawbacks of the MySQL implementation.
Here are some specific SHOW differences:
- SHOW ENGINES: use DATA_DICTIONARY
- SHOW CREATE TABLE: specifies the AUTO_INCREMENT at CREATE/ALTER TABLE time, not the current AUTO_INCREMENT
- SHOW TABLE STATUS: only shows tables in the current table cache
- SHOW [ENGINE] STATUS: use the DATA_DICTIONARY tables for that engine
- ALTER TABLE UPGRADE
- REPAIR TABLE
- CREATE FUNCTION
- CONVERT
- SET NAMES
- Multi-table delete and multi-table update code was removed and can be accomplished through subqueries. More detailed information can be found in the Data Manipulation Language section.
- crypt()
- bit_length()
- bit_count()
- CIPHER
- CLIENT
- CODE
- CONTRIBUTORS
- CPU
- DEFINER
- DES_KEY_FILE
- ENGINES
- EVERY
- IO
- IPC
- ISSUSER
- There is no requirement for a ‘mysql’ schema.
- There is no SET datatype, use ENUM.
- There is no SET NAMES command, UTF-8 by default
- There is no CHARSET or CHARACTER SET commands, everything defaults to UTF8
- There is no TIME type, use DATETIME or INT.
- There is no TINYINT, SMALLINT or MEDIUMINT. Integer operations have been optimized around 32 and 64 bit integers.
- There are no TINYBLOB, MEDIUMBLOB and LONGBLOB datatypes. We have optimized a single BLOB container.
- There are no TINYTEXT, MEDIUMTEXT and LONGTEXT datatypes. Use TEXT or BLOB.
- There is no UNSIGNED (as per the standard).
- There are no spatial data types GEOMETRY, POINT, LINESTRING & POLYGON (go use Postgres).
- No YEAR field type.
- There are no FULLTEXT indexes for the MyISAM storage engine (the only engine FULLTEXT was supported in). Look at either Lucene, Sphinx, or Solr.
- No “dual” table.
- The “LOCAL” keyword in “LOAD DATA LOCAL INFILE” is not supported