mysql存储引擎memory，ndb，innodb之选择

mysql的innodb和cluster的NDB引擎都支持事务，在有共同的特性外，也有不同之处：

以mysql cluster NDB 7.3和MySQL 5.6之InnoDB为例：

ndb7.3基于mysql5.6，包括支持innodb1.1，因此可以在cluster里使用innodb表，但这些表不是集群的。

MySQL Cluster NDB存储引擎用分布式, shared-nothing的架构实现，这使其和innodb有不少不同之处。

比如事务、外键、表限制等，具体见下表：

Theseare shown in the following table:

Feature	InnoDB 1.1	MySQLCluster NDB 7.3, MySQL Cluster NDB7.4
MySQLServer Version	5.6	5.6
InnoDB Version	InnoDB 5.6.23	InnoDB 5.6.23
MySQLCluster Version	N/A	NDB 7.3.9
StorageLimits	64TB	3TB (Practicalupper limit based on 48 data nodes with 64GB RAM each; can beincreased with disk-based data and BLOBs)
ForeignKeys	Yes	Priorto MySQL Cluster NDB 7.3: No. (Ignored, as with MyISAM ) Availablein MySQL Cluster NDB 7.3.
Transactions	Allstandard types	READCOMMITTED
MVCC	Yes	No
DataCompression	Yes	No (MySQLCluster checkpoint and backup files can be compressed)
LargeRow Support (> 14K)	Supportedfor VARBINARY , VARCHAR , BLOB ,and TEXT columns	Supportedfor BLOB and TEXT columns only (Usingthese types to store very large amounts of data can lower MySQLCluster performance)
ReplicationSupport	Asynchronousand semisynchronous replication using MySQL Replication	Automaticsynchronous replication within a MySQL Cluster. Asynchronousreplication between MySQL Clusters, using MySQL Replication
Scaleoutfor Read Operations	Yes(MySQL Replication)	Yes(Automatic partitioning in MySQL Cluster; MySQL Replication)
Scaleoutfor Write Operations	Requiresapplication-level partitioning (sharding)	Yes(Automatic partitioning in MySQL Cluster is transparent toapplications)
HighAvailability (HA)	Requiresadditional software	Yes(Designed for 99.999% uptime)
NodeFailure Recovery and Failover	Requiresadditional software	Automatic (Keyelement in MySQL Cluster architecture)
Timefor Node Failure Recovery	30seconds or longer	Typically< 1 second
Real-TimePerformance	No	Yes
In-MemoryTables	No	Yes (Somedata can optionally be stored on disk; both in-memory and diskdata storage are durable)
NoSQLAccess to Storage Engine	Nativememcached interface in development (see the MySQL Dev ZonearticleMySQLCluster 7.2 (DMR2): NoSQL, Key/Value, Memcached)	Yes MultipleAPIs, including Memcached, Node.js/JavaScript, Java, JPA, C++,and HTTP/REST
Concurrentand Parallel Writes	Notsupported	Upto 48 writers, optimized for concurrent writes
ConflictDetection and Resolution (Multiple Replication Masters)	No	Yes
HashIndexes	No	Yes
OnlineAddition of Nodes	Read-onlyreplicas using MySQL Replication	Yes(all node types)
OnlineUpgrades	No	Yes
OnlineSchema Modifications	Yes,as part of MySQL 5.6.	Yes.

数据驱动型应用负载于innodb和NDB存储引擎之主要不同：

Workload	InnoDB	MySQLCluster ( NDB )
High-VolumeOLTP Applications	Yes	Yes
DSSApplications (data marts, analytics)	Yes	Limited(Join operations across OLTP datasets not exceeding 3TB in size)
CustomApplications	Yes	Yes
PackagedApplications	Yes	Limited(should be mostly primary key access). MySQLCluster NDB 7.3 supports foreign keys.
In-NetworkTelecoms Applications (HLR, HSS, SDP)	No	Yes
SessionManagement and Caching	Yes	Yes
E-CommerceApplications	Yes	Yes
UserProfile Management, AAA Protocol	Yes	Yes

这两种存储引擎适合的应用场景

Preferredapplication requirements for InnoDB	Preferredapplication requirements for NDB
Foreignkeys Note MySQLCluster NDB 7.3 supports foreign keys. Fulltable scans Verylarge databases, rows, or transactions Transactionsother than READCOMMITTED	Writescaling 99.999%uptime Onlineaddition of nodes and online schema operations MultipleSQL and NoSQL APIs (seeMySQLCluster APIs: Overview and Concepts) Real-timeperformance Limiteduse of BLOB columns Foreignkeys are supported, although their use may have an impact onperformance at high throughput

2

如何选择memory存储引擎或mysql cluster：

When to Use MEMORY or MySQL Cluster.

Developers looking to deploy applications that use the MEMORY storage engine for important, highly available, or frequently updated data should consider whether MySQL Cluster is a better choice. A typical use case for the MEMORY engine involves these

characteristics:

• Operations involving transient, non-critical data such as session management or caching. When the MySQL server halts or restarts, the data in MEMORY tables is lost.

• In-memory storage for fast access and low latency. Data volume can fit entirely in memory without causing the operating system to swap out virtual memory pages.

• A read-only or read-mostly data access pattern (limited updates).

MySQL Cluster offers the same features as the MEMORY engine with higher performance levels, and provides additional features not available with MEMORY:

• Row-level locking and multiple-thread operation for low contention between clients.

• Scalability even with statement mixes that include writes.

• Optional disk-backed operation for data durability.

• Shared-nothing architecture and multiple-host operation with no single point of failure, enabling 99.999% availability.

• Automatic data distribution across nodes; application developers need not craft custom sharding or partitioning solutions.

• Support for variable-length data types (including BLOB and TEXT) not supported by MEMORY.

MEMORY存储引擎和MySQL Cluster的更多细节对比参见白皮书《Scaling Web Services with MySQL Cluster: An Alternative to the MySQL Memory Storage Engine》

Table 15.4 

MEMORY 

Storage Engine Features

Storagelimits	RAM	Transactions	No	Lockinggranularity	Table
MVCC	No	Geospatialdata type support	No	Geospatialindexing support	No
B-treeindexes	Yes	T-treeindexes	No	Hashindexes	Yes
Full-textsearch indexes	No	Clusteredindexes	No	Datacaches	N/A
Indexcaches	N/A	Compresseddata	No	Encrypteddata[a]	Yes
Clusterdatabase support	No	Replicationsupport[b]	Yes	Foreignkey support	No
Backup/ point-in-time recovery[c]	Yes	Querycache support	Yes	Updatestatistics for data dictionary	Yes
[a]Implemented in the server (via encryption functions), ratherthan in the storage engine. Implemented in the server, rather than in the storage engine. [c]Implemented in the server, rather than in the storage engine.

3

myisam, memory, ndb, archive, innodb存储引擎功能汇总：
[b]Table 15.1 StorageEngines Feature Summary

Feature	MyISAM	Memory	InnoDB	Archive	NDB
Storagelimits	256TB	RAM	64TB	None	384EB
Transactions	No	No	Yes	No	Yes
Lockinggranularity	Table	Table	Row	Table	Row
MVCC	No	No	Yes	No	No
Geospatialdata type support	Yes	No	Yes	Yes	Yes
Geospatialindexing support	Yes	No	Yes[a]	No	No
B-treeindexes	Yes	Yes	Yes	No	No
T-treeindexes	No	No	No	No	Yes
Hashindexes	No	Yes	No[b]	No	Yes
Full-textsearch indexes	Yes	No	Yes[c]	No	No
Clustered indexes	No	No	Yes	No	No
Datacaches	No	N/A	Yes	No	Yes
Indexcaches	Yes	N/A	Yes	No	Yes
Compresseddata	Yes[d]	No	Yes[e]	Yes	No
Encrypteddata[f]	Yes	Yes	Yes	Yes	Yes
Clusterdatabase support	No	No	No	No	Yes
Replicationsupport[g]	Yes	Yes	Yes	Yes	Yes
Foreignkey support	No	No	Yes	No	No
Backup/ point-in-time recovery[h]	Yes	Yes	Yes	Yes	Yes
Querycache support	Yes	Yes	Yes	Yes	Yes
Updatestatistics for data dictionary	Yes	Yes	Yes	Yes	Yes
[a]InnoDB support for geospatial indexing is available in MySQL5.7.5 and higher. [b]InnoDB utilizes hash indexes internally for its AdaptiveHash Index feature. [c]InnoDB support for FULLTEXT indexes is available in MySQL5.6.4 and higher. [d]Compressed MyISAM tables are supported only when using thecompressed row format. Tables using the compressed row formatwith MyISAM are read only. [e]Compressed InnoDB tables require the InnoDB Barracuda fileformat. [f]Implemented in the server (via encryption functions), ratherthan in the storage engine. [g]Implemented in the server, rather than in the storageengine. [h]Implemented in the server, rather than in the storageengine.

要选mysql cluster要根据ndb存储引擎的特征和应用场景做详细测试，安装简测见我的博文《centos65安装简测mysql
cluster 7.3.7》http://blog.csdn.net/beiigang/article/details/43485585